diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18636 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 3101,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0638297872340426e-05,
+      "loss": 4.6342,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-05,
+      "loss": 4.6483,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1914893617021275e-05,
+      "loss": 4.6701,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-05,
+      "loss": 4.5565,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.319148936170213e-05,
+      "loss": 4.4473,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.382978723404255e-05,
+      "loss": 4.3015,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.446808510638298e-05,
+      "loss": 4.308,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-05,
+      "loss": 4.1786,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.574468085106382e-05,
+      "loss": 4.2147,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010638297872340425,
+      "loss": 4.2331,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011702127659574468,
+      "loss": 4.1476,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001276595744680851,
+      "loss": 4.1404,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013829787234042554,
+      "loss": 4.1148,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014893617021276596,
+      "loss": 4.1409,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001595744680851064,
+      "loss": 4.1197,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017021276595744682,
+      "loss": 4.0744,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018085106382978726,
+      "loss": 4.0125,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019148936170212765,
+      "loss": 4.0387,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020212765957446807,
+      "loss": 4.0033,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002127659574468085,
+      "loss": 3.9891,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022340425531914892,
+      "loss": 3.8875,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023404255319148937,
+      "loss": 4.0952,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002446808510638298,
+      "loss": 4.0212,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002553191489361702,
+      "loss": 3.9252,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026595744680851064,
+      "loss": 3.8274,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002765957446808511,
+      "loss": 3.9679,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028723404255319153,
+      "loss": 3.9235,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002978723404255319,
+      "loss": 3.8548,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030851063829787236,
+      "loss": 3.9201,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003191489361702128,
+      "loss": 3.8505,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003297872340425532,
+      "loss": 3.7114,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034042553191489364,
+      "loss": 3.8871,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003510638297872341,
+      "loss": 3.7322,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003617021276595745,
+      "loss": 3.7742,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003723404255319149,
+      "loss": 3.8589,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003829787234042553,
+      "loss": 3.8215,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039361702127659574,
+      "loss": 3.9244,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040425531914893613,
+      "loss": 3.7586,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004148936170212766,
+      "loss": 3.7317,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000425531914893617,
+      "loss": 3.8096,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043617021276595746,
+      "loss": 3.7728,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044680851063829785,
+      "loss": 3.7629,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004574468085106383,
+      "loss": 3.8167,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046808510638297874,
+      "loss": 3.7099,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004787234042553192,
+      "loss": 3.6397,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004893617021276596,
+      "loss": 3.7662,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 3.6969,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005106382978723404,
+      "loss": 3.68,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005212765957446809,
+      "loss": 3.6187,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005319148936170213,
+      "loss": 3.6053,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005425531914893617,
+      "loss": 3.7706,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005531914893617022,
+      "loss": 3.6709,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005638297872340426,
+      "loss": 3.7204,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005744680851063831,
+      "loss": 3.8184,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005851063829787234,
+      "loss": 3.7567,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005957446808510638,
+      "loss": 3.6134,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006063829787234043,
+      "loss": 3.7236,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006170212765957447,
+      "loss": 3.6425,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006276595744680851,
+      "loss": 3.7258,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006382978723404256,
+      "loss": 3.6305,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000648936170212766,
+      "loss": 3.691,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006595744680851064,
+      "loss": 3.5683,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006702127659574469,
+      "loss": 3.5715,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006808510638297873,
+      "loss": 3.5824,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006914893617021278,
+      "loss": 3.5241,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007021276595744682,
+      "loss": 3.4804,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007127659574468085,
+      "loss": 3.5855,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000723404255319149,
+      "loss": 3.4676,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007340425531914894,
+      "loss": 3.6303,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007446808510638298,
+      "loss": 3.6311,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007553191489361702,
+      "loss": 3.5639,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007659574468085106,
+      "loss": 3.6574,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000776595744680851,
+      "loss": 3.5481,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007872340425531915,
+      "loss": 3.5343,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007978723404255319,
+      "loss": 3.5138,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008085106382978723,
+      "loss": 3.5036,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008191489361702128,
+      "loss": 3.5825,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008297872340425531,
+      "loss": 3.5222,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008404255319148936,
+      "loss": 3.5668,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000851063829787234,
+      "loss": 3.6167,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008617021276595744,
+      "loss": 3.5627,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008723404255319149,
+      "loss": 3.5699,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008829787234042553,
+      "loss": 3.4169,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008936170212765957,
+      "loss": 3.5708,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009042553191489362,
+      "loss": 3.5,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009148936170212766,
+      "loss": 3.5372,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000925531914893617,
+      "loss": 3.5185,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009361702127659575,
+      "loss": 3.4574,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009468085106382979,
+      "loss": 3.6023,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009574468085106384,
+      "loss": 3.4911,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009680851063829787,
+      "loss": 3.4604,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009787234042553192,
+      "loss": 3.4781,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009893617021276596,
+      "loss": 3.5277,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 3.562,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999997271192761,
+      "loss": 3.433,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999989084774027,
+      "loss": 3.5221,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999997544075273,
+      "loss": 3.4968,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999956339143761,
+      "loss": 3.4706,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999931779967977,
+      "loss": 3.5191,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999901763252177,
+      "loss": 3.5578,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999866289029128,
+      "loss": 3.524,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999825357337555,
+      "loss": 3.4544,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999977896822213,
+      "loss": 3.4774,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999727121733492,
+      "loss": 3.4459,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999669817928227,
+      "loss": 3.4277,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999607056868892,
+      "loss": 3.4952,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999538838623982,
+      "loss": 3.6091,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999465163267967,
+      "loss": 3.4673,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999386030881264,
+      "loss": 3.488,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999301441550243,
+      "loss": 3.4814,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999921139536724,
+      "loss": 3.5256,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999911589243054,
+      "loss": 3.4824,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999014932844388,
+      "loss": 3.3502,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998908516718984,
+      "loss": 3.5416,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999879664417048,
+      "loss": 3.4455,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998679315320993,
+      "loss": 3.4724,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998556530298585,
+      "loss": 3.477,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998428289237282,
+      "loss": 3.512,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998294592277062,
+      "loss": 3.4368,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998155439563858,
+      "loss": 3.3966,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998010831249557,
+      "loss": 3.4602,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997860767492002,
+      "loss": 3.3586,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997705248454994,
+      "loss": 3.3943,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997544274308282,
+      "loss": 3.3518,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997377845227576,
+      "loss": 3.3641,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997205961394533,
+      "loss": 3.421,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997028622996773,
+      "loss": 3.387,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996845830227863,
+      "loss": 3.5264,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996657583287325,
+      "loss": 3.4043,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996463882380635,
+      "loss": 3.4886,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996264727719223,
+      "loss": 3.4543,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996060119520469,
+      "loss": 3.508,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999585005800771,
+      "loss": 3.3412,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999563454341023,
+      "loss": 3.3768,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999541357596327,
+      "loss": 3.4162,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995187155908021,
+      "loss": 3.4034,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994955283491624,
+      "loss": 3.2749,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994717958967177,
+      "loss": 3.4489,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994475182593721,
+      "loss": 3.3973,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994226954636253,
+      "loss": 3.4231,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999397327536572,
+      "loss": 3.4176,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999371414505902,
+      "loss": 3.3445,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993449563998997,
+      "loss": 3.4304,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999317953247445,
+      "loss": 3.4304,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992904050780123,
+      "loss": 3.4697,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992623119216711,
+      "loss": 3.3699,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992336738090858,
+      "loss": 3.3622,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992044907715153,
+      "loss": 3.4186,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991747628408138,
+      "loss": 3.4145,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991444900494298,
+      "loss": 3.3931,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991136724304068,
+      "loss": 3.4401,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990823100173832,
+      "loss": 3.3557,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990504028445917,
+      "loss": 3.4629,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990179509468595,
+      "loss": 3.4617,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989849543596088,
+      "loss": 3.3905,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989514131188558,
+      "loss": 3.4278,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998917327261212,
+      "loss": 3.4224,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988826968238826,
+      "loss": 3.3131,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988475218446675,
+      "loss": 3.4012,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988118023619612,
+      "loss": 3.3399,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987755384147522,
+      "loss": 3.3905,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987387300426235,
+      "loss": 3.342,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998701377285752,
+      "loss": 3.4265,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986634801849093,
+      "loss": 3.5024,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998625038781461,
+      "loss": 3.343,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998586053117367,
+      "loss": 3.2839,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985465232351806,
+      "loss": 3.3791,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985064491780495,
+      "loss": 3.3057,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998465830989716,
+      "loss": 3.3701,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984246687145155,
+      "loss": 3.3101,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983829623973774,
+      "loss": 3.4455,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983407120838252,
+      "loss": 3.3255,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982979178199763,
+      "loss": 3.3808,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982545796525415,
+      "loss": 3.3383,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982106976288252,
+      "loss": 3.3145,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981662717967257,
+      "loss": 3.3312,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998121302204735,
+      "loss": 3.3968,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980757889019385,
+      "loss": 3.4161,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980297319380146,
+      "loss": 3.3306,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997983131363236,
+      "loss": 3.4388,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997935987228468,
+      "loss": 3.3122,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978882995851695,
+      "loss": 3.4535,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978400684853928,
+      "loss": 3.3042,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977912939817832,
+      "loss": 3.3255,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977419761275793,
+      "loss": 3.3271,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976921149766123,
+      "loss": 3.3428,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976417105833069,
+      "loss": 3.3865,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997590763002681,
+      "loss": 3.203,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997539272290345,
+      "loss": 3.3354,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974872385025016,
+      "loss": 3.4559,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974346616959476,
+      "loss": 3.2985,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973815419280713,
+      "loss": 3.271,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973278792568545,
+      "loss": 3.3101,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997273673740871,
+      "loss": 3.3593,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972189254392873,
+      "loss": 3.2578,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971636344118626,
+      "loss": 3.3937,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971078007189483,
+      "loss": 3.39,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997051424421488,
+      "loss": 3.3297,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996994505581018,
+      "loss": 3.3204,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996937044259666,
+      "loss": 3.3319,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968790405201532,
+      "loss": 3.3498,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009968204944257911,
+      "loss": 3.3379,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967614060404847,
+      "loss": 3.4223,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967017754287301,
+      "loss": 3.3036,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966416026556154,
+      "loss": 3.4388,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996580887786821,
+      "loss": 3.4202,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996519630888618,
+      "loss": 3.2944,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099645783202787,
+      "loss": 3.3782,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996395491272032,
+      "loss": 3.3786,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963326086891498,
+      "loss": 3.346,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962691843478618,
+      "loss": 3.3138,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996205218317397,
+      "loss": 3.3341,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961407106675753,
+      "loss": 3.3469,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996075661468809,
+      "loss": 3.3133,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960100707921003,
+      "loss": 3.2872,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995943938709043,
+      "loss": 3.1956,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995877265291822,
+      "loss": 3.288,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958100506132126,
+      "loss": 3.2854,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957422947465813,
+      "loss": 3.2294,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995673997765885,
+      "loss": 3.3196,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956051597456719,
+      "loss": 3.317,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955357807610796,
+      "loss": 3.3675,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954658608878374,
+      "loss": 3.2897,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953954002022642,
+      "loss": 3.4666,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953243987812694,
+      "loss": 3.3784,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952528567023528,
+      "loss": 3.4241,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951807740436042,
+      "loss": 3.2678,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951081508837035,
+      "loss": 3.3365,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950349873019204,
+      "loss": 3.3133,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949612833781147,
+      "loss": 3.3916,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948870391927362,
+      "loss": 3.2874,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948122548268236,
+      "loss": 3.3705,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994736930362006,
+      "loss": 3.2848,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946610658805018,
+      "loss": 3.2503,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945846614651188,
+      "loss": 3.2746,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945077171992544,
+      "loss": 3.283,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944302331668946,
+      "loss": 3.2819,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943522094526152,
+      "loss": 3.3079,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994273646141581,
+      "loss": 3.2843,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941945433195455,
+      "loss": 3.2227,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941149010728513,
+      "loss": 3.3917,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940347194884293,
+      "loss": 3.2955,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939539986538004,
+      "loss": 3.3559,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938727386570726,
+      "loss": 3.4155,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937909395869432,
+      "loss": 3.451,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937086015326978,
+      "loss": 3.2875,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936257245842105,
+      "loss": 3.2404,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993542308831943,
+      "loss": 3.3602,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934583543669453,
+      "loss": 3.2482,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933738612808562,
+      "loss": 3.2974,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932888296659017,
+      "loss": 3.3117,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932032596148956,
+      "loss": 3.3091,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931171512212395,
+      "loss": 3.2625,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993030504578923,
+      "loss": 3.394,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929433197825225,
+      "loss": 3.3877,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928555969272027,
+      "loss": 3.3144,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927673361087145,
+      "loss": 3.1886,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926785374233971,
+      "loss": 3.3557,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992589200968176,
+      "loss": 3.2626,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924993268405641,
+      "loss": 3.2298,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924089151386612,
+      "loss": 3.3708,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923179659611537,
+      "loss": 3.2225,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922264794073143,
+      "loss": 3.3191,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009921344555770032,
+      "loss": 3.2591,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920418945706662,
+      "loss": 3.2962,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991948796489336,
+      "loss": 3.2612,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918551614346309,
+      "loss": 3.1553,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917609895087561,
+      "loss": 3.374,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916662808145023,
+      "loss": 3.2346,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991571035455246,
+      "loss": 3.3486,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099147525353495,
+      "loss": 3.2302,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913789351581625,
+      "loss": 3.3121,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912820804300168,
+      "loss": 3.254,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911846894562323,
+      "loss": 3.2408,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910867623431137,
+      "loss": 3.2829,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909882991975504,
+      "loss": 3.2671,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990889300127017,
+      "loss": 3.2862,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907897652395735,
+      "loss": 3.3517,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906896946438647,
+      "loss": 3.2885,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905890884491196,
+      "loss": 3.2477,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904879467651524,
+      "loss": 3.2502,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903862697023614,
+      "loss": 3.2599,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902840573717292,
+      "loss": 3.225,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901813098848239,
+      "loss": 3.3786,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900780273537956,
+      "loss": 3.2848,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00098997420989138,
+      "loss": 3.28,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898698576108961,
+      "loss": 3.283,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897649706262473,
+      "loss": 3.2013,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896595490519195,
+      "loss": 3.2768,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989553593002983,
+      "loss": 3.2638,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894471025950916,
+      "loss": 3.3133,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893400779444816,
+      "loss": 3.278,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892325191679728,
+      "loss": 3.3244,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891244263829685,
+      "loss": 3.2978,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000989015799707454,
+      "loss": 3.2365,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988906639259998,
+      "loss": 3.4179,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887969451597516,
+      "loss": 3.2566,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886867175264485,
+      "loss": 3.2744,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885759564804044,
+      "loss": 3.3493,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988464662142518,
+      "loss": 3.2392,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883528346342693,
+      "loss": 3.3646,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882404740777207,
+      "loss": 3.2564,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881275805955162,
+      "loss": 3.318,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880141543108815,
+      "loss": 3.2642,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879001953476245,
+      "loss": 3.2873,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877857038301334,
+      "loss": 3.3153,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876706798833787,
+      "loss": 3.2249,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875551236329116,
+      "loss": 3.2809,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874390352048645,
+      "loss": 3.3248,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873224147259503,
+      "loss": 3.2696,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987205262323463,
+      "loss": 3.2808,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870875781252772,
+      "loss": 3.2963,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986969362259848,
+      "loss": 3.2172,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868506148562107,
+      "loss": 3.2371,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867313360439804,
+      "loss": 3.2849,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866115259533532,
+      "loss": 3.2535,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864911847151043,
+      "loss": 3.249,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863703124605892,
+      "loss": 3.2645,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862489093217422,
+      "loss": 3.1771,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861269754310779,
+      "loss": 3.2371,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098600451092169,
+      "loss": 3.3113,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985881515927251,
+      "loss": 3.3088,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857579905820134,
+      "loss": 3.2945,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856339350208074,
+      "loss": 3.2578,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855093493790425,
+      "loss": 3.3328,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985384233792707,
+      "loss": 3.2641,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852585883983673,
+      "loss": 3.2082,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851324133331681,
+      "loss": 3.3341,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850057087348327,
+      "loss": 3.1994,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984878474741662,
+      "loss": 3.324,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847507114925345,
+      "loss": 3.3105,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984622419126907,
+      "loss": 3.2988,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844935977848135,
+      "loss": 3.2556,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843642476068653,
+      "loss": 3.2291,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842343687342515,
+      "loss": 3.2839,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841039613087373,
+      "loss": 3.288,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839730254726656,
+      "loss": 3.262,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838415613689562,
+      "loss": 3.2763,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837095691411046,
+      "loss": 3.3024,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983577048933184,
+      "loss": 3.1771,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834440008898424,
+      "loss": 3.3032,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009833104251563056,
+      "loss": 3.1399,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831763218783743,
+      "loss": 3.2278,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983041691202425,
+      "loss": 3.2783,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829065332754106,
+      "loss": 3.2507,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827708482448587,
+      "loss": 3.3164,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826346362588727,
+      "loss": 3.3307,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824978974661313,
+      "loss": 3.2609,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982360632015888,
+      "loss": 3.3845,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822228400579711,
+      "loss": 3.402,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820845217427836,
+      "loss": 3.2383,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819456772213033,
+      "loss": 3.28,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981806306645082,
+      "loss": 3.2134,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816664101662458,
+      "loss": 3.3105,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981525987937495,
+      "loss": 3.3483,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981385040112104,
+      "loss": 3.3435,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00098124356684392,
+      "loss": 3.2571,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811015682873647,
+      "loss": 3.3675,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809590445974327,
+      "loss": 3.335,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808159959296921,
+      "loss": 3.1917,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806724224402832,
+      "loss": 3.2378,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805283242859202,
+      "loss": 3.0905,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803837016238894,
+      "loss": 3.2316,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802385546120499,
+      "loss": 3.2589,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800928834088327,
+      "loss": 3.2974,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799466881732415,
+      "loss": 3.1248,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797999690648517,
+      "loss": 3.2047,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796527262438104,
+      "loss": 3.2954,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795049598708369,
+      "loss": 3.1246,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793566701072213,
+      "loss": 3.1975,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979207857114825,
+      "loss": 3.2068,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790585210560812,
+      "loss": 3.219,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789086620939935,
+      "loss": 3.3036,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787582803921365,
+      "loss": 3.2138,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978607376114655,
+      "loss": 3.2552,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784559494262649,
+      "loss": 3.3408,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783040004922514,
+      "loss": 3.2212,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009781515294784706,
+      "loss": 3.2271,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977998536551348,
+      "loss": 3.1993,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778450218778788,
+      "loss": 3.2573,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977690985625628,
+      "loss": 3.2584,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775364279627293,
+      "loss": 3.2358,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773813490578864,
+      "loss": 3.3188,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977225749080371,
+      "loss": 3.2439,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770696282000244,
+      "loss": 3.3837,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976912986587256,
+      "loss": 3.329,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767558244130438,
+      "loss": 3.2112,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765981418489336,
+      "loss": 3.1936,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097643993906704,
+      "loss": 3.1779,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762812162400445,
+      "loss": 3.2095,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000976121973541197,
+      "loss": 3.1911,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759622111443145,
+      "loss": 3.2793,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758019292237812,
+      "loss": 3.3302,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756411279545485,
+      "loss": 3.2008,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754798075121348,
+      "loss": 3.1425,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975317968072625,
+      "loss": 3.244,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751556098126704,
+      "loss": 3.2707,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974992732909489,
+      "loss": 3.346,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748293375408646,
+      "loss": 3.2123,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746654238851469,
+      "loss": 3.3,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745009921212514,
+      "loss": 3.2673,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743360424286591,
+      "loss": 3.3869,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741705749874165,
+      "loss": 3.3019,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740045899781352,
+      "loss": 3.198,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738380875819914,
+      "loss": 3.2414,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736710679807265,
+      "loss": 3.3015,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973503531356646,
+      "loss": 3.3285,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097333547789262,
+      "loss": 3.2705,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731669077720827,
+      "loss": 3.2424,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729978211790325,
+      "loss": 3.1655,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728282182980308,
+      "loss": 3.2811,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009726580993142036,
+      "loss": 3.2577,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724874644132392,
+      "loss": 3.2445,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723163137813897,
+      "loss": 3.219,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721446476054701,
+      "loss": 3.1759,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719724660728575,
+      "loss": 3.1746,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717997693714924,
+      "loss": 3.1689,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716265576898771,
+      "loss": 3.2055,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009714528312170761,
+      "loss": 3.3049,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712785901427157,
+      "loss": 3.2404,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000971103834656984,
+      "loss": 3.2936,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970928564950631,
+      "loss": 3.2243,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707527812149672,
+      "loss": 3.203,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705764836418646,
+      "loss": 3.2894,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703996724237565,
+      "loss": 3.2231,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702223477536359,
+      "loss": 3.257,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700445098250566,
+      "loss": 3.2252,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698661588321333,
+      "loss": 3.2733,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696872949695399,
+      "loss": 3.2988,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695079184325104,
+      "loss": 3.2411,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693280294168385,
+      "loss": 3.3203,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691476281188771,
+      "loss": 3.1778,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689667147355384,
+      "loss": 3.2067,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687852894642932,
+      "loss": 3.2865,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686033525031719,
+      "loss": 3.3049,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684209040507623,
+      "loss": 3.2493,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682379443062116,
+      "loss": 3.2165,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968054473469224,
+      "loss": 3.3911,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678704917400627,
+      "loss": 3.3445,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676859993195475,
+      "loss": 3.2885,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675009964090562,
+      "loss": 3.1688,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673154832105237,
+      "loss": 3.1851,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671294599264421,
+      "loss": 3.1956,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669429267598602,
+      "loss": 3.1736,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667558839143826,
+      "loss": 3.2575,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665683315941714,
+      "loss": 3.21,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966380270003944,
+      "loss": 3.3244,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661916993489741,
+      "loss": 3.0961,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660026198350907,
+      "loss": 3.1973,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009658130316686785,
+      "loss": 3.1458,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656229350566775,
+      "loss": 3.1948,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654323302065821,
+      "loss": 3.3106,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652412173264423,
+      "loss": 3.2163,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650495966248618,
+      "loss": 3.1401,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648574683109992,
+      "loss": 3.2895,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964664832594567,
+      "loss": 3.345,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644716896858314,
+      "loss": 3.1442,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642780397956124,
+      "loss": 3.2268,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640838831352829,
+      "loss": 3.3284,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638892199167699,
+      "loss": 3.2956,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636940503525524,
+      "loss": 3.2382,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634983746556627,
+      "loss": 3.2367,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963302193039685,
+      "loss": 3.2799,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631055057187563,
+      "loss": 3.3044,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629083129075652,
+      "loss": 3.1931,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627106148213521,
+      "loss": 3.214,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625124116759091,
+      "loss": 3.2194,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623137036875794,
+      "loss": 3.1515,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621144910732573,
+      "loss": 3.3106,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619147740503879,
+      "loss": 3.2071,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617145528369673,
+      "loss": 3.2313,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615138276515409,
+      "loss": 3.163,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613125987132053,
+      "loss": 3.1963,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611108662416064,
+      "loss": 3.342,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609086304569396,
+      "loss": 3.0808,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00096070589157995,
+      "loss": 3.2444,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605026498319319,
+      "loss": 3.255,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602989054347282,
+      "loss": 3.1596,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600946586107305,
+      "loss": 3.1907,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598899095828791,
+      "loss": 3.2257,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000959684658574662,
+      "loss": 3.2619,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594789058101153,
+      "loss": 3.2864,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592726515138233,
+      "loss": 3.1435,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590658959109169,
+      "loss": 3.3117,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588586392270746,
+      "loss": 3.1556,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958650881688522,
+      "loss": 3.098,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584426235220312,
+      "loss": 3.2365,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582338649549205,
+      "loss": 3.2857,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958024606215055,
+      "loss": 3.1553,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578148475308452,
+      "loss": 3.207,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576045891312475,
+      "loss": 3.2474,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573938312457637,
+      "loss": 3.1386,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571825741044411,
+      "loss": 3.3134,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569708179378715,
+      "loss": 3.1482,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567585629771917,
+      "loss": 3.3314,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565458094540828,
+      "loss": 3.2479,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563325576007701,
+      "loss": 3.2286,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956118807650023,
+      "loss": 3.2899,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559045598351543,
+      "loss": 3.1887,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556898143900207,
+      "loss": 3.1997,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554745715490214,
+      "loss": 3.2341,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552588315470991,
+      "loss": 3.1827,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955042594619739,
+      "loss": 3.0738,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548258610029684,
+      "loss": 3.1148,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546086309333572,
+      "loss": 3.2751,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954390904648017,
+      "loss": 3.204,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541726823846011,
+      "loss": 3.1927,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539539643813039,
+      "loss": 3.1719,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537347508768613,
+      "loss": 3.2563,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535150421105495,
+      "loss": 3.2237,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953294838322186,
+      "loss": 3.1499,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009530741397521283,
+      "loss": 3.3012,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528529466412737,
+      "loss": 3.2816,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526312592310598,
+      "loss": 3.1262,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524090777634632,
+      "loss": 3.2743,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009521864024810003,
+      "loss": 3.2179,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519632336267262,
+      "loss": 3.2283,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517395714442348,
+      "loss": 3.2818,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515154161776582,
+      "loss": 3.1753,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512907680716676,
+      "loss": 3.2247,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951065627371471,
+      "loss": 3.1627,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508399943228153,
+      "loss": 3.2219,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506138691719832,
+      "loss": 3.2528,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503872521657964,
+      "loss": 3.1971,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501601435516118,
+      "loss": 3.1477,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499325435773243,
+      "loss": 3.2172,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497044524913639,
+      "loss": 3.2184,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494758705426977,
+      "loss": 3.2353,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000949246797980828,
+      "loss": 3.2723,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490172350557925,
+      "loss": 3.3113,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487871820181646,
+      "loss": 3.2191,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485566391190525,
+      "loss": 3.3133,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483256066100989,
+      "loss": 3.1997,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480940847434813,
+      "loss": 3.1995,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947862073771911,
+      "loss": 3.1442,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476295739486332,
+      "loss": 3.1558,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473965855274268,
+      "loss": 3.2046,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471631087626042,
+      "loss": 3.1634,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469291439090103,
+      "loss": 3.208,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466946912220233,
+      "loss": 3.0875,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464597509575538,
+      "loss": 3.2013,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462243233720443,
+      "loss": 3.2517,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009459884087224693,
+      "loss": 3.1121,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457520072663352,
+      "loss": 3.3217,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009455151192616796,
+      "loss": 3.307,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452777449670712,
+      "loss": 3.0893,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450398846416093,
+      "loss": 3.1934,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448015385449243,
+      "loss": 3.2269,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445627069371758,
+      "loss": 3.2274,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443233900790544,
+      "loss": 3.25,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094408358823178,
+      "loss": 3.1615,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438433016571015,
+      "loss": 3.1383,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436025306172974,
+      "loss": 3.2473,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433612753751748,
+      "loss": 3.2358,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009431195361940693,
+      "loss": 3.238,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428773133378444,
+      "loss": 3.1777,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426346070708924,
+      "loss": 3.2386,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423914176581327,
+      "loss": 3.1866,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421477453650118,
+      "loss": 3.1697,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419035904575039,
+      "loss": 3.1742,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416589532021094,
+      "loss": 3.1645,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009414138338658556,
+      "loss": 3.2007,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411682327162959,
+      "loss": 3.1113,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009409221500215095,
+      "loss": 3.2193,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406755860501015,
+      "loss": 3.2195,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404285410712017,
+      "loss": 3.2781,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401810153544659,
+      "loss": 3.2341,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399330091700736,
+      "loss": 3.2489,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396845227887294,
+      "loss": 3.1825,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394355564816619,
+      "loss": 3.1748,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391861105206236,
+      "loss": 3.1069,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389361851778904,
+      "loss": 3.2267,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009386857807262614,
+      "loss": 3.1479,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009384348974390591,
+      "loss": 3.1389,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000938183535590128,
+      "loss": 3.1824,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379316954538356,
+      "loss": 3.1621,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937679377305071,
+      "loss": 3.279,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374265814192453,
+      "loss": 3.0674,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371733080722911,
+      "loss": 3.1519,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369195575406619,
+      "loss": 3.2336,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366653301013322,
+      "loss": 3.1189,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364106260317973,
+      "loss": 3.253,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361554456100721,
+      "loss": 3.0478,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358997891146923,
+      "loss": 3.0754,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356436568247127,
+      "loss": 3.291,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353870490197074,
+      "loss": 3.2246,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351299659797698,
+      "loss": 3.2036,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000934872407985512,
+      "loss": 3.1685,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346143753180644,
+      "loss": 3.175,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343558682590756,
+      "loss": 3.2161,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340968870907118,
+      "loss": 3.1747,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933837432095657,
+      "loss": 3.1042,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335775035571122,
+      "loss": 3.2374,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333171017587955,
+      "loss": 3.153,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330562269849414,
+      "loss": 3.1003,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327948795203005,
+      "loss": 3.1474,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009325330596501397,
+      "loss": 3.1469,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322707676602413,
+      "loss": 3.0854,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009320080038369031,
+      "loss": 3.1842,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317447684669377,
+      "loss": 3.1676,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009314810618376727,
+      "loss": 3.1911,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00093121688423695,
+      "loss": 3.2545,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309522359531251,
+      "loss": 3.175,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306871172750679,
+      "loss": 3.2474,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009304215284921618,
+      "loss": 3.2022,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301554698943024,
+      "loss": 3.2059,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009298889417718992,
+      "loss": 3.1692,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296219444158738,
+      "loss": 3.0845,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009293544781176597,
+      "loss": 3.0929,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009290865431692025,
+      "loss": 3.0687,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009288181398629594,
+      "loss": 3.1947,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009285492684918988,
+      "loss": 3.1432,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282799293494999,
+      "loss": 3.17,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280101227297525,
+      "loss": 3.2083,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277398489271567,
+      "loss": 3.1439,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009274691082367227,
+      "loss": 3.2192,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00092719790095397,
+      "loss": 3.1,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269262273749277,
+      "loss": 3.2859,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009266540877961336,
+      "loss": 3.2624,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263814825146341,
+      "loss": 3.2615,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009261084118279846,
+      "loss": 3.1447,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258348760342477,
+      "loss": 3.2187,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925560875431994,
+      "loss": 3.1887,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009252864103203014,
+      "loss": 3.2343,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925011480998755,
+      "loss": 3.1922,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247360877674462,
+      "loss": 3.114,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244602309269733,
+      "loss": 3.2385,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009241839107784403,
+      "loss": 3.167,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239071276234567,
+      "loss": 3.1323,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236298817641379,
+      "loss": 3.2066,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000923352173503104,
+      "loss": 3.232,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009230740031434801,
+      "loss": 3.2253,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227953709888953,
+      "loss": 3.2202,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225162773434831,
+      "loss": 3.1586,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222367225118806,
+      "loss": 3.1456,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009219567067992282,
+      "loss": 3.1921,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009216762305111696,
+      "loss": 3.1783,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921395293953851,
+      "loss": 3.2321,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921113897433921,
+      "loss": 3.1836,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009208320412585305,
+      "loss": 3.2514,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009205497257353321,
+      "loss": 3.1775,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202669511724793,
+      "loss": 3.2639,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009199837178786272,
+      "loss": 3.1824,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009197000261629314,
+      "loss": 3.2205,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919415876335048,
+      "loss": 3.1927,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191312687051328,
+      "loss": 3.1568,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188462035838418,
+      "loss": 3.2212,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185606812823299,
+      "loss": 3.1717,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009182747021122515,
+      "loss": 3.096,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179882663857593,
+      "loss": 3.1658,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009177013744155043,
+      "loss": 3.196,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009174140265146356,
+      "loss": 3.0531,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009171262229968003,
+      "loss": 3.2155,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168379641761425,
+      "loss": 3.0872,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916549250367303,
+      "loss": 3.2644,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009162600818854199,
+      "loss": 3.223,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159704590461268,
+      "loss": 3.1838,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156803821655542,
+      "loss": 3.2219,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153898515603272,
+      "loss": 3.0999,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150988675475669,
+      "loss": 3.1274,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148074304448888,
+      "loss": 3.1811,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009145155405704032,
+      "loss": 3.123,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009142231982427147,
+      "loss": 3.2761,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139304037809215,
+      "loss": 3.1384,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136371575046156,
+      "loss": 3.0785,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000913343459733882,
+      "loss": 3.0953,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009130493107892986,
+      "loss": 3.1517,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009127547109919355,
+      "loss": 3.1906,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009124596606633551,
+      "loss": 3.2565,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121641601256119,
+      "loss": 3.1926,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118682097012513,
+      "loss": 3.2777,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009115718097133099,
+      "loss": 3.1001,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112749604853152,
+      "loss": 3.0461,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000910977662341285,
+      "loss": 3.1728,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009106799156057268,
+      "loss": 3.2072,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009103817206036382,
+      "loss": 3.2435,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100830776605056,
+      "loss": 3.159,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000909783987102305,
+      "loss": 3.1201,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009094844492555003,
+      "loss": 3.2132,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009091844644470439,
+      "loss": 3.3439,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009088840330043764,
+      "loss": 3.0841,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009085831552554252,
+      "loss": 3.2024,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082818315286055,
+      "loss": 3.1352,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009079800621528189,
+      "loss": 3.1806,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009076778474574538,
+      "loss": 3.2105,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009073751877723841,
+      "loss": 3.164,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009070720834279702,
+      "loss": 3.1757,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009067685347550571,
+      "loss": 3.2549,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009064645420849753,
+      "loss": 3.2135,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061601057495398,
+      "loss": 3.0742,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009058552260810496,
+      "loss": 3.2726,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055499034122881,
+      "loss": 3.194,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052441380765217,
+      "loss": 3.2482,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009049379304075007,
+      "loss": 3.215,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046312807394573,
+      "loss": 3.1798,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009043241894071068,
+      "loss": 3.1854,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040166567456467,
+      "loss": 3.109,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009037086830907555,
+      "loss": 3.1606,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034002687785938,
+      "loss": 3.2186,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009030914141458027,
+      "loss": 3.2178,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009027821195295043,
+      "loss": 3.1832,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009024723852673005,
+      "loss": 3.1508,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021622116972736,
+      "loss": 3.2739,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000901851599157985,
+      "loss": 3.1511,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015405479884754,
+      "loss": 3.2107,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012290585282643,
+      "loss": 3.2317,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009171311173496,
+      "loss": 3.216,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009006047660962072,
+      "loss": 3.1985,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009002919638057907,
+      "loss": 3.1895,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008999787245875309,
+      "loss": 3.1949,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008996650487833356,
+      "loss": 3.2006,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993509367355892,
+      "loss": 3.1389,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899036388787152,
+      "loss": 3.1898,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987214052813603,
+      "loss": 3.2189,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008984059865620262,
+      "loss": 3.1866,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898090132973436,
+      "loss": 3.1726,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008977738448603514,
+      "loss": 3.2042,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897457122568008,
+      "loss": 3.1121,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971399664421153,
+      "loss": 3.162,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008968223768288566,
+      "loss": 3.0626,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008965043540748884,
+      "loss": 3.1483,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961858985273396,
+      "loss": 3.1053,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008958670105338118,
+      "loss": 3.2203,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955476904423785,
+      "loss": 3.1752,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895227938601585,
+      "loss": 3.1947,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949077553604477,
+      "loss": 3.1118,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008945871410684539,
+      "loss": 3.0393,
+      "step": 727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008942660960755614,
+      "loss": 3.0737,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008939446207321982,
+      "loss": 3.0708,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000893622715389262,
+      "loss": 3.0915,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933003803981199,
+      "loss": 3.2752,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008929776161106079,
+      "loss": 3.2326,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008926544228790305,
+      "loss": 3.1029,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923308010561608,
+      "loss": 3.2492,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008920067509952392,
+      "loss": 3.174,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008916822730499738,
+      "loss": 3.1198,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913573675745396,
+      "loss": 3.1225,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910320349235785,
+      "loss": 3.2323,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907062754521985,
+      "loss": 3.1342,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008903800895159736,
+      "loss": 3.0011,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008900534774709432,
+      "loss": 3.1642,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008897264396736116,
+      "loss": 3.1101,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008893989764809484,
+      "loss": 3.1353,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889071088250387,
+      "loss": 3.1125,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887427753398248,
+      "loss": 3.1624,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008884140381076229,
+      "loss": 3.1345,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880848769126055,
+      "loss": 3.16,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008877552921140599,
+      "loss": 3.0959,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874252840717351,
+      "loss": 3.1872,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008870948531458424,
+      "loss": 3.128,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867639996970551,
+      "loss": 3.117,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864327240865072,
+      "loss": 3.1673,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861010266757932,
+      "loss": 3.1724,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857689078269688,
+      "loss": 3.1527,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008854363679025491,
+      "loss": 3.1595,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851034072655093,
+      "loss": 3.0984,
+      "step": 756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008847700262792834,
+      "loss": 3.2066,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844362253077645,
+      "loss": 3.0298,
+      "step": 758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008841020047153039,
+      "loss": 3.1346,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008837673648667111,
+      "loss": 3.2158,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834323061272529,
+      "loss": 3.2147,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883096828862654,
+      "loss": 3.1216,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827609334390951,
+      "loss": 3.161,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008824246202232141,
+      "loss": 3.1573,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008820878895821044,
+      "loss": 3.1125,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817507418833152,
+      "loss": 3.1191,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000881413177494851,
+      "loss": 3.2914,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000881075196785171,
+      "loss": 3.1838,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008807368001231889,
+      "loss": 3.1319,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803979878782725,
+      "loss": 3.1906,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800587604202428,
+      "loss": 3.1493,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008797191181193748,
+      "loss": 3.1917,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008793790613463954,
+      "loss": 3.1883,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790385904724847,
+      "loss": 3.2167,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008786977058692745,
+      "loss": 3.1888,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008783564079088476,
+      "loss": 3.1834,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008780146969637391,
+      "loss": 3.1596,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776725734069342,
+      "loss": 3.1191,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008773300376118683,
+      "loss": 3.2015,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008769870899524275,
+      "loss": 3.1108,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766437308029467,
+      "loss": 3.1214,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762999605382105,
+      "loss": 3.1988,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008759557795334518,
+      "loss": 3.1074,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875611188164352,
+      "loss": 3.111,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008752661868070408,
+      "loss": 3.1031,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008749207758380949,
+      "loss": 3.2252,
+      "step": 786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008745749556345382,
+      "loss": 3.1009,
+      "step": 787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008742287265738416,
+      "loss": 3.1845,
+      "step": 788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008738820890339216,
+      "loss": 3.1353,
+      "step": 789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735350433931414,
+      "loss": 3.2036,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008731875900303093,
+      "loss": 3.0921,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008728397293246785,
+      "loss": 3.1436,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008724914616559467,
+      "loss": 3.2682,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008721427874042563,
+      "loss": 3.0338,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717937069501931,
+      "loss": 3.1637,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008714442206747866,
+      "loss": 3.1643,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008710943289595088,
+      "loss": 3.1439,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707440321862747,
+      "loss": 3.221,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008703933307374412,
+      "loss": 3.0803,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008700422249958068,
+      "loss": 3.1256,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008696907153446119,
+      "loss": 3.2808,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008693388021675368,
+      "loss": 3.2426,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689864858487033,
+      "loss": 3.1693,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008686337667726722,
+      "loss": 3.0742,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008682806453244447,
+      "loss": 3.1588,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008679271218894611,
+      "loss": 3.0458,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008675731968536002,
+      "loss": 3.097,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008672188706031792,
+      "loss": 3.1333,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008668641435249534,
+      "loss": 3.1224,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665090160061155,
+      "loss": 3.0124,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661534884342953,
+      "loss": 3.1,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008657975611975593,
+      "loss": 3.1514,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008654412346844102,
+      "loss": 3.2539,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008650845092837867,
+      "loss": 3.1879,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008647273853850624,
+      "loss": 3.1263,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008643698633780466,
+      "loss": 3.2353,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008640119436529825,
+      "loss": 3.0071,
+      "step": 817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863653626600548,
+      "loss": 3.1986,
+      "step": 818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008632949126118537,
+      "loss": 3.1859,
+      "step": 819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862935802078445,
+      "loss": 3.1353,
+      "step": 820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008625762953922986,
+      "loss": 3.0941,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008622163929458245,
+      "loss": 3.1265,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008618560951318646,
+      "loss": 3.0203,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008614954023436919,
+      "loss": 3.1504,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008611343149750112,
+      "loss": 3.1862,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008607728334199574,
+      "loss": 3.1274,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008604109580730958,
+      "loss": 3.1856,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860048689329422,
+      "loss": 3.1329,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008596860275843602,
+      "loss": 3.1424,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593229732337642,
+      "loss": 3.2196,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008589595266739164,
+      "loss": 3.105,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008585956883015264,
+      "loss": 3.1552,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008582314585137327,
+      "loss": 3.071,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008578668377081001,
+      "loss": 3.1382,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008575018262826209,
+      "loss": 3.2505,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008571364246357131,
+      "loss": 3.1711,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000856770633166221,
+      "loss": 3.1385,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008564044522734146,
+      "loss": 3.1129,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008560378823569885,
+      "loss": 3.1546,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008556709238170623,
+      "loss": 3.1097,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553035770541797,
+      "loss": 3.0547,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549358424693077,
+      "loss": 3.0809,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008545677204638375,
+      "loss": 3.262,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008541992114395825,
+      "loss": 3.0546,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008538303157987786,
+      "loss": 3.1745,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000853461033944084,
+      "loss": 3.0752,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008530913662785784,
+      "loss": 3.1343,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008527213132057624,
+      "loss": 3.1101,
+      "step": 848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008523508751295573,
+      "loss": 3.0981,
+      "step": 849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008519800524543049,
+      "loss": 3.2034,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008516088455847666,
+      "loss": 3.1987,
+      "step": 851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008512372549261231,
+      "loss": 3.1722,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008508652808839743,
+      "loss": 3.1902,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504929238643381,
+      "loss": 3.0393,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000850120184273651,
+      "loss": 3.23,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008497470625187666,
+      "loss": 3.2037,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849373559006956,
+      "loss": 3.137,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008489996741459068,
+      "loss": 3.1141,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008486254083437228,
+      "loss": 3.1107,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008482507620089236,
+      "loss": 3.1405,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008478757355504446,
+      "loss": 3.038,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008475003293776354,
+      "loss": 3.0702,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008471245439002605,
+      "loss": 3.148,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008467483795284987,
+      "loss": 3.0591,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008463718366729415,
+      "loss": 3.1634,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008459949157445944,
+      "loss": 3.1593,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008456176171548753,
+      "loss": 3.1324,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008452399413156138,
+      "loss": 3.2056,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008448618886390522,
+      "loss": 2.9817,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008444834595378434,
+      "loss": 3.1205,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008441046544250515,
+      "loss": 3.287,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008437254737141508,
+      "loss": 3.1449,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008433459178190259,
+      "loss": 3.1297,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008429659871539708,
+      "loss": 3.1621,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008425856821336885,
+      "loss": 3.0941,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008422050031732904,
+      "loss": 3.0605,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008418239506882965,
+      "loss": 3.1756,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008414425250946345,
+      "loss": 3.1358,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008410607268086388,
+      "loss": 3.1289,
+      "step": 879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008406785562470511,
+      "loss": 3.0407,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008402960138270193,
+      "loss": 3.1026,
+      "step": 881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008399130999660973,
+      "loss": 3.1565,
+      "step": 882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008395298150822443,
+      "loss": 3.0614,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008391461595938245,
+      "loss": 3.149,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008387621339196068,
+      "loss": 3.1641,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008383777384787639,
+      "loss": 3.0724,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837992973690872,
+      "loss": 3.0762,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837607839975911,
+      "loss": 3.0978,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837222337754263,
+      "loss": 3.1856,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008368364674467126,
+      "loss": 3.0643,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000836450229474446,
+      "loss": 3.0858,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008360636242590509,
+      "loss": 3.1731,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008356766522225156,
+      "loss": 3.007,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008352893137872291,
+      "loss": 3.1555,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083490160937598,
+      "loss": 3.175,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008345135394119566,
+      "loss": 3.2132,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341251043187461,
+      "loss": 3.1449,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008337363045203345,
+      "loss": 3.0853,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008333471404411054,
+      "loss": 3.0724,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329576125058406,
+      "loss": 3.1662,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008325677211397185,
+      "loss": 3.1085,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008321774667683147,
+      "loss": 3.0701,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008317868498176005,
+      "loss": 3.183,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008313958707139433,
+      "loss": 3.2582,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000831004529884106,
+      "loss": 3.1251,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008306128277552459,
+      "loss": 3.1495,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008302207647549147,
+      "loss": 3.1012,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008298283413110585,
+      "loss": 3.1086,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294355578520161,
+      "loss": 3.2025,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008290424148065199,
+      "loss": 3.1181,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008286489126036943,
+      "loss": 3.1626,
+      "step": 911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008282550516730562,
+      "loss": 3.1296,
+      "step": 912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008278608324445139,
+      "loss": 3.0505,
+      "step": 913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008274662553483662,
+      "loss": 3.0958,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008270713208153034,
+      "loss": 3.0558,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008266760292764058,
+      "loss": 3.1519,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008262803811631427,
+      "loss": 3.2194,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008258843769073734,
+      "loss": 3.1349,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008254880169413454,
+      "loss": 3.1735,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000825091301697695,
+      "loss": 3.2133,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008246942316094457,
+      "loss": 3.2733,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008242968071100086,
+      "loss": 3.1952,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008238990286331816,
+      "loss": 3.1738,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008235008966131491,
+      "loss": 3.1843,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008231024114844814,
+      "loss": 3.1048,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822703573682134,
+      "loss": 3.276,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008223043836414477,
+      "loss": 2.973,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219048417981472,
+      "loss": 3.1652,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215049485883419,
+      "loss": 3.0866,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008211047044485243,
+      "loss": 3.1993,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00082070410981557,
+      "loss": 3.0771,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008203031651267374,
+      "loss": 3.0962,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008199018708196664,
+      "loss": 3.1122,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195002273323792,
+      "loss": 3.1143,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000819098235103279,
+      "loss": 3.155,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008186958945711491,
+      "loss": 3.1124,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008182932061751538,
+      "loss": 3.1153,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008178901703548367,
+      "loss": 3.2448,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008174867875501203,
+      "loss": 3.0312,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008170830582013064,
+      "loss": 3.147,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008166789827490749,
+      "loss": 3.1787,
+      "step": 941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008162745616344831,
+      "loss": 3.1906,
+      "step": 942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008158697952989662,
+      "loss": 3.158,
+      "step": 943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008154646841843358,
+      "loss": 3.2005,
+      "step": 944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008150592287327799,
+      "loss": 3.0267,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008146534293868626,
+      "loss": 3.1975,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008142472865895231,
+      "loss": 3.341,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008138408007840754,
+      "loss": 3.1228,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134339724142082,
+      "loss": 2.9714,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008130268019239839,
+      "loss": 3.0103,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126192897578387,
+      "loss": 3.0808,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000812211436360581,
+      "loss": 3.1613,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118032421773926,
+      "loss": 3.1547,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113947076538264,
+      "loss": 3.105,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008109858332358074,
+      "loss": 3.127,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008105766193696311,
+      "loss": 3.1654,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008101670665019643,
+      "loss": 3.092,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809757175079843,
+      "loss": 2.9747,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008093469455506731,
+      "loss": 3.1088,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008089363783622297,
+      "loss": 3.0744,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808525473962656,
+      "loss": 3.0731,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008081142328004637,
+      "loss": 3.0788,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077026553245319,
+      "loss": 3.0818,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807290741984107,
+      "loss": 3.1427,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008068784932288018,
+      "loss": 3.1554,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000806465909508595,
+      "loss": 3.2164,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008060529912738315,
+      "loss": 3.1714,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056397389752208,
+      "loss": 3.1612,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008052261530638374,
+      "loss": 3.0517,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008048122339911195,
+      "loss": 3.1334,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008043979822088696,
+      "loss": 3.0299,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008039833981692529,
+      "loss": 3.0943,
+      "step": 972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035684823247972,
+      "loss": 3.0451,
+      "step": 973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008031532351283929,
+      "loss": 3.0618,
+      "step": 974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008027376570332916,
+      "loss": 3.0121,
+      "step": 975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008023217484931065,
+      "loss": 3.1398,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008019055099618111,
+      "loss": 3.0997,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014889418937394,
+      "loss": 3.0648,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008010720447435849,
+      "loss": 3.0828,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008006548189664005,
+      "loss": 3.1623,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008002372650175978,
+      "loss": 3.1404,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007998193833529461,
+      "loss": 3.1595,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007994011744285731,
+      "loss": 3.042,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007989826387009634,
+      "loss": 3.1718,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007985637766269582,
+      "loss": 3.0619,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007981445886637552,
+      "loss": 3.065,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007977250752689074,
+      "loss": 3.2228,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007973052369003238,
+      "loss": 3.0692,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796885074016267,
+      "loss": 3.1822,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007964645870753547,
+      "loss": 2.99,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796043776536558,
+      "loss": 2.9929,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007956226428592013,
+      "loss": 3.0954,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952011865029613,
+      "loss": 3.0538,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007947794079278677,
+      "loss": 3.1497,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007943573075943014,
+      "loss": 3.0602,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007939348859629943,
+      "loss": 3.0602,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007935121434950297,
+      "loss": 3.1235,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007930890806518403,
+      "loss": 3.1367,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792665697895209,
+      "loss": 3.0514,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007922419956872677,
+      "loss": 3.1376,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007918179744904974,
+      "loss": 3.1626,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007913936347677265,
+      "loss": 3.177,
+      "step": 1002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007909689769821319,
+      "loss": 3.0485,
+      "step": 1003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000790544001597237,
+      "loss": 3.1168,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007901187090769123,
+      "loss": 3.0795,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007896930998853744,
+      "loss": 3.1658,
+      "step": 1006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007892671744871852,
+      "loss": 3.2363,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007888409333472524,
+      "loss": 3.1998,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007884143769308277,
+      "loss": 3.1804,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007879875057035071,
+      "loss": 3.0356,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007875603201312307,
+      "loss": 3.0654,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007871328206802811,
+      "loss": 3.1805,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007867050078172836,
+      "loss": 3.034,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007862768820092062,
+      "loss": 3.2897,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007858484437233573,
+      "loss": 3.144,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007854196934273876,
+      "loss": 3.1677,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007849906315892881,
+      "loss": 3.0343,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007845612586773889,
+      "loss": 3.056,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007841315751603609,
+      "loss": 3.2044,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007837015815072135,
+      "loss": 3.1644,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007832712781872944,
+      "loss": 3.0538,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007828406656702893,
+      "loss": 3.0974,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007824097444262223,
+      "loss": 3.0861,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007819785149254532,
+      "loss": 3.0875,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007815469776386793,
+      "loss": 3.2169,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000781115133036933,
+      "loss": 2.9995,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007806829815915831,
+      "loss": 3.166,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007802505237743323,
+      "loss": 3.0596,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007798177600572183,
+      "loss": 3.2259,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007793846909126126,
+      "loss": 3.1378,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007789513168132205,
+      "loss": 3.1834,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007785176382320793,
+      "loss": 3.1325,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007780836556425592,
+      "loss": 3.049,
+      "step": 1033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007776493695183622,
+      "loss": 3.1205,
+      "step": 1034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007772147803335217,
+      "loss": 3.0787,
+      "step": 1035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007767798885624013,
+      "loss": 3.0951,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007763446946796957,
+      "loss": 3.0578,
+      "step": 1037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000775909199160429,
+      "loss": 3.2135,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007754734024799544,
+      "loss": 3.0614,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007750373051139538,
+      "loss": 3.2251,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007746009075384376,
+      "loss": 3.1591,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007741642102297439,
+      "loss": 3.047,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007737272136645377,
+      "loss": 3.1564,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007732899183198106,
+      "loss": 3.1602,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007728523246728807,
+      "loss": 3.1825,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007724144332013913,
+      "loss": 3.1697,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007719762443833114,
+      "loss": 3.0913,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007715377586969334,
+      "loss": 3.1,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007710989766208751,
+      "loss": 3.1468,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706598986340769,
+      "loss": 3.1147,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007702205252158024,
+      "loss": 3.1456,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007697808568456379,
+      "loss": 3.077,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007693408940034917,
+      "loss": 3.0937,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007689006371695928,
+      "loss": 3.0653,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007684600868244919,
+      "loss": 3.0712,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007680192434490598,
+      "loss": 3.1854,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007675781075244872,
+      "loss": 3.1133,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007671366795322839,
+      "loss": 3.2718,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007666949599542788,
+      "loss": 3.1929,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007662529492726187,
+      "loss": 3.1687,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007658106479697689,
+      "loss": 3.1177,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007653680565285106,
+      "loss": 3.1298,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007649251754319432,
+      "loss": 3.16,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007644820051634812,
+      "loss": 3.0801,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007640385462068553,
+      "loss": 3.1097,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000763594799046111,
+      "loss": 3.085,
+      "step": 1066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007631507641656084,
+      "loss": 3.1306,
+      "step": 1067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007627064420500216,
+      "loss": 3.0576,
+      "step": 1068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762261833184339,
+      "loss": 3.1704,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007618169380538605,
+      "loss": 3.1164,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007613717571442,
+      "loss": 3.1107,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007609262909412824,
+      "loss": 3.1444,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007604805399313441,
+      "loss": 3.1992,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007600345046009329,
+      "loss": 3.1339,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007595881854369065,
+      "loss": 3.1166,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007591415829264321,
+      "loss": 3.1498,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007586946975569871,
+      "loss": 3.0164,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007582475298163565,
+      "loss": 3.0314,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007578000801926345,
+      "loss": 3.1255,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007573523491742228,
+      "loss": 3.0196,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007569043372498297,
+      "loss": 3.1274,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007564560449084706,
+      "loss": 3.1706,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000756007472639467,
+      "loss": 3.1046,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007555586209324455,
+      "loss": 3.0539,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007551094902773383,
+      "loss": 3.1828,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007546600811643815,
+      "loss": 3.0649,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542103940841158,
+      "loss": 3.0852,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007537604295273847,
+      "loss": 3.0199,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007533101879853347,
+      "loss": 3.0488,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000752859669949415,
+      "loss": 3.0799,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007524088759113764,
+      "loss": 3.0552,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519578063632706,
+      "loss": 3.0455,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007515064617974506,
+      "loss": 2.952,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007510548427065694,
+      "loss": 3.1681,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007506029495835793,
+      "loss": 2.8964,
+      "step": 1095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000750150782921732,
+      "loss": 3.0239,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007496983432145781,
+      "loss": 3.1688,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007492456309559656,
+      "loss": 3.1186,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007487926466400402,
+      "loss": 3.1467,
+      "step": 1099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000748339390761245,
+      "loss": 3.0971,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007478858638143189,
+      "loss": 3.0757,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007474320662942972,
+      "loss": 2.9974,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007469779986965099,
+      "loss": 3.0129,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465236615165825,
+      "loss": 3.0395,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007460690552504346,
+      "loss": 3.1422,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007456141803942786,
+      "loss": 3.0378,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007451590374446217,
+      "loss": 3.0111,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007447036268982623,
+      "loss": 3.1783,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007442479492522916,
+      "loss": 3.1422,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007437920050040921,
+      "loss": 3.0651,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007433357946513377,
+      "loss": 3.1142,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742879318691992,
+      "loss": 3.0761,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007424225776243092,
+      "loss": 3.1252,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007419655719468326,
+      "loss": 3.0767,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007415083021583943,
+      "loss": 3.0959,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007410507687581149,
+      "loss": 3.1198,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007405929722454026,
+      "loss": 3.2023,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007401349131199525,
+      "loss": 3.0741,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007396765918817467,
+      "loss": 3.0279,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007392180090310535,
+      "loss": 3.0025,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007387591650684267,
+      "loss": 3.0746,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007383000604947045,
+      "loss": 2.9987,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007378406958110104,
+      "loss": 3.0962,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007373810715187516,
+      "loss": 3.1824,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007369211881196181,
+      "loss": 3.1199,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007364610461155835,
+      "loss": 3.0521,
+      "step": 1126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007360006460089032,
+      "loss": 3.1112,
+      "step": 1127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007355399883021144,
+      "loss": 3.1647,
+      "step": 1128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007350790734980358,
+      "loss": 3.0312,
+      "step": 1129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007346179020997662,
+      "loss": 3.1435,
+      "step": 1130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007341564746106848,
+      "loss": 3.137,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007336947915344503,
+      "loss": 3.1715,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007332328533750004,
+      "loss": 3.0619,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007327706606365512,
+      "loss": 3.1128,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007323082138235965,
+      "loss": 3.071,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007318455134409076,
+      "loss": 3.1462,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007313825599935326,
+      "loss": 3.1969,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007309193539867958,
+      "loss": 3.1087,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007304558959262972,
+      "loss": 3.1813,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000729992186317912,
+      "loss": 3.0638,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295282256677896,
+      "loss": 3.1572,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007290640144823535,
+      "loss": 3.1287,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007285995532683016,
+      "loss": 3.0065,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007281348425326033,
+      "loss": 3.0787,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000727669882782501,
+      "loss": 3.1719,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007272046745255093,
+      "loss": 3.0915,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007267392182694134,
+      "loss": 3.0758,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007262735145222696,
+      "loss": 3.088,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007258075637924039,
+      "loss": 3.0903,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007253413665884124,
+      "loss": 3.0108,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007248749234191602,
+      "loss": 3.2068,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007244082347937804,
+      "loss": 3.1151,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007239413012216744,
+      "loss": 3.1276,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000723474123212511,
+      "loss": 3.0918,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230067012762257,
+      "loss": 3.0569,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007225390359230202,
+      "loss": 3.0195,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007220711276633618,
+      "loss": 3.1027,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216029770079832,
+      "loss": 3.1856,
+      "step": 1158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007211345844678817,
+      "loss": 3.1284,
+      "step": 1159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000720665950554318,
+      "loss": 3.1003,
+      "step": 1160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007201970757788173,
+      "loss": 3.0087,
+      "step": 1161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007197279606531668,
+      "loss": 3.0512,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007192586056894164,
+      "loss": 3.0592,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007187890113998782,
+      "loss": 2.9189,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007183191782971248,
+      "loss": 3.0864,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007178491068939896,
+      "loss": 3.1149,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007173787977035667,
+      "loss": 3.207,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007169082512392092,
+      "loss": 3.0532,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007164374680145293,
+      "loss": 3.1443,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159664485433976,
+      "loss": 3.0538,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007154951933399426,
+      "loss": 3.1209,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007150237029185505,
+      "loss": 3.0327,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007145519777938635,
+      "loss": 3.0307,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007140800184807806,
+      "loss": 3.1571,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007136078254944559,
+      "loss": 3.0505,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007131353993502993,
+      "loss": 3.1092,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007126627405639744,
+      "loss": 3.0155,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007121898496513994,
+      "loss": 3.177,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007117167271287453,
+      "loss": 3.144,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000711243373512436,
+      "loss": 3.0596,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007107697893191482,
+      "loss": 3.1476,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007102959750658098,
+      "loss": 3.1964,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007098219312695998,
+      "loss": 3.0527,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007093476584479479,
+      "loss": 2.9954,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007088731571185336,
+      "loss": 3.0227,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007083984277992862,
+      "loss": 3.0334,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007079234710083836,
+      "loss": 3.0424,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007074482872642518,
+      "loss": 3.2096,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007069728770855651,
+      "loss": 2.9819,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007064972409912443,
+      "loss": 3.1233,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007060213795004571,
+      "loss": 3.0901,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007055452931326172,
+      "loss": 3.1231,
+      "step": 1192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000705068982407384,
+      "loss": 3.0763,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007045924478446612,
+      "loss": 3.0411,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007041156899645976,
+      "loss": 3.1527,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007036387092875849,
+      "loss": 3.0503,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007031615063342587,
+      "loss": 3.0452,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007026840816254969,
+      "loss": 3.1566,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007022064356824196,
+      "loss": 2.98,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007017285690263882,
+      "loss": 3.0441,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000701250482179005,
+      "loss": 3.0945,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007007721756621129,
+      "loss": 3.1214,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007002936499977944,
+      "loss": 3.0313,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000699814905708371,
+      "loss": 3.0989,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006993359433164032,
+      "loss": 3.0662,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006988567633446896,
+      "loss": 3.0349,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006983773663162658,
+      "loss": 3.0612,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006978977527544048,
+      "loss": 3.1051,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006974179231826157,
+      "loss": 3.1274,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006969378781246436,
+      "loss": 3.0939,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006964576181044686,
+      "loss": 3.0474,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006959771436463054,
+      "loss": 3.0022,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000695496455274603,
+      "loss": 3.1512,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006950155535140438,
+      "loss": 2.9828,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000694534438889543,
+      "loss": 3.2127,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006940531119262482,
+      "loss": 3.0085,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006935715731495389,
+      "loss": 2.984,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006930898230850255,
+      "loss": 3.1242,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006926078622585495,
+      "loss": 3.1205,
+      "step": 1219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006921256911961819,
+      "loss": 2.9006,
+      "step": 1220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006916433104242237,
+      "loss": 2.9827,
+      "step": 1221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006911607204692045,
+      "loss": 2.9848,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006906779218578822,
+      "loss": 3.1294,
+      "step": 1223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006901949151172426,
+      "loss": 3.0188,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006897117007744985,
+      "loss": 3.0948,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006892282793570896,
+      "loss": 3.1303,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006887446513926813,
+      "loss": 2.9729,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006882608174091647,
+      "loss": 3.1655,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006877767779346556,
+      "loss": 3.086,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006872925334974942,
+      "loss": 3.046,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006868080846262443,
+      "loss": 3.1174,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000686323431849693,
+      "loss": 3.121,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006858385756968501,
+      "loss": 3.1125,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006853535166969469,
+      "loss": 3.0082,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006848682553794364,
+      "loss": 3.1261,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006843827922739928,
+      "loss": 3.0955,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006838971279105098,
+      "loss": 2.9833,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006834112628191014,
+      "loss": 3.0071,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006829251975301003,
+      "loss": 2.996,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000682438932574058,
+      "loss": 3.0915,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006819524684817438,
+      "loss": 3.032,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006814658057841444,
+      "loss": 3.1378,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006809789450124633,
+      "loss": 3.0946,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006804918866981203,
+      "loss": 3.0112,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006800046313727503,
+      "loss": 3.1827,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000679517179568204,
+      "loss": 3.0607,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006790295318165461,
+      "loss": 3.0524,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006785416886500556,
+      "loss": 3.0564,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006780536506012238,
+      "loss": 3.0946,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006775654182027559,
+      "loss": 2.9987,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006770769919875687,
+      "loss": 2.9657,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006765883724887905,
+      "loss": 3.016,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006760995602397606,
+      "loss": 3.1252,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006756105557740289,
+      "loss": 3.1555,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000675121359625355,
+      "loss": 2.9418,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006746319723277076,
+      "loss": 3.1459,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006741423944152641,
+      "loss": 3.0297,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006736526264224101,
+      "loss": 3.0183,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006731626688837387,
+      "loss": 3.0877,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006726725223340494,
+      "loss": 3.0838,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006721821873083486,
+      "loss": 3.1321,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006716916643418483,
+      "loss": 3.1675,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006712009539699655,
+      "loss": 3.0546,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006707100567283216,
+      "loss": 3.1347,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006702189731527424,
+      "loss": 3.1976,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006697277037792568,
+      "loss": 3.2056,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006692362491440965,
+      "loss": 3.0932,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006687446097836956,
+      "loss": 3.0685,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006682527862346898,
+      "loss": 3.0554,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006677607790339156,
+      "loss": 3.0366,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00066726858871841,
+      "loss": 3.0326,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006667762158254104,
+      "loss": 3.1387,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006662836608923526,
+      "loss": 3.0064,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000665790924456872,
+      "loss": 2.9997,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006652980070568015,
+      "loss": 3.1072,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006648049092301717,
+      "loss": 3.1586,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006643116315152102,
+      "loss": 2.9579,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006638181744503409,
+      "loss": 3.0929,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006633245385741837,
+      "loss": 3.0596,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006628307244255534,
+      "loss": 3.0813,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006623367325434591,
+      "loss": 3.0371,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006618425634671046,
+      "loss": 3.2216,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006613482177358866,
+      "loss": 3.0331,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006608536958893948,
+      "loss": 3.0069,
+      "step": 1284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006603589984674113,
+      "loss": 3.1178,
+      "step": 1285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006598641260099095,
+      "loss": 3.0886,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006593690790570538,
+      "loss": 3.1143,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006588738581491997,
+      "loss": 3.0963,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000658378463826892,
+      "loss": 3.1132,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006578828966308649,
+      "loss": 3.0465,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006573871571020412,
+      "loss": 3.2051,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006568912457815321,
+      "loss": 3.0496,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006563951632106365,
+      "loss": 3.082,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006558989099308394,
+      "loss": 3.0829,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006554024864838126,
+      "loss": 3.084,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006549058934114141,
+      "loss": 2.9944,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006544091312556861,
+      "loss": 3.0538,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006539122005588562,
+      "loss": 3.1085,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006534151018633355,
+      "loss": 3.1244,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006529178357117186,
+      "loss": 2.9979,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000652420402646783,
+      "loss": 3.2041,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006519228032114882,
+      "loss": 3.1081,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006514250379489753,
+      "loss": 2.9579,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006509271074025668,
+      "loss": 2.9982,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000650429012115765,
+      "loss": 3.1354,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006499307526322523,
+      "loss": 3.0675,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006494323294958902,
+      "loss": 3.0228,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006489337432507194,
+      "loss": 3.0756,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006484349944409579,
+      "loss": 3.0405,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006479360836110016,
+      "loss": 3.0741,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006474370113054229,
+      "loss": 3.0694,
+      "step": 1311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006469377780689708,
+      "loss": 3.0888,
+      "step": 1312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006464383844465698,
+      "loss": 3.1279,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006459388309833191,
+      "loss": 3.056,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006454391182244936,
+      "loss": 2.9782,
+      "step": 1315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006449392467155403,
+      "loss": 3.0448,
+      "step": 1316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000644439217002081,
+      "loss": 2.9042,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006439390296299095,
+      "loss": 3.1383,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006434386851449914,
+      "loss": 3.0881,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006429381840934643,
+      "loss": 2.9917,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000642437527021637,
+      "loss": 3.0037,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006419367144759876,
+      "loss": 2.9882,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006414357470031647,
+      "loss": 3.0614,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006409346251499858,
+      "loss": 3.0582,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006404333494634366,
+      "loss": 3.1805,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006399319204906715,
+      "loss": 3.0535,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006394303387790112,
+      "loss": 3.0443,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000638928604875944,
+      "loss": 3.1679,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006384267193291238,
+      "loss": 3.0605,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00063792468268637,
+      "loss": 3.085,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006374224954956673,
+      "loss": 3.051,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006369201583051645,
+      "loss": 3.1854,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006364176716631739,
+      "loss": 3.0971,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006359150361181715,
+      "loss": 3.0792,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006354122522187952,
+      "loss": 3.0754,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006349093205138454,
+      "loss": 3.0266,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006344062415522836,
+      "loss": 3.0694,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006339030158832318,
+      "loss": 3.0421,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006333996440559725,
+      "loss": 3.1528,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006328961266199476,
+      "loss": 3.0281,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006323924641247575,
+      "loss": 3.0623,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006318886571201618,
+      "loss": 3.131,
+      "step": 1342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006313847061560772,
+      "loss": 3.0032,
+      "step": 1343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006308806117825777,
+      "loss": 3.0947,
+      "step": 1344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000630376374549894,
+      "loss": 3.11,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006298719950084124,
+      "loss": 3.0504,
+      "step": 1346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006293674737086748,
+      "loss": 2.9727,
+      "step": 1347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006288628112013779,
+      "loss": 3.0949,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006283580080373721,
+      "loss": 3.1186,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006278530647676617,
+      "loss": 3.1722,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000627347981943404,
+      "loss": 3.0309,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006268427601159082,
+      "loss": 2.979,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006263373998366358,
+      "loss": 3.0596,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000625831901657199,
+      "loss": 3.1554,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006253262661293602,
+      "loss": 3.0397,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006248204938050329,
+      "loss": 3.0766,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006243145852362788,
+      "loss": 3.066,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006238085409753085,
+      "loss": 3.0168,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006233023615744812,
+      "loss": 3.1468,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006227960475863033,
+      "loss": 3.0554,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006222895995634279,
+      "loss": 3.096,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006217830180586548,
+      "loss": 3.009,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006212763036249293,
+      "loss": 3.0356,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006207694568153417,
+      "loss": 3.1752,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006202624781831269,
+      "loss": 3.0707,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006197553682816637,
+      "loss": 3.0132,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006192481276644742,
+      "loss": 3.1999,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006187407568852231,
+      "loss": 3.0769,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006182332564977173,
+      "loss": 3.159,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006177256270559052,
+      "loss": 3.1692,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006172178691138756,
+      "loss": 3.0733,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006167099832258583,
+      "loss": 2.9234,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000616201969946222,
+      "loss": 3.1062,
+      "step": 1373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156938298294752,
+      "loss": 3.0761,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006151855634302643,
+      "loss": 3.1389,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006146771713033736,
+      "loss": 3.0394,
+      "step": 1376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000614168654003725,
+      "loss": 3.1137,
+      "step": 1377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006136600120863766,
+      "loss": 3.0202,
+      "step": 1378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006131512461065227,
+      "loss": 3.0739,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000612642356619493,
+      "loss": 2.9689,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121333441807522,
+      "loss": 3.0292,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006116242093458988,
+      "loss": 3.0665,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006111149526706653,
+      "loss": 3.2103,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006106055747109169,
+      "loss": 3.1359,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006100960760226515,
+      "loss": 3.044,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006095864571619982,
+      "loss": 3.0796,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006090767186852182,
+      "loss": 3.0373,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006085668611487022,
+      "loss": 2.9907,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080568851089717,
+      "loss": 3.0673,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000607546791122677,
+      "loss": 3.0096,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006070365797465976,
+      "loss": 2.9772,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006065262515376407,
+      "loss": 3.0824,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006060158070528413,
+      "loss": 3.0273,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006055052468493614,
+      "loss": 3.1353,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006049945714844888,
+      "loss": 3.0389,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006044837815156376,
+      "loss": 3.1989,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006039728775003467,
+      "loss": 3.1275,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006034618599962796,
+      "loss": 3.1088,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006029507295612234,
+      "loss": 3.074,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006024394867530889,
+      "loss": 3.0859,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006019281321299092,
+      "loss": 3.0061,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006014166662498396,
+      "loss": 3.0759,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006009050896711569,
+      "loss": 3.0543,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006003934029522584,
+      "loss": 3.0918,
+      "step": 1404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005998816066516622,
+      "loss": 3.0662,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005993697013280055,
+      "loss": 3.1404,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005988576875400447,
+      "loss": 3.0755,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005983455658466547,
+      "loss": 3.0057,
+      "step": 1408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005978333368068279,
+      "loss": 3.1473,
+      "step": 1409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005973210009796739,
+      "loss": 3.1848,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005968085589244192,
+      "loss": 3.1033,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000596296011200406,
+      "loss": 3.1661,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000595783358367092,
+      "loss": 3.0962,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000595270600984049,
+      "loss": 3.0928,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005947577396109641,
+      "loss": 3.0603,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005942447748076368,
+      "loss": 3.177,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005937317071339801,
+      "loss": 3.1661,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005932185371500189,
+      "loss": 2.9701,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005927052654158902,
+      "loss": 3.0091,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005921918924918419,
+      "loss": 3.0127,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005916784189382322,
+      "loss": 2.9747,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005911648453155292,
+      "loss": 3.1756,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005906511721843102,
+      "loss": 3.1236,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005901374001052614,
+      "loss": 3.0054,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005896235296391765,
+      "loss": 3.1186,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005891095613469572,
+      "loss": 3.1214,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005885954957896114,
+      "loss": 3.065,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005880813335282536,
+      "loss": 3.0566,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005875670751241035,
+      "loss": 3.0606,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005870527211384862,
+      "loss": 3.0932,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005865382721328308,
+      "loss": 3.1343,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005860237286686698,
+      "loss": 3.0928,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005855090913076395,
+      "loss": 3.1559,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005849943606114781,
+      "loss": 2.933,
+      "step": 1434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005844795371420264,
+      "loss": 3.0358,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005839646214612257,
+      "loss": 2.9998,
+      "step": 1436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005834496141311181,
+      "loss": 3.0734,
+      "step": 1437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005829345157138462,
+      "loss": 3.1063,
+      "step": 1438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005824193267716517,
+      "loss": 3.0889,
+      "step": 1439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005819040478668748,
+      "loss": 3.1706,
+      "step": 1440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005813886795619546,
+      "loss": 2.9339,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000580873222419427,
+      "loss": 3.1482,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005803576770019259,
+      "loss": 3.0178,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005798420438721804,
+      "loss": 2.9704,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000579326323593016,
+      "loss": 3.1023,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005788105167273532,
+      "loss": 3.0428,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005782946238382069,
+      "loss": 3.2117,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005777786454886862,
+      "loss": 3.0794,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005772625822419933,
+      "loss": 3.0719,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005767464346614226,
+      "loss": 3.1018,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005762302033103616,
+      "loss": 3.0715,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005757138887522884,
+      "loss": 3.0777,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005751974915507719,
+      "loss": 3.0715,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000574681012269472,
+      "loss": 3.0248,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005741644514721374,
+      "loss": 3.054,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005736478097226057,
+      "loss": 3.134,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005731310875848037,
+      "loss": 3.0249,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005726142856227452,
+      "loss": 2.9153,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005720974044005313,
+      "loss": 3.1941,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00057158044448235,
+      "loss": 2.8834,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005710634064324747,
+      "loss": 3.1203,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005705462908152641,
+      "loss": 3.0503,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005700290981951621,
+      "loss": 3.0421,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005695118291366959,
+      "loss": 3.13,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005689944842044768,
+      "loss": 3.0117,
+      "step": 1465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005684770639631986,
+      "loss": 3.0296,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005679595689776372,
+      "loss": 3.0899,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005674419998126504,
+      "loss": 3.1906,
+      "step": 1468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005669243570331766,
+      "loss": 3.0326,
+      "step": 1469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005664066412042348,
+      "loss": 3.0663,
+      "step": 1470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000565888852890924,
+      "loss": 3.0769,
+      "step": 1471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005653709926584214,
+      "loss": 2.981,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005648530610719838,
+      "loss": 3.0556,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005643350586969452,
+      "loss": 3.0006,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000563816986098717,
+      "loss": 2.9804,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005632988438427875,
+      "loss": 3.0426,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005627806324947206,
+      "loss": 3.0099,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005622623526201561,
+      "loss": 3.0803,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005617440047848081,
+      "loss": 3.0705,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005612255895544651,
+      "loss": 3.0752,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005607071074949894,
+      "loss": 3.061,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005601885591723162,
+      "loss": 2.9979,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005596699451524523,
+      "loss": 3.0146,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005591512660014773,
+      "loss": 2.9966,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005586325222855411,
+      "loss": 3.0872,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005581137145708643,
+      "loss": 3.0498,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005575948434237378,
+      "loss": 3.043,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005570759094105209,
+      "loss": 3.0099,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005565569130976422,
+      "loss": 3.0225,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000556037855051598,
+      "loss": 3.0615,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005555187358389518,
+      "loss": 2.992,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005549995560263345,
+      "loss": 2.9793,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005544803161804425,
+      "loss": 2.9814,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005539610168680381,
+      "loss": 2.9823,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005534416586559483,
+      "loss": 3.0588,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005529222421110643,
+      "loss": 3.1007,
+      "step": 1496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005524027678003415,
+      "loss": 3.0229,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005518832362907982,
+      "loss": 3.0701,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005513636481495142,
+      "loss": 3.1175,
+      "step": 1499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005508440039436325,
+      "loss": 3.0859,
+      "step": 1500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005503243042403564,
+      "loss": 3.0346,
+      "step": 1501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005498045496069499,
+      "loss": 3.0914,
+      "step": 1502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005492847406107373,
+      "loss": 3.0146,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000548764877819102,
+      "loss": 3.0938,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005482449617994859,
+      "loss": 3.1775,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005477249931193897,
+      "loss": 2.996,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005472049723463704,
+      "loss": 3.144,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005466849000480432,
+      "loss": 3.0742,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000546164776792079,
+      "loss": 3.0056,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005456446031462036,
+      "loss": 3.0698,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005451243796781992,
+      "loss": 3.0242,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005446041069559009,
+      "loss": 3.0596,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005440837855471988,
+      "loss": 2.9423,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005435634160200354,
+      "loss": 3.137,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005430429989424062,
+      "loss": 3.0918,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005425225348823581,
+      "loss": 2.9521,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005420020244079895,
+      "loss": 3.1038,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005414814680874497,
+      "loss": 3.0777,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005409608664889376,
+      "loss": 3.116,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005404402201807021,
+      "loss": 3.1519,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005399195297310403,
+      "loss": 2.976,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005393987957082979,
+      "loss": 2.9668,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005388780186808678,
+      "loss": 3.0607,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005383571992171904,
+      "loss": 2.9927,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005378363378857517,
+      "loss": 3.0627,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000537315435255084,
+      "loss": 3.0229,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005367944918937644,
+      "loss": 3.0677,
+      "step": 1527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005362735083704144,
+      "loss": 3.0336,
+      "step": 1528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005357524852536995,
+      "loss": 3.0344,
+      "step": 1529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005352314231123284,
+      "loss": 3.0691,
+      "step": 1530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005347103225150525,
+      "loss": 3.0995,
+      "step": 1531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005341891840306646,
+      "loss": 3.043,
+      "step": 1532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005336680082279996,
+      "loss": 3.1109,
+      "step": 1533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005331467956759331,
+      "loss": 3.1535,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00053262554694338,
+      "loss": 3.1274,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005321042625992955,
+      "loss": 3.082,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005315829432126732,
+      "loss": 2.9839,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005310615893525454,
+      "loss": 3.0494,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005305402015879817,
+      "loss": 3.16,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005300187804880886,
+      "loss": 3.069,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005294973266220094,
+      "loss": 3.1059,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005289758405589228,
+      "loss": 3.0879,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005284543228680428,
+      "loss": 3.0113,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005279327741186179,
+      "loss": 2.9862,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005274111948799306,
+      "loss": 2.989,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005268895857212967,
+      "loss": 2.9869,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005263679472120639,
+      "loss": 3.0002,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005258462799216134,
+      "loss": 3.1005,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005253245844193564,
+      "loss": 3.0549,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005248028612747358,
+      "loss": 3.0631,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005242811110572242,
+      "loss": 3.0545,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005237593343363239,
+      "loss": 3.012,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005232375316815663,
+      "loss": 3.1145,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005227157036625107,
+      "loss": 2.9893,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005221938508487445,
+      "loss": 3.1043,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005216719738098822,
+      "loss": 3.0395,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005211500731155641,
+      "loss": 3.0252,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000520628149335457,
+      "loss": 3.208,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005201062030392527,
+      "loss": 2.9452,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005195842347966673,
+      "loss": 2.9919,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005190622451774414,
+      "loss": 3.2363,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005185402347513382,
+      "loss": 2.9515,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005180182040881444,
+      "loss": 3.1865,
+      "step": 1563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005174961537576685,
+      "loss": 3.0288,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005169740843297399,
+      "loss": 3.0286,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005164519963742097,
+      "loss": 2.978,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005159298904609489,
+      "loss": 3.1858,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005154077671598479,
+      "loss": 3.1213,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005148856270408163,
+      "loss": 2.9191,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000514363470673782,
+      "loss": 3.1281,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005138412986286905,
+      "loss": 3.0459,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005133191114755046,
+      "loss": 3.1015,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005127969097842037,
+      "loss": 2.9895,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005122746941247828,
+      "loss": 2.8967,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005117524650672523,
+      "loss": 3.0047,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005112302231816369,
+      "loss": 3.0885,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000510707969037976,
+      "loss": 2.9419,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005101857032063215,
+      "loss": 2.9858,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005096634262567388,
+      "loss": 3.0178,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005091411387593051,
+      "loss": 3.0522,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005086188412841093,
+      "loss": 3.0727,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005080965344012508,
+      "loss": 2.9431,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005075742186808396,
+      "loss": 3.1487,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005070518946929954,
+      "loss": 3.0792,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005065295630078465,
+      "loss": 3.038,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005060072241955304,
+      "loss": 3.1427,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005054848788261912,
+      "loss": 2.9739,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005049625274699814,
+      "loss": 3.0901,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005044401706970591,
+      "loss": 3.0522,
+      "step": 1589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005039178090775889,
+      "loss": 3.0714,
+      "step": 1590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005033954431817405,
+      "loss": 3.1198,
+      "step": 1591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005028730735796882,
+      "loss": 3.0186,
+      "step": 1592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005023507008416099,
+      "loss": 2.956,
+      "step": 1593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005018283255376882,
+      "loss": 3.0333,
+      "step": 1594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005013059482381071,
+      "loss": 3.0538,
+      "step": 1595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005007835695130536,
+      "loss": 3.0524,
+      "step": 1596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005002611899327161,
+      "loss": 2.994,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004997388100672839,
+      "loss": 3.0579,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004992164304869464,
+      "loss": 3.1068,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004986940517618929,
+      "loss": 2.9313,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000498171674462312,
+      "loss": 3.0254,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004976492991583901,
+      "loss": 2.9862,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497126926420312,
+      "loss": 3.0674,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004966045568182596,
+      "loss": 3.0076,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000496082190922411,
+      "loss": 3.0272,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000495559829302941,
+      "loss": 2.9872,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004950374725300187,
+      "loss": 2.9138,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004945151211738089,
+      "loss": 3.0225,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004939927758044698,
+      "loss": 3.0014,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004934704369921535,
+      "loss": 3.1137,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004929481053070045,
+      "loss": 3.1653,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004924257813191605,
+      "loss": 3.0077,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004919034655987492,
+      "loss": 3.0573,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004913811587158907,
+      "loss": 3.0242,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004908588612406949,
+      "loss": 2.9322,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004903365737432612,
+      "loss": 3.0584,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004898142967936787,
+      "loss": 3.0416,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004892920309620243,
+      "loss": 3.0031,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004887697768183632,
+      "loss": 3.0567,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048824753493274787,
+      "loss": 3.0432,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004877253058752172,
+      "loss": 3.0726,
+      "step": 1621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048720309021579617,
+      "loss": 2.9625,
+      "step": 1622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048668088852449543,
+      "loss": 3.0257,
+      "step": 1623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004861587013713096,
+      "loss": 3.0122,
+      "step": 1624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004856365293262181,
+      "loss": 3.0537,
+      "step": 1625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004851143729591838,
+      "loss": 2.9975,
+      "step": 1626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048459223284015216,
+      "loss": 2.9803,
+      "step": 1627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004840701095390512,
+      "loss": 3.0082,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048354800362579035,
+      "loss": 3.0969,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048302591567026015,
+      "loss": 2.9955,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004825038462423317,
+      "loss": 3.0114,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048198179591185555,
+      "loss": 3.0072,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048145976524866175,
+      "loss": 3.0886,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048093775482255886,
+      "loss": 2.9942,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004804157652033328,
+      "loss": 3.109,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004798937969607474,
+      "loss": 3.1006,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047937185066454307,
+      "loss": 3.1287,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004788499268844359,
+      "loss": 2.9758,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047832802619011787,
+      "loss": 2.9055,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004778061491512555,
+      "loss": 3.1382,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047728429633748934,
+      "loss": 3.1108,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047676246831843377,
+      "loss": 3.0322,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004762406656636761,
+      "loss": 3.0995,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000475718888942776,
+      "loss": 3.1605,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047519713872526436,
+      "loss": 3.0694,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004746754155806437,
+      "loss": 3.1324,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004741537200783867,
+      "loss": 3.1137,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047363205278793606,
+      "loss": 3.0662,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004731104142787035,
+      "loss": 3.0213,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004725888051200695,
+      "loss": 3.0387,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004720672258813822,
+      "loss": 3.0977,
+      "step": 1651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004715456771319573,
+      "loss": 3.03,
+      "step": 1652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004710241594410773,
+      "loss": 2.9586,
+      "step": 1653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004705026733779907,
+      "loss": 2.9972,
+      "step": 1654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046998121951191164,
+      "loss": 3.0412,
+      "step": 1655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004694597984120185,
+      "loss": 3.0339,
+      "step": 1656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046893841064745475,
+      "loss": 3.0468,
+      "step": 1657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046841705678732685,
+      "loss": 2.8813,
+      "step": 1658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004678957374007046,
+      "loss": 3.0304,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046737445305662005,
+      "loss": 3.0143,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046685320432406704,
+      "loss": 3.1202,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004663319917720004,
+      "loss": 3.1355,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046581081596933543,
+      "loss": 3.0634,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046528967748494767,
+      "loss": 3.1022,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046476857688767154,
+      "loss": 3.0691,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004642475147463007,
+      "loss": 3.0587,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046372649162958575,
+      "loss": 3.0191,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046320550810623575,
+      "loss": 3.0348,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000462684564744916,
+      "loss": 2.8724,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004621636621142483,
+      "loss": 3.0413,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004616428007828097,
+      "loss": 3.1145,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004611219813191322,
+      "loss": 3.0672,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046060120429170214,
+      "loss": 2.9739,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046008047026895976,
+      "loss": 3.0421,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045955977981929796,
+      "loss": 2.9575,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045903913351106235,
+      "loss": 3.024,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045851853191255056,
+      "loss": 3.0853,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004579979755920107,
+      "loss": 3.1108,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045747746511764207,
+      "loss": 2.919,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045695700105759387,
+      "loss": 2.9118,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045643658397996443,
+      "loss": 3.0169,
+      "step": 1681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045591621445280117,
+      "loss": 2.9364,
+      "step": 1682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004553958930440991,
+      "loss": 2.9591,
+      "step": 1683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045487562032180096,
+      "loss": 3.0402,
+      "step": 1684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004543553968537964,
+      "loss": 2.9278,
+      "step": 1685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045383522320792115,
+      "loss": 3.1347,
+      "step": 1686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004533150999519567,
+      "loss": 3.097,
+      "step": 1687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045279502765362974,
+      "loss": 3.1253,
+      "step": 1688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045227500688061064,
+      "loss": 3.1184,
+      "step": 1689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004517550382005142,
+      "loss": 3.0718,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045123512218089813,
+      "loss": 3.0429,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004507152593892627,
+      "loss": 3.073,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045019545039305013,
+      "loss": 3.0526,
+      "step": 1693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044967569575964377,
+      "loss": 3.0145,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044915599605636756,
+      "loss": 2.9175,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044863635185048585,
+      "loss": 3.0591,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000448116763709202,
+      "loss": 3.0837,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044759723219965836,
+      "loss": 2.8756,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004470777578889358,
+      "loss": 3.0695,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044655834134405193,
+      "loss": 3.0251,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004460389831319621,
+      "loss": 2.9982,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004455196838195576,
+      "loss": 2.9777,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004450004439736656,
+      "loss": 3.0557,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044448126416104817,
+      "loss": 3.122,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044396214494840224,
+      "loss": 3.0234,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004434430869023579,
+      "loss": 2.981,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004429240905894792,
+      "loss": 2.9933,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044240515657626225,
+      "loss": 3.0672,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044188628542913557,
+      "loss": 3.092,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004413674777144591,
+      "loss": 3.1125,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044084873399852286,
+      "loss": 3.0403,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004403300548475478,
+      "loss": 3.0858,
+      "step": 1712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000439811440827684,
+      "loss": 3.0594,
+      "step": 1713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004392928925050105,
+      "loss": 3.0819,
+      "step": 1714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043877441044553483,
+      "loss": 3.1082,
+      "step": 1715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043825599521519213,
+      "loss": 3.0391,
+      "step": 1716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043773764737984407,
+      "loss": 2.9452,
+      "step": 1717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004372193675052795,
+      "loss": 3.0538,
+      "step": 1718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004367011561572126,
+      "loss": 3.1534,
+      "step": 1719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043618301390128294,
+      "loss": 3.0767,
+      "step": 1720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043566494130305503,
+      "loss": 3.0509,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004351469389280163,
+      "loss": 3.1267,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043462900734157864,
+      "loss": 3.0583,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043411114710907616,
+      "loss": 2.9949,
+      "step": 1724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004335933587957651,
+      "loss": 3.0482,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043307564296682355,
+      "loss": 3.011,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043255800018734974,
+      "loss": 2.941,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043204043102236284,
+      "loss": 3.1476,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043152293603680145,
+      "loss": 2.9093,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004310055157955232,
+      "loss": 3.067,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043048817086330406,
+      "loss": 2.9741,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004299709018048381,
+      "loss": 3.0324,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000429453709184736,
+      "loss": 3.0182,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042893659356752547,
+      "loss": 3.0237,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042841955551765,
+      "loss": 2.9598,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004279025955994686,
+      "loss": 3.0957,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000427385714377255,
+      "loss": 3.0926,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004268689124151964,
+      "loss": 3.0598,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042635219027739433,
+      "loss": 2.8831,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004258355485278628,
+      "loss": 3.0757,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042531898773052806,
+      "loss": 3.0549,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000424802508449228,
+      "loss": 2.9274,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042428611124771184,
+      "loss": 3.1607,
+      "step": 1743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004237697966896385,
+      "loss": 3.0961,
+      "step": 1744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004232535653385774,
+      "loss": 3.0516,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004227374177580069,
+      "loss": 3.0271,
+      "step": 1746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042222135451131383,
+      "loss": 3.0806,
+      "step": 1747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042170537616179326,
+      "loss": 3.0571,
+      "step": 1748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000421189483272647,
+      "loss": 3.1212,
+      "step": 1749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004206736764069841,
+      "loss": 3.0898,
+      "step": 1750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042015795612781965,
+      "loss": 2.9249,
+      "step": 1751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041964232299807416,
+      "loss": 3.0087,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004191267775805728,
+      "loss": 3.0509,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004186113204380455,
+      "loss": 3.0686,
+      "step": 1754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004180959521331253,
+      "loss": 2.9948,
+      "step": 1755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041758067322834847,
+      "loss": 3.069,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004170654842861538,
+      "loss": 2.9485,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041655038586888184,
+      "loss": 2.987,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041603537853877453,
+      "loss": 2.921,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041552046285797374,
+      "loss": 3.0334,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041500563938852187,
+      "loss": 3.0094,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004144909086923606,
+      "loss": 3.0848,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004139762713313304,
+      "loss": 3.1046,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041346172786716943,
+      "loss": 3.0315,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041294727886151384,
+      "loss": 2.9769,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004124329248758966,
+      "loss": 2.8875,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041191866647174655,
+      "loss": 3.0335,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041140450421038864,
+      "loss": 3.0669,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004108904386530429,
+      "loss": 3.0423,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004103764703608237,
+      "loss": 3.0543,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040986259989473885,
+      "loss": 3.0163,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040934882781568994,
+      "loss": 2.9653,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040883515468447093,
+      "loss": 2.9816,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040832158106176783,
+      "loss": 3.0217,
+      "step": 1774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004078081075081581,
+      "loss": 3.0161,
+      "step": 1775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040729473458410984,
+      "loss": 3.0695,
+      "step": 1776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040678146284998126,
+      "loss": 2.9458,
+      "step": 1777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040626829286602006,
+      "loss": 3.0527,
+      "step": 1778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004057552251923633,
+      "loss": 3.0455,
+      "step": 1779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040524226038903596,
+      "loss": 3.0758,
+      "step": 1780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040472939901595113,
+      "loss": 2.945,
+      "step": 1781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040421664163290826,
+      "loss": 3.0661,
+      "step": 1782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040370398879959406,
+      "loss": 3.0918,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040319144107558085,
+      "loss": 2.9262,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004026789990203261,
+      "loss": 3.1366,
+      "step": 1785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004021666631931722,
+      "loss": 3.0151,
+      "step": 1786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004016544341533455,
+      "loss": 2.9829,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040114231245995537,
+      "loss": 3.1643,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040063029867199454,
+      "loss": 2.9312,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040011839334833777,
+      "loss": 3.0575,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003996065970477415,
+      "loss": 3.1644,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003990949103288433,
+      "loss": 2.98,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003985833337501605,
+      "loss": 3.0514,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039807186787009085,
+      "loss": 2.9997,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003975605132469111,
+      "loss": 3.0471,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003970492704387766,
+      "loss": 2.9109,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003965381400037205,
+      "loss": 3.0622,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003960271224996534,
+      "loss": 3.0571,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003955162184843625,
+      "loss": 2.9172,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039500542851551135,
+      "loss": 3.0112,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039449475315063876,
+      "loss": 2.9617,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003939841929471587,
+      "loss": 2.9458,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003934737484623595,
+      "loss": 2.9544,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003929634202534026,
+      "loss": 3.0041,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039245320887732306,
+      "loss": 3.055,
+      "step": 1805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003919431148910284,
+      "loss": 2.9519,
+      "step": 1806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003914331388512978,
+      "loss": 3.0726,
+      "step": 1807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039092328131478207,
+      "loss": 3.0479,
+      "step": 1808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039041354283800186,
+      "loss": 2.9681,
+      "step": 1809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038990392397734866,
+      "loss": 2.8479,
+      "step": 1810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003893944252890832,
+      "loss": 3.1741,
+      "step": 1811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003888850473293348,
+      "loss": 3.2186,
+      "step": 1812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003883757906541012,
+      "loss": 2.9676,
+      "step": 1813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038786665581924803,
+      "loss": 3.0731,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003873576433805071,
+      "loss": 3.0639,
+      "step": 1815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003868487538934774,
+      "loss": 3.0478,
+      "step": 1816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038633998791362344,
+      "loss": 3.1099,
+      "step": 1817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038583134599627503,
+      "loss": 3.1182,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003853228286966265,
+      "loss": 2.9793,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003848144365697358,
+      "loss": 2.9576,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003843061701705248,
+      "loss": 3.0191,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038379803005377807,
+      "loss": 2.9585,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003832900167741418,
+      "loss": 2.9305,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003827821308861244,
+      "loss": 3.0868,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003822743729440951,
+      "loss": 2.9702,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003817667435022828,
+      "loss": 2.9975,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038125924311477696,
+      "loss": 3.0134,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003807518723355259,
+      "loss": 2.9048,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038024463171833635,
+      "loss": 3.0188,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037973752181687335,
+      "loss": 2.9559,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003792305431846584,
+      "loss": 3.0226,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037872369637507075,
+      "loss": 3.0042,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037821698194134525,
+      "loss": 3.0322,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037771040043657204,
+      "loss": 3.0426,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037720395241369663,
+      "loss": 3.1529,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037669763842551883,
+      "loss": 3.0636,
+      "step": 1836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003761914590246915,
+      "loss": 3.1257,
+      "step": 1837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003756854147637213,
+      "loss": 2.9742,
+      "step": 1838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003751795061949671,
+      "loss": 2.9918,
+      "step": 1839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037467373387063967,
+      "loss": 3.0208,
+      "step": 1840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003741680983428013,
+      "loss": 2.9824,
+      "step": 1841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037366260016336435,
+      "loss": 3.053,
+      "step": 1842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003731572398840918,
+      "loss": 3.0878,
+      "step": 1843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037265201805659616,
+      "loss": 2.9306,
+      "step": 1844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037214693523233836,
+      "loss": 3.03,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037164199196262795,
+      "loss": 3.0958,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037113718879862236,
+      "loss": 3.0278,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037063252629132527,
+      "loss": 2.9306,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037012800499158766,
+      "loss": 2.9284,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036962362545010606,
+      "loss": 2.9922,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036911938821742224,
+      "loss": 2.9026,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000368615293843923,
+      "loss": 3.0902,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036811134287983836,
+      "loss": 3.0636,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003676075358752426,
+      "loss": 2.9908,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003671038733800526,
+      "loss": 3.0211,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003666003559440276,
+      "loss": 3.1285,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036609698411676817,
+      "loss": 2.9157,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003655937584477165,
+      "loss": 3.1373,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003650906794861546,
+      "loss": 2.9216,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003645877477812048,
+      "loss": 3.0186,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036408496388182855,
+      "loss": 2.9971,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036358232833682614,
+      "loss": 3.012,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003630798416948358,
+      "loss": 3.0446,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003625775045043328,
+      "loss": 3.0709,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003620753173136301,
+      "loss": 3.0212,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036157328067087626,
+      "loss": 3.0276,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000361071395124056,
+      "loss": 2.9619,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036056966122098864,
+      "loss": 2.9513,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003600680795093286,
+      "loss": 2.9513,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035956665053656333,
+      "loss": 3.0124,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035906537485001436,
+      "loss": 2.9435,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003585642529968353,
+      "loss": 3.0594,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035806328552401236,
+      "loss": 3.1515,
+      "step": 1873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003575624729783632,
+      "loss": 2.9218,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035706181590653563,
+      "loss": 3.0231,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035656131485500863,
+      "loss": 3.0318,
+      "step": 1876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003560609703700906,
+      "loss": 2.9199,
+      "step": 1877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000355560782997919,
+      "loss": 3.004,
+      "step": 1878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003550607532844596,
+      "loss": 2.9658,
+      "step": 1879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003545608817755065,
+      "loss": 3.0329,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035406116901668083,
+      "loss": 3.0939,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035356161555343034,
+      "loss": 3.0727,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003530622219310293,
+      "loss": 3.0371,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035256298869457717,
+      "loss": 3.0027,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003520639163889986,
+      "loss": 3.0154,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003515650055590422,
+      "loss": 2.9916,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035106625674928067,
+      "loss": 3.1125,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035056767050410973,
+      "loss": 2.9848,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003500692473677479,
+      "loss": 2.9393,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034957098788423525,
+      "loss": 3.0837,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003490728925974334,
+      "loss": 2.9663,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003485749620510247,
+      "loss": 3.0494,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034807719678851183,
+      "loss": 2.9739,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034757959735321695,
+      "loss": 2.9143,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003470821642882813,
+      "loss": 3.0686,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003465848981366647,
+      "loss": 2.9852,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000346087799441144,
+      "loss": 3.0318,
+      "step": 1897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000345590868744314,
+      "loss": 2.98,
+      "step": 1898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034509410658858605,
+      "loss": 3.0489,
+      "step": 1899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003445975135161873,
+      "loss": 2.9283,
+      "step": 1900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003441010900691608,
+      "loss": 3.0145,
+      "step": 1901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003436048367893636,
+      "loss": 3.1413,
+      "step": 1902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034310875421846785,
+      "loss": 3.0131,
+      "step": 1903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034261284289795885,
+      "loss": 3.0597,
+      "step": 1904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034211710336913526,
+      "loss": 2.9564,
+      "step": 1905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003416215361731081,
+      "loss": 2.9669,
+      "step": 1906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003411261418508005,
+      "loss": 3.0239,
+      "step": 1907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003406309209429463,
+      "loss": 3.0335,
+      "step": 1908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034013587399009073,
+      "loss": 2.9823,
+      "step": 1909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003396410015325888,
+      "loss": 2.9674,
+      "step": 1910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033914630411060516,
+      "loss": 3.074,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033865178226411355,
+      "loss": 2.9136,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033815743653289556,
+      "loss": 3.0276,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033766326745654105,
+      "loss": 2.9807,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003371692755744468,
+      "loss": 3.003,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033667546142581627,
+      "loss": 2.9641,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003361818255496589,
+      "loss": 3.034,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003356883684847899,
+      "loss": 3.0796,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033519509076982844,
+      "loss": 2.8993,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003347019929431986,
+      "loss": 2.9786,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003342090755431281,
+      "loss": 3.0816,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003337163391076473,
+      "loss": 2.8282,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003332237841745898,
+      "loss": 2.9438,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033273141128159,
+      "loss": 2.961,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003322392209660845,
+      "loss": 3.0489,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033174721376531027,
+      "loss": 3.0393,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033125539021630434,
+      "loss": 3.1161,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033076375085590336,
+      "loss": 2.9245,
+      "step": 1928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033027229622074335,
+      "loss": 2.9634,
+      "step": 1929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032978102684725767,
+      "loss": 2.9443,
+      "step": 1930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003292899432716785,
+      "loss": 2.9973,
+      "step": 1931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032879904603003455,
+      "loss": 3.0658,
+      "step": 1932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032830833565815164,
+      "loss": 2.9142,
+      "step": 1933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032781781269165144,
+      "loss": 3.0179,
+      "step": 1934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003273274776659507,
+      "loss": 3.0021,
+      "step": 1935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032683733111626143,
+      "loss": 2.9573,
+      "step": 1936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003263473735775899,
+      "loss": 3.0651,
+      "step": 1937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003258576055847359,
+      "loss": 3.0061,
+      "step": 1938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003253680276722924,
+      "loss": 2.9704,
+      "step": 1939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032487864037464515,
+      "loss": 3.0617,
+      "step": 1940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003243894442259712,
+      "loss": 3.0372,
+      "step": 1941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003239004397602394,
+      "loss": 2.9965,
+      "step": 1942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032341162751120956,
+      "loss": 3.0533,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003229230080124313,
+      "loss": 3.0163,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003224345817972442,
+      "loss": 3.0055,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032194634939877633,
+      "loss": 2.9956,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003214583113499446,
+      "loss": 3.0877,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003209704681834539,
+      "loss": 3.049,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000320482820431796,
+      "loss": 2.9062,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003199953686272497,
+      "loss": 3.035,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031950811330188,
+      "loss": 2.9661,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031902105498753683,
+      "loss": 2.9624,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003185341942158557,
+      "loss": 2.9594,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003180475315182563,
+      "loss": 3.0432,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031756106742594206,
+      "loss": 3.1525,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031707480246989986,
+      "loss": 2.9335,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031658873718089876,
+      "loss": 2.9949,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003161028720894902,
+      "loss": 2.9114,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003156172077260073,
+      "loss": 3.0832,
+      "step": 1959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031513174462056347,
+      "loss": 2.9536,
+      "step": 1960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031464648330305315,
+      "loss": 2.9074,
+      "step": 1961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003141614243031501,
+      "loss": 2.9378,
+      "step": 1962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000313676568150307,
+      "loss": 3.0663,
+      "step": 1963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031319191537375573,
+      "loss": 3.0681,
+      "step": 1964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031270746650250585,
+      "loss": 3.0036,
+      "step": 1965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003122232220653444,
+      "loss": 3.0603,
+      "step": 1966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031173918259083547,
+      "loss": 3.1021,
+      "step": 1967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031125534860731877,
+      "loss": 3.0143,
+      "step": 1968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031077172064291047,
+      "loss": 3.0942,
+      "step": 1969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031028829922550156,
+      "loss": 2.9399,
+      "step": 1970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030980508488275747,
+      "loss": 2.9959,
+      "step": 1971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003093220781421178,
+      "loss": 2.9112,
+      "step": 1972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003088392795307956,
+      "loss": 2.9099,
+      "step": 1973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030835668957577634,
+      "loss": 3.0439,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030787430880381805,
+      "loss": 3.014,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030739213774145054,
+      "loss": 2.9846,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030691017691497445,
+      "loss": 3.0015,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003064284268504613,
+      "loss": 3.1128,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003059468880737519,
+      "loss": 3.0578,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000305465561110457,
+      "loss": 3.0458,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003049844464859561,
+      "loss": 2.8905,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030450354472539697,
+      "loss": 3.0303,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030402285635369465,
+      "loss": 2.9889,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003035423818955316,
+      "loss": 3.0298,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030306212187535653,
+      "loss": 2.9803,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003025820768173843,
+      "loss": 2.9755,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003021022472455952,
+      "loss": 3.0368,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003016226336837342,
+      "loss": 2.9577,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003011432366553106,
+      "loss": 2.9414,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030066405668359686,
+      "loss": 3.0183,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003001850942916291,
+      "loss": 2.9249,
+      "step": 1991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002997063500022057,
+      "loss": 2.8891,
+      "step": 1992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002992278243378871,
+      "loss": 2.9473,
+      "step": 1993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002987495178209951,
+      "loss": 3.0543,
+      "step": 1994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029827143097361196,
+      "loss": 3.0664,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029779356431758054,
+      "loss": 2.9663,
+      "step": 1996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002973159183745031,
+      "loss": 2.948,
+      "step": 1997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002968384936657413,
+      "loss": 3.0375,
+      "step": 1998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002963612907124151,
+      "loss": 3.0224,
+      "step": 1999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029588431003540273,
+      "loss": 3.0302,
+      "step": 2000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000295407552155339,
+      "loss": 2.8375,
+      "step": 2001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002949310175926162,
+      "loss": 2.9326,
+      "step": 2002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029445470686738286,
+      "loss": 2.8996,
+      "step": 2003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029397862049954305,
+      "loss": 2.9643,
+      "step": 2004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029350275900875587,
+      "loss": 2.9842,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000293027122914435,
+      "loss": 3.083,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002925517127357482,
+      "loss": 2.9163,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002920765289916164,
+      "loss": 2.9816,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002916015722007136,
+      "loss": 3.0947,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002911268428814664,
+      "loss": 2.9812,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029065234155205223,
+      "loss": 3.0146,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002901780687304004,
+      "loss": 3.0945,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002897040249341904,
+      "loss": 2.9406,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028923021068085185,
+      "loss": 3.1377,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028875662648756407,
+      "loss": 3.007,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002882832728712551,
+      "loss": 3.0117,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002878101503486009,
+      "loss": 3.0507,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028733725943602575,
+      "loss": 3.0476,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002868646006497008,
+      "loss": 3.0376,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028639217450554413,
+      "loss": 2.9679,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002859199815192195,
+      "loss": 2.9964,
+      "step": 2021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028544802220613677,
+      "loss": 2.9891,
+      "step": 2022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028497629708144965,
+      "loss": 3.0143,
+      "step": 2023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002845048066600574,
+      "loss": 3.1311,
+      "step": 2024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002840335514566025,
+      "loss": 3.0746,
+      "step": 2025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028356253198547076,
+      "loss": 3.044,
+      "step": 2026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002830917487607909,
+      "loss": 3.0865,
+      "step": 2027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028262120229643334,
+      "loss": 2.9643,
+      "step": 2028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028215089310601037,
+      "loss": 2.9327,
+      "step": 2029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028168082170287526,
+      "loss": 2.9635,
+      "step": 2030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028121098860012174,
+      "loss": 2.9134,
+      "step": 2031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002807413943105835,
+      "loss": 2.8916,
+      "step": 2032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028027203934683333,
+      "loss": 3.0413,
+      "step": 2033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002798029242211828,
+      "loss": 2.9152,
+      "step": 2034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002793340494456822,
+      "loss": 3.0455,
+      "step": 2035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027886541553211864,
+      "loss": 2.8945,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002783970229920169,
+      "loss": 3.093,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027792887233663846,
+      "loss": 3.0055,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027746096407698,
+      "loss": 3.0259,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002769932987237744,
+      "loss": 3.017,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000276525876787489,
+      "loss": 3.0311,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027605869877832555,
+      "loss": 3.0381,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027559176520621957,
+      "loss": 2.9847,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027512507658083996,
+      "loss": 3.0596,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002746586334115877,
+      "loss": 3.0624,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027419243620759623,
+      "loss": 2.9991,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002737264854777306,
+      "loss": 2.9825,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002732607817305866,
+      "loss": 2.9672,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027279532547449083,
+      "loss": 3.0677,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000272330117217499,
+      "loss": 2.9827,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002718651574673968,
+      "loss": 2.8948,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002714004467316984,
+      "loss": 3.0818,
+      "step": 2052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027093598551764626,
+      "loss": 2.9807,
+      "step": 2053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002704717743322104,
+      "loss": 3.0181,
+      "step": 2054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027000781368208813,
+      "loss": 2.9183,
+      "step": 2055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026954410407370276,
+      "loss": 3.0111,
+      "step": 2056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026908064601320435,
+      "loss": 2.9956,
+      "step": 2057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026861744000646753,
+      "loss": 2.9386,
+      "step": 2058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002681544865590926,
+      "loss": 3.0226,
+      "step": 2059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026769178617640384,
+      "loss": 3.004,
+      "step": 2060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002672293393634491,
+      "loss": 3.0147,
+      "step": 2061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002667671466249997,
+      "loss": 3.0513,
+      "step": 2062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002663052084655497,
+      "loss": 2.974,
+      "step": 2063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002658435253893152,
+      "loss": 3.037,
+      "step": 2064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000265382097900234,
+      "loss": 2.9431,
+      "step": 2065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002649209265019644,
+      "loss": 2.9869,
+      "step": 2066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002644600116978857,
+      "loss": 3.1417,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026399935399109686,
+      "loss": 2.8785,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026353895388441656,
+      "loss": 2.953,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002630788118803819,
+      "loss": 3.0281,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026261892848124864,
+      "loss": 2.8884,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026215930418898957,
+      "loss": 2.9059,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026169993950529545,
+      "loss": 2.9845,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026124083493157335,
+      "loss": 3.0439,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026078199096894636,
+      "loss": 3.0791,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002603234081182533,
+      "loss": 3.0204,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025986508688004756,
+      "loss": 3.0038,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025940702775459747,
+      "loss": 2.9754,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000258949231241885,
+      "loss": 3.0829,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025849169784160575,
+      "loss": 3.0017,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025803442805316743,
+      "loss": 3.0307,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025757742237569105,
+      "loss": 3.2095,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025712068130800817,
+      "loss": 2.8445,
+      "step": 2083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025666420534866253,
+      "loss": 2.936,
+      "step": 2084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002562079949959079,
+      "loss": 2.9698,
+      "step": 2085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002557520507477085,
+      "loss": 2.9476,
+      "step": 2086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025529637310173794,
+      "loss": 2.9895,
+      "step": 2087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002548409625553785,
+      "loss": 2.9552,
+      "step": 2088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002543858196057214,
+      "loss": 2.992,
+      "step": 2089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025393094474956556,
+      "loss": 3.0614,
+      "step": 2090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002534763384834174,
+      "loss": 2.9832,
+      "step": 2091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025302200130348997,
+      "loss": 3.0143,
+      "step": 2092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025256793370570296,
+      "loss": 2.976,
+      "step": 2093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002521141361856811,
+      "loss": 2.9384,
+      "step": 2094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000251660609238755,
+      "loss": 2.9267,
+      "step": 2095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025120735335995976,
+      "loss": 2.982,
+      "step": 2096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002507543690440345,
+      "loss": 2.9642,
+      "step": 2097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025030165678542203,
+      "loss": 3.0299,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000249849217078268,
+      "loss": 2.9858,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024939705041642076,
+      "loss": 2.9526,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002489451572934306,
+      "loss": 2.9258,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024849353820254944,
+      "loss": 3.0276,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002480421936367294,
+      "loss": 3.0632,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002475911240886237,
+      "loss": 2.9666,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024714033005058514,
+      "loss": 3.0133,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002466898120146654,
+      "loss": 2.9171,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002462395704726155,
+      "loss": 3.0408,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024578960591588436,
+      "loss": 3.0178,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024533991883561866,
+      "loss": 2.9598,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002448905097226619,
+      "loss": 3.0697,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024444137906755466,
+      "loss": 3.0123,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024399252736053312,
+      "loss": 2.9767,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024354395509152937,
+      "loss": 2.9085,
+      "step": 2113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024309566275017026,
+      "loss": 3.0397,
+      "step": 2114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024264765082577733,
+      "loss": 3.0722,
+      "step": 2115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002421999198073655,
+      "loss": 2.9826,
+      "step": 2116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002417524701836436,
+      "loss": 3.1104,
+      "step": 2117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024130530244301308,
+      "loss": 3.003,
+      "step": 2118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024085841707356787,
+      "loss": 2.9723,
+      "step": 2119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024041181456309368,
+      "loss": 2.9896,
+      "step": 2120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002399654953990671,
+      "loss": 2.9113,
+      "step": 2121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023951946006865577,
+      "loss": 3.0495,
+      "step": 2122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023907370905871755,
+      "loss": 3.141,
+      "step": 2123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002386282428558001,
+      "loss": 3.1159,
+      "step": 2124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023818306194613954,
+      "loss": 3.0215,
+      "step": 2125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023773816681566117,
+      "loss": 3.0394,
+      "step": 2126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023729355794997848,
+      "loss": 3.0696,
+      "step": 2127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002368492358343919,
+      "loss": 3.0445,
+      "step": 2128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002364052009538892,
+      "loss": 2.8801,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023596145379314477,
+      "loss": 3.0546,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002355179948365189,
+      "loss": 3.0436,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023507482456805695,
+      "loss": 2.9476,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023463194347148942,
+      "loss": 3.0322,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002341893520302313,
+      "loss": 3.1251,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023374705072738122,
+      "loss": 2.9204,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023330504004572116,
+      "loss": 2.9108,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023286332046771618,
+      "loss": 2.9389,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002324218924755129,
+      "loss": 3.0271,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002319807565509402,
+      "loss": 2.9233,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002315399131755081,
+      "loss": 2.857,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023109936283040723,
+      "loss": 2.9899,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023065910599650853,
+      "loss": 2.9945,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002302191431543621,
+      "loss": 2.9589,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022977947478419758,
+      "loss": 2.9217,
+      "step": 2144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022934010136592314,
+      "loss": 2.9698,
+      "step": 2145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022890102337912489,
+      "loss": 2.9944,
+      "step": 2146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022846224130306664,
+      "loss": 2.8664,
+      "step": 2147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022802375561668876,
+      "loss": 2.9602,
+      "step": 2148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002275855667986086,
+      "loss": 3.0715,
+      "step": 2149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002271476753271194,
+      "loss": 2.9527,
+      "step": 2150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022671008168018947,
+      "loss": 3.0897,
+      "step": 2151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002262727863354624,
+      "loss": 2.8329,
+      "step": 2152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022583578977025632,
+      "loss": 2.9913,
+      "step": 2153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022539909246156254,
+      "loss": 3.0448,
+      "step": 2154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022496269488604637,
+      "loss": 2.9724,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002245265975200458,
+      "loss": 3.066,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022409080083957107,
+      "loss": 3.0518,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022365530532030442,
+      "loss": 3.0112,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002232201114375988,
+      "loss": 3.0275,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022278521966647848,
+      "loss": 2.9123,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022235063048163777,
+      "loss": 2.9829,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002219163443574408,
+      "loss": 3.0084,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002214823617679207,
+      "loss": 3.0088,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002210486831867796,
+      "loss": 3.0065,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022061530908738737,
+      "loss": 3.0751,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022018223994278182,
+      "loss": 3.0111,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021974947622566788,
+      "loss": 3.0092,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000219317018408417,
+      "loss": 2.8981,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021888486696306704,
+      "loss": 3.0211,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021845302236132082,
+      "loss": 3.0363,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002180214850745467,
+      "loss": 2.9035,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002175902555737778,
+      "loss": 3.0546,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021715933432971064,
+      "loss": 3.0502,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021672872181270576,
+      "loss": 2.9704,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021629841849278675,
+      "loss": 2.9581,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002158684248396392,
+      "loss": 2.9997,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021543874132261116,
+      "loss": 3.0065,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021500936841071212,
+      "loss": 2.9316,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021458030657261234,
+      "loss": 3.0218,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021415155627664285,
+      "loss": 2.9801,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021372311799079414,
+      "loss": 2.9763,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002132949921827165,
+      "loss": 3.0426,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000212867179319719,
+      "loss": 3.0166,
+      "step": 2183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021243967986876933,
+      "loss": 3.1126,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021201249429649282,
+      "loss": 2.9752,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002115856230691725,
+      "loss": 3.0076,
+      "step": 2186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021115906665274776,
+      "loss": 2.9811,
+      "step": 2187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002107328255128148,
+      "loss": 2.8905,
+      "step": 2188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021030690011462566,
+      "loss": 3.123,
+      "step": 2189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020988129092308765,
+      "loss": 2.9789,
+      "step": 2190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020945599840276307,
+      "loss": 3.0336,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002090310230178682,
+      "loss": 2.9602,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020860636523227345,
+      "loss": 2.9224,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020818202550950277,
+      "loss": 3.0064,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020775800431273235,
+      "loss": 3.0718,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002073343021047911,
+      "loss": 2.9981,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020691091934815997,
+      "loss": 3.0412,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020648785650497053,
+      "loss": 2.9645,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020606511403700574,
+      "loss": 2.959,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020564269240569872,
+      "loss": 2.947,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020522059207213228,
+      "loss": 2.9229,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020479881349703882,
+      "loss": 3.0351,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020437735714079897,
+      "loss": 2.9342,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020395622346344212,
+      "loss": 3.0457,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002035354129246454,
+      "loss": 2.9259,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020311492598373306,
+      "loss": 3.0576,
+      "step": 2206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020269476309967628,
+      "loss": 2.9979,
+      "step": 2207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002022749247310926,
+      "loss": 3.0062,
+      "step": 2208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002018554113362449,
+      "loss": 3.1259,
+      "step": 2209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020143622337304184,
+      "loss": 3.0445,
+      "step": 2210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020101736129903664,
+      "loss": 2.9889,
+      "step": 2211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002005988255714269,
+      "loss": 3.0232,
+      "step": 2212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000200180616647054,
+      "loss": 3.0114,
+      "step": 2213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019976273498240233,
+      "loss": 2.9615,
+      "step": 2214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019934518103359943,
+      "loss": 2.9627,
+      "step": 2215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019892795525641504,
+      "loss": 2.9036,
+      "step": 2216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019851105810626075,
+      "loss": 2.9029,
+      "step": 2217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019809449003818896,
+      "loss": 3.0143,
+      "step": 2218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019767825150689378,
+      "loss": 3.0351,
+      "step": 2219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019726234296670852,
+      "loss": 3.0498,
+      "step": 2220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019684676487160725,
+      "loss": 2.9823,
+      "step": 2221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019643151767520284,
+      "loss": 3.0251,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019601660183074716,
+      "loss": 2.9443,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019560201779113056,
+      "loss": 3.0551,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001951877660088806,
+      "loss": 3.0153,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001947738469361628,
+      "loss": 3.0105,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019436026102477927,
+      "loss": 3.1001,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019394700872616856,
+      "loss": 3.0785,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019353409049140515,
+      "loss": 2.9726,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001931215067711984,
+      "loss": 3.0491,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019270925801589308,
+      "loss": 3.0011,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019229734467546812,
+      "loss": 3.0184,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019188576719953633,
+      "loss": 3.0245,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000191474526037344,
+      "loss": 2.9614,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019106362163777047,
+      "loss": 2.9894,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001906530544493269,
+      "loss": 2.9044,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019024282492015698,
+      "loss": 3.0238,
+      "step": 2237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018983293349803565,
+      "loss": 3.0916,
+      "step": 2238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001894233806303689,
+      "loss": 3.0402,
+      "step": 2239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018901416676419275,
+      "loss": 3.0408,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001886052923461737,
+      "loss": 2.998,
+      "step": 2241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018819675782260766,
+      "loss": 3.071,
+      "step": 2242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018778856363941903,
+      "loss": 2.9758,
+      "step": 2243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001873807102421614,
+      "loss": 3.0011,
+      "step": 2244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018697319807601604,
+      "loss": 3.0058,
+      "step": 2245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018656602758579193,
+      "loss": 2.9636,
+      "step": 2246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001861591992159247,
+      "loss": 2.9928,
+      "step": 2247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018575271341047706,
+      "loss": 3.0081,
+      "step": 2248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853465706131374,
+      "loss": 3.0589,
+      "step": 2249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018494077126722004,
+      "loss": 3.006,
+      "step": 2250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018453531581566436,
+      "loss": 3.0028,
+      "step": 2251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018413020470103392,
+      "loss": 2.8148,
+      "step": 2252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018372543836551696,
+      "loss": 2.987,
+      "step": 2253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001833210172509252,
+      "loss": 2.9602,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018291694179869357,
+      "loss": 3.0809,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001825132124498796,
+      "loss": 3.0523,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018210982964516343,
+      "loss": 3.0286,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018170679382484617,
+      "loss": 2.9352,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018130410542885084,
+      "loss": 2.8767,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018090176489672104,
+      "loss": 2.9517,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018049977266762073,
+      "loss": 3.1135,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018009812918033376,
+      "loss": 3.035,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017969683487326278,
+      "loss": 2.9725,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017929589018443015,
+      "loss": 3.1007,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017889529555147583,
+      "loss": 2.94,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017849505141165817,
+      "loss": 3.016,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017809515820185284,
+      "loss": 3.1001,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001776956163585526,
+      "loss": 2.962,
+      "step": 2268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017729642631786615,
+      "loss": 3.0345,
+      "step": 2269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017689758851551873,
+      "loss": 2.9199,
+      "step": 2270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000176499103386851,
+      "loss": 2.9824,
+      "step": 2271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017610097136681852,
+      "loss": 2.9771,
+      "step": 2272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001757031928899917,
+      "loss": 2.9406,
+      "step": 2273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001753057683905545,
+      "loss": 3.0116,
+      "step": 2274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001749086983023051,
+      "loss": 2.9935,
+      "step": 2275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017451198305865452,
+      "loss": 2.9904,
+      "step": 2276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017411562309262656,
+      "loss": 2.8838,
+      "step": 2277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017371961883685721,
+      "loss": 2.999,
+      "step": 2278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017332397072359436,
+      "loss": 3.0218,
+      "step": 2279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017292867918469656,
+      "loss": 2.9838,
+      "step": 2280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017253374465163386,
+      "loss": 2.9777,
+      "step": 2281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017213916755548625,
+      "loss": 3.0549,
+      "step": 2282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001717449483269437,
+      "loss": 2.9166,
+      "step": 2283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017135108739630572,
+      "loss": 2.9988,
+      "step": 2284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017095758519348014,
+      "loss": 2.8792,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001705644421479839,
+      "loss": 2.931,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017017165868894168,
+      "loss": 3.1211,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016977923524508532,
+      "loss": 2.8426,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001693871722447542,
+      "loss": 2.9636,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016899547011589417,
+      "loss": 2.9774,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016860412928605678,
+      "loss": 2.984,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016821315018239962,
+      "loss": 3.0069,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016782253323168546,
+      "loss": 3.0502,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001674322788602815,
+      "loss": 2.9448,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016704238749415957,
+      "loss": 2.8851,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016665285955889465,
+      "loss": 2.9744,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016626369547966563,
+      "loss": 3.0195,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001658748956812539,
+      "loss": 2.9763,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016548646058804346,
+      "loss": 2.9291,
+      "step": 2299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016509839062402005,
+      "loss": 2.9048,
+      "step": 2300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016471068621277103,
+      "loss": 3.0275,
+      "step": 2301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016432334777748443,
+      "loss": 2.9473,
+      "step": 2302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001639363757409491,
+      "loss": 2.997,
+      "step": 2303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016354977052555393,
+      "loss": 3.0604,
+      "step": 2304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001631635325532873,
+      "loss": 2.9697,
+      "step": 2305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016277766224573703,
+      "loss": 2.8906,
+      "step": 2306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016239216002408909,
+      "loss": 2.8221,
+      "step": 2307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000162007026309128,
+      "loss": 2.9882,
+      "step": 2308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016162226152123633,
+      "loss": 2.9281,
+      "step": 2309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001612378660803933,
+      "loss": 2.9962,
+      "step": 2310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016085384040617545,
+      "loss": 3.0537,
+      "step": 2311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016047018491775584,
+      "loss": 2.9698,
+      "step": 2312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001600869000339028,
+      "loss": 2.8588,
+      "step": 2313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015970398617298076,
+      "loss": 3.0183,
+      "step": 2314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000159321443752949,
+      "loss": 2.9484,
+      "step": 2315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015893927319136125,
+      "loss": 2.9958,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015855747490536572,
+      "loss": 3.0281,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015817604931170348,
+      "loss": 2.9203,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001577949968267096,
+      "loss": 2.9671,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001574143178663116,
+      "loss": 3.0136,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015703401284602913,
+      "loss": 3.0214,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015665408218097415,
+      "loss": 2.9904,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015627452628584932,
+      "loss": 2.9521,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015589534557494867,
+      "loss": 2.9794,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015551654046215668,
+      "loss": 2.8941,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015513811136094787,
+      "loss": 3.0037,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015476005868438614,
+      "loss": 3.0091,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001543823828451249,
+      "loss": 2.8751,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001540050842554056,
+      "loss": 3.0481,
+      "step": 2329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015362816332705848,
+      "loss": 3.0061,
+      "step": 2330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015325162047150138,
+      "loss": 2.9696,
+      "step": 2331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001528754560997395,
+      "loss": 3.0944,
+      "step": 2332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001524996706223647,
+      "loss": 2.9096,
+      "step": 2333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001521242644495557,
+      "loss": 2.9247,
+      "step": 2334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015174923799107654,
+      "loss": 2.8893,
+      "step": 2335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015137459165627737,
+      "loss": 2.9509,
+      "step": 2336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015100032585409334,
+      "loss": 3.0873,
+      "step": 2337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000150626440993044,
+      "loss": 3.0549,
+      "step": 2338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015025293748123354,
+      "loss": 2.9201,
+      "step": 2339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014987981572634913,
+      "loss": 3.0575,
+      "step": 2340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014950707613566201,
+      "loss": 3.0316,
+      "step": 2341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014913471911602583,
+      "loss": 3.098,
+      "step": 2342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014876274507387693,
+      "loss": 3.0198,
+      "step": 2343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014839115441523355,
+      "loss": 2.9397,
+      "step": 2344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001480199475456952,
+      "loss": 2.9235,
+      "step": 2345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014764912487044273,
+      "loss": 2.8298,
+      "step": 2346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014727868679423762,
+      "loss": 2.9837,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014690863372142154,
+      "loss": 3.0424,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014653896605591583,
+      "loss": 2.9656,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001461696842012214,
+      "loss": 3.0639,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014580078856041756,
+      "loss": 2.9402,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001454322795361625,
+      "loss": 2.8472,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001450641575306922,
+      "loss": 3.0676,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014469642294582047,
+      "loss": 2.9727,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014432907618293768,
+      "loss": 2.9911,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014396211764301144,
+      "loss": 2.9573,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014359554772658552,
+      "loss": 3.0336,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014322936683377902,
+      "loss": 3.0624,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014286357536428697,
+      "loss": 3.0345,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014249817371737918,
+      "loss": 2.8598,
+      "step": 2360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001421331622919,
+      "loss": 2.9227,
+      "step": 2361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014176854148626743,
+      "loss": 3.074,
+      "step": 2362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014140431169847368,
+      "loss": 2.9493,
+      "step": 2363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014104047332608378,
+      "loss": 2.9295,
+      "step": 2364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001406770267662358,
+      "loss": 2.9723,
+      "step": 2365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014031397241564,
+      "loss": 3.0595,
+      "step": 2366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013995131067057821,
+      "loss": 3.1249,
+      "step": 2367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013958904192690424,
+      "loss": 2.8877,
+      "step": 2368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001392271665800427,
+      "loss": 2.9537,
+      "step": 2369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013886568502498886,
+      "loss": 3.1024,
+      "step": 2370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013850459765630803,
+      "loss": 2.9398,
+      "step": 2371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013814390486813555,
+      "loss": 3.1041,
+      "step": 2372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013778360705417553,
+      "loss": 3.007,
+      "step": 2373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013742370460770144,
+      "loss": 3.03,
+      "step": 2374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013706419792155507,
+      "loss": 3.0065,
+      "step": 2375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013670508738814613,
+      "loss": 2.9617,
+      "step": 2376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013634637339945217,
+      "loss": 2.9803,
+      "step": 2377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013598805634701744,
+      "loss": 2.9807,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013563013662195355,
+      "loss": 3.1233,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013527261461493767,
+      "loss": 2.986,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013491549071621345,
+      "loss": 3.0072,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001345587653155898,
+      "loss": 2.982,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013420243880244088,
+      "loss": 2.9045,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013384651156570483,
+      "loss": 2.8894,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001334909839938846,
+      "loss": 3.0116,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013313585647504666,
+      "loss": 2.9891,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001327811293968208,
+      "loss": 2.9274,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013242680314639994,
+      "loss": 3.0107,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013207287811053893,
+      "loss": 2.9991,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013171935467555523,
+      "loss": 2.9722,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013136623322732783,
+      "loss": 2.9219,
+      "step": 2391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013101351415129682,
+      "loss": 2.972,
+      "step": 2392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001306611978324632,
+      "loss": 3.0182,
+      "step": 2393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001303092846553882,
+      "loss": 2.9949,
+      "step": 2394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012995777500419313,
+      "loss": 3.0154,
+      "step": 2395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012960666926255883,
+      "loss": 3.066,
+      "step": 2396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001292559678137253,
+      "loss": 3.0283,
+      "step": 2397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012890567104049112,
+      "loss": 3.0363,
+      "step": 2398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001285557793252135,
+      "loss": 3.0181,
+      "step": 2399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012820629304980685,
+      "loss": 2.9237,
+      "step": 2400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012785721259574368,
+      "loss": 3.0805,
+      "step": 2401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012750853834405342,
+      "loss": 3.0003,
+      "step": 2402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012716027067532164,
+      "loss": 2.9884,
+      "step": 2403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012681240996969084,
+      "loss": 2.8644,
+      "step": 2404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012646495660685863,
+      "loss": 2.9898,
+      "step": 2405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012611791096607845,
+      "loss": 3.0008,
+      "step": 2406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012577127342615859,
+      "loss": 3.0083,
+      "step": 2407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012542504436546182,
+      "loss": 2.902,
+      "step": 2408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001250792241619051,
+      "loss": 3.0661,
+      "step": 2409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012473381319295929,
+      "loss": 2.9876,
+      "step": 2410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012438881183564799,
+      "loss": 2.9041,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012404422046654833,
+      "loss": 3.0132,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001237000394617896,
+      "loss": 3.0102,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001233562691970533,
+      "loss": 2.9431,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012301291004757265,
+      "loss": 3.0831,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012266996238813173,
+      "loss": 3.1117,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001223274265930659,
+      "loss": 2.9782,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012198530303626087,
+      "loss": 3.0267,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012164359209115234,
+      "loss": 2.937,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012130229413072558,
+      "loss": 3.0182,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012096140952751528,
+      "loss": 2.9749,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012062093865360457,
+      "loss": 3.1412,
+      "step": 2422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012028088188062525,
+      "loss": 3.0087,
+      "step": 2423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011994123957975722,
+      "loss": 3.0568,
+      "step": 2424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011960201212172766,
+      "loss": 3.0225,
+      "step": 2425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001192631998768111,
+      "loss": 2.9839,
+      "step": 2426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011892480321482912,
+      "loss": 2.9538,
+      "step": 2427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011858682250514913,
+      "loss": 2.843,
+      "step": 2428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011824925811668486,
+      "loss": 2.8762,
+      "step": 2429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011791211041789563,
+      "loss": 3.0867,
+      "step": 2430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011757537977678589,
+      "loss": 3.0366,
+      "step": 2431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011723906656090499,
+      "loss": 2.8965,
+      "step": 2432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011690317113734616,
+      "loss": 2.9752,
+      "step": 2433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011656769387274713,
+      "loss": 2.9655,
+      "step": 2434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011623263513328903,
+      "loss": 3.0377,
+      "step": 2435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011589799528469614,
+      "loss": 2.8227,
+      "step": 2436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011556377469223562,
+      "loss": 3.0122,
+      "step": 2437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011522997372071658,
+      "loss": 3.031,
+      "step": 2438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011489659273449071,
+      "loss": 3.0258,
+      "step": 2439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011456363209745091,
+      "loss": 2.9032,
+      "step": 2440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011423109217303129,
+      "loss": 2.9713,
+      "step": 2441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011389897332420685,
+      "loss": 3.082,
+      "step": 2442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011356727591349303,
+      "loss": 3.1176,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001132360003029449,
+      "loss": 2.9265,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011290514685415749,
+      "loss": 3.0288,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011257471592826502,
+      "loss": 2.9406,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001122447078859402,
+      "loss": 2.892,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011191512308739443,
+      "loss": 3.0309,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001115859618923773,
+      "loss": 2.8802,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011125722466017545,
+      "loss": 2.9471,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011092891174961323,
+      "loss": 2.9395,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011060102351905172,
+      "loss": 2.9258,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011027356032638836,
+      "loss": 3.0303,
+      "step": 2453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010994652252905696,
+      "loss": 2.9573,
+      "step": 2454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010961991048402648,
+      "loss": 2.9424,
+      "step": 2455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010929372454780152,
+      "loss": 2.9713,
+      "step": 2456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010896796507642154,
+      "loss": 3.0484,
+      "step": 2457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010864263242546046,
+      "loss": 2.9851,
+      "step": 2458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010831772695002639,
+      "loss": 2.9491,
+      "step": 2459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010799324900476092,
+      "loss": 2.913,
+      "step": 2460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010766919894383926,
+      "loss": 2.9232,
+      "step": 2461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010734557712096943,
+      "loss": 3.0172,
+      "step": 2462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010702238388939211,
+      "loss": 3.0077,
+      "step": 2463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010669961960188007,
+      "loss": 2.8919,
+      "step": 2464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001063772846107381,
+      "loss": 2.996,
+      "step": 2465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001060553792678019,
+      "loss": 3.036,
+      "step": 2466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010573390392443872,
+      "loss": 2.9301,
+      "step": 2467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010541285893154617,
+      "loss": 3.0561,
+      "step": 2468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010509224463955247,
+      "loss": 3.0746,
+      "step": 2469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010477206139841506,
+      "loss": 2.9147,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010445230955762158,
+      "loss": 2.9636,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010413298946618832,
+      "loss": 2.9895,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010381410147266047,
+      "loss": 2.9386,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010349564592511163,
+      "loss": 2.9239,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010317762317114349,
+      "loss": 2.8971,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010286003355788487,
+      "loss": 2.9796,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001025428774319922,
+      "loss": 2.9482,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010222615513964872,
+      "loss": 2.9667,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010190986702656402,
+      "loss": 2.9424,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010159401343797381,
+      "loss": 2.9607,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001012785947186397,
+      "loss": 2.9733,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010096361121284808,
+      "loss": 3.0743,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010064906326441093,
+      "loss": 2.933,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010033495121666441,
+      "loss": 2.9653,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010002127541246909,
+      "loss": 2.93,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.970803619420937e-05,
+      "loss": 2.9404,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.939523390379285e-05,
+      "loss": 3.0332,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.908286888265038e-05,
+      "loss": 3.0192,
+      "step": 2488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.877094147173566e-05,
+      "loss": 2.9352,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.845945201152456e-05,
+      "loss": 3.0256,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.81484008420151e-05,
+      "loss": 3.0573,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.783778830272643e-05,
+      "loss": 2.8965,
+      "step": 2492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.75276147326995e-05,
+      "loss": 2.87,
+      "step": 2493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.721788047049584e-05,
+      "loss": 3.1011,
+      "step": 2494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.690858585419738e-05,
+      "loss": 2.8365,
+      "step": 2495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.659973122140625e-05,
+      "loss": 3.0665,
+      "step": 2496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.629131690924465e-05,
+      "loss": 3.0355,
+      "step": 2497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.598334325435348e-05,
+      "loss": 2.9965,
+      "step": 2498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.56758105928932e-05,
+      "loss": 2.7877,
+      "step": 2499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.536871926054275e-05,
+      "loss": 2.9378,
+      "step": 2500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.506206959249941e-05,
+      "loss": 2.9917,
+      "step": 2501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.475586192347818e-05,
+      "loss": 3.0148,
+      "step": 2502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.445009658771204e-05,
+      "loss": 2.9331,
+      "step": 2503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.414477391895044e-05,
+      "loss": 2.9547,
+      "step": 2504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.383989425046024e-05,
+      "loss": 2.8813,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.353545791502466e-05,
+      "loss": 2.9638,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.32314652449428e-05,
+      "loss": 3.0125,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.292791657202986e-05,
+      "loss": 2.9995,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.262481222761587e-05,
+      "loss": 2.9715,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.232215254254628e-05,
+      "loss": 2.9096,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.201993784718105e-05,
+      "loss": 3.0248,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.171816847139447e-05,
+      "loss": 2.9797,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.141684474457474e-05,
+      "loss": 2.9278,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.11159669956237e-05,
+      "loss": 2.9463,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.081553555295602e-05,
+      "loss": 2.8546,
+      "step": 2515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.051555074449979e-05,
+      "loss": 2.9304,
+      "step": 2516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.021601289769504e-05,
+      "loss": 2.9021,
+      "step": 2517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.991692233949433e-05,
+      "loss": 2.879,
+      "step": 2518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.961827939636197e-05,
+      "loss": 3.0957,
+      "step": 2519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.93200843942733e-05,
+      "loss": 3.037,
+      "step": 2520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.902233765871509e-05,
+      "loss": 3.04,
+      "step": 2521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.87250395146848e-05,
+      "loss": 2.8949,
+      "step": 2522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.842819028669013e-05,
+      "loss": 3.0356,
+      "step": 2523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.813179029874874e-05,
+      "loss": 2.9268,
+      "step": 2524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.783583987438826e-05,
+      "loss": 3.046,
+      "step": 2525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.754033933664496e-05,
+      "loss": 2.8973,
+      "step": 2526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.724528900806467e-05,
+      "loss": 2.951,
+      "step": 2527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.695068921070154e-05,
+      "loss": 3.0853,
+      "step": 2528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.665654026611797e-05,
+      "loss": 2.8596,
+      "step": 2529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.636284249538445e-05,
+      "loss": 3.0803,
+      "step": 2530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.606959621907851e-05,
+      "loss": 2.9141,
+      "step": 2531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.577680175728536e-05,
+      "loss": 3.0723,
+      "step": 2532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.548445942959682e-05,
+      "loss": 2.8846,
+      "step": 2533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.519256955511129e-05,
+      "loss": 2.9113,
+      "step": 2534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.490113245243314e-05,
+      "loss": 2.9769,
+      "step": 2535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.461014843967286e-05,
+      "loss": 2.9244,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.431961783444587e-05,
+      "loss": 3.0576,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.40295409538731e-05,
+      "loss": 3.0594,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.373991811458026e-05,
+      "loss": 3.1361,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.345074963269706e-05,
+      "loss": 2.894,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.316203582385778e-05,
+      "loss": 2.9287,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.287377700319981e-05,
+      "loss": 3.0721,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.258597348536451e-05,
+      "loss": 3.0984,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.229862558449591e-05,
+      "loss": 2.9446,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.201173361424086e-05,
+      "loss": 2.8784,
+      "step": 2545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.172529788774847e-05,
+      "loss": 3.0218,
+      "step": 2546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.143931871767013e-05,
+      "loss": 3.0434,
+      "step": 2547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.115379641615828e-05,
+      "loss": 2.9223,
+      "step": 2548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.086873129486722e-05,
+      "loss": 3.0921,
+      "step": 2549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.05841236649521e-05,
+      "loss": 2.8617,
+      "step": 2550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.02999738370686e-05,
+      "loss": 2.9834,
+      "step": 2551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.001628212137291e-05,
+      "loss": 2.9699,
+      "step": 2552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.97330488275208e-05,
+      "loss": 2.8652,
+      "step": 2553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.9450274264668e-05,
+      "loss": 3.0043,
+      "step": 2554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.916795874146943e-05,
+      "loss": 2.9487,
+      "step": 2555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.888610256607892e-05,
+      "loss": 2.9269,
+      "step": 2556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.8604706046149e-05,
+      "loss": 2.956,
+      "step": 2557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.832376948883047e-05,
+      "loss": 3.0668,
+      "step": 2558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.80432932007718e-05,
+      "loss": 3.0969,
+      "step": 2559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.776327748811946e-05,
+      "loss": 2.983,
+      "step": 2560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.748372265651699e-05,
+      "loss": 2.9572,
+      "step": 2561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.720462901110476e-05,
+      "loss": 2.9519,
+      "step": 2562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.692599685651996e-05,
+      "loss": 3.0298,
+      "step": 2563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.664782649689611e-05,
+      "loss": 2.8902,
+      "step": 2564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.637011823586221e-05,
+      "loss": 2.9122,
+      "step": 2565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.609287237654344e-05,
+      "loss": 2.9359,
+      "step": 2566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.581608922155986e-05,
+      "loss": 3.0138,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.553976907302684e-05,
+      "loss": 3.009,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.526391223255386e-05,
+      "loss": 3.0433,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.498851900124509e-05,
+      "loss": 2.9578,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.47135896796986e-05,
+      "loss": 2.8991,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.443912456800605e-05,
+      "loss": 2.9429,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.416512396575226e-05,
+      "loss": 2.8824,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.38915881720154e-05,
+      "loss": 2.9238,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.361851748536585e-05,
+      "loss": 2.923,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.334591220386649e-05,
+      "loss": 2.8918,
+      "step": 2576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.307377262507237e-05,
+      "loss": 2.9484,
+      "step": 2577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.280209904602996e-05,
+      "loss": 3.0292,
+      "step": 2578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.253089176327737e-05,
+      "loss": 3.0476,
+      "step": 2579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.226015107284328e-05,
+      "loss": 2.9272,
+      "step": 2580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.198987727024753e-05,
+      "loss": 2.9541,
+      "step": 2581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.172007065050012e-05,
+      "loss": 3.0449,
+      "step": 2582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.145073150810117e-05,
+      "loss": 2.9918,
+      "step": 2583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.118186013704065e-05,
+      "loss": 2.9181,
+      "step": 2584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.091345683079753e-05,
+      "loss": 3.008,
+      "step": 2585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.064552188234047e-05,
+      "loss": 2.8182,
+      "step": 2586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.03780555841263e-05,
+      "loss": 3.0356,
+      "step": 2587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.011105822810076e-05,
+      "loss": 2.9909,
+      "step": 2588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984453010569758e-05,
+      "loss": 3.0316,
+      "step": 2589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.957847150783847e-05,
+      "loss": 2.9661,
+      "step": 2590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.931288272493213e-05,
+      "loss": 3.0554,
+      "step": 2591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.904776404687497e-05,
+      "loss": 2.9906,
+      "step": 2592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.878311576305013e-05,
+      "loss": 3.0954,
+      "step": 2593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.85189381623273e-05,
+      "loss": 3.0395,
+      "step": 2594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.825523153306223e-05,
+      "loss": 2.8909,
+      "step": 2595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.799199616309703e-05,
+      "loss": 3.0653,
+      "step": 2596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.77292323397588e-05,
+      "loss": 2.8906,
+      "step": 2597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.746694034986039e-05,
+      "loss": 3.0369,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.720512047969957e-05,
+      "loss": 2.9826,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.694377301505866e-05,
+      "loss": 2.964,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.668289824120454e-05,
+      "loss": 2.9066,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642249644288778e-05,
+      "loss": 3.0262,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.616256790434305e-05,
+      "loss": 3.0305,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.590311290928824e-05,
+      "loss": 2.8784,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.564413174092443e-05,
+      "loss": 2.9627,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.53856246819356e-05,
+      "loss": 3.0501,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.512759201448798e-05,
+      "loss": 3.0993,
+      "step": 2607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.487003402023012e-05,
+      "loss": 2.9568,
+      "step": 2608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.461295098029268e-05,
+      "loss": 3.0823,
+      "step": 2609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.435634317528737e-05,
+      "loss": 2.9869,
+      "step": 2610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.410021088530771e-05,
+      "loss": 2.9163,
+      "step": 2611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.384455438992797e-05,
+      "loss": 3.0476,
+      "step": 2612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.358937396820291e-05,
+      "loss": 2.9848,
+      "step": 2613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.333466989866787e-05,
+      "loss": 2.9712,
+      "step": 2614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.30804424593382e-05,
+      "loss": 3.0385,
+      "step": 2615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.282669192770896e-05,
+      "loss": 3.0005,
+      "step": 2616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257341858075466e-05,
+      "loss": 2.9973,
+      "step": 2617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.232062269492905e-05,
+      "loss": 3.0865,
+      "step": 2618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.206830454616447e-05,
+      "loss": 2.9491,
+      "step": 2619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.181646440987204e-05,
+      "loss": 3.1126,
+      "step": 2620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1565102560941e-05,
+      "loss": 2.9415,
+      "step": 2621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.131421927373859e-05,
+      "loss": 3.0046,
+      "step": 2622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.106381482210976e-05,
+      "loss": 3.0096,
+      "step": 2623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0813889479376495e-05,
+      "loss": 2.9674,
+      "step": 2624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.056444351833812e-05,
+      "loss": 2.8852,
+      "step": 2625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.031547721127062e-05,
+      "loss": 2.8914,
+      "step": 2626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.006699082992645e-05,
+      "loss": 2.9373,
+      "step": 2627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9818984645534144e-05,
+      "loss": 2.9106,
+      "step": 2628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.957145892879828e-05,
+      "loss": 2.9669,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.932441394989857e-05,
+      "loss": 2.9628,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.907784997849053e-05,
+      "loss": 2.9238,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.883176728370415e-05,
+      "loss": 3.0738,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.858616613414442e-05,
+      "loss": 2.9423,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.834104679789076e-05,
+      "loss": 2.9514,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8096409542496284e-05,
+      "loss": 2.9262,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.785225463498828e-05,
+      "loss": 3.1032,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.760858234186739e-05,
+      "loss": 3.0563,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.736539292910758e-05,
+      "loss": 2.9757,
+      "step": 2638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.712268666215559e-05,
+      "loss": 2.9081,
+      "step": 2639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.688046380593098e-05,
+      "loss": 2.9893,
+      "step": 2640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6638724624825335e-05,
+      "loss": 2.9282,
+      "step": 2641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.639746938270262e-05,
+      "loss": 2.9993,
+      "step": 2642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.615669834289849e-05,
+      "loss": 2.9588,
+      "step": 2643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5916411768220044e-05,
+      "loss": 3.022,
+      "step": 2644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5676609920945584e-05,
+      "loss": 2.9411,
+      "step": 2645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5437293062824243e-05,
+      "loss": 2.9284,
+      "step": 2646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.51984614550759e-05,
+      "loss": 2.8598,
+      "step": 2647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.496011535839063e-05,
+      "loss": 2.9484,
+      "step": 2648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4722255032928826e-05,
+      "loss": 3.0204,
+      "step": 2649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.448488073832047e-05,
+      "loss": 2.9856,
+      "step": 2650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.424799273366482e-05,
+      "loss": 2.9646,
+      "step": 2651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.401159127753075e-05,
+      "loss": 2.9785,
+      "step": 2652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.377567662795579e-05,
+      "loss": 2.8884,
+      "step": 2653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.354024904244631e-05,
+      "loss": 3.0357,
+      "step": 2654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3305308777976677e-05,
+      "loss": 2.9599,
+      "step": 2655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3070856090989824e-05,
+      "loss": 2.8807,
+      "step": 2656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.283689123739599e-05,
+      "loss": 3.0019,
+      "step": 2657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.260341447257327e-05,
+      "loss": 2.8991,
+      "step": 2658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2370426051366884e-05,
+      "loss": 3.0585,
+      "step": 2659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.213792622808905e-05,
+      "loss": 3.0097,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.190591525651878e-05,
+      "loss": 3.0201,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.16743933899011e-05,
+      "loss": 2.9984,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1443360880947584e-05,
+      "loss": 2.9434,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121281798183547e-05,
+      "loss": 2.9726,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.098276494420756e-05,
+      "loss": 2.8647,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.075320201917211e-05,
+      "loss": 2.8781,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0524129457302394e-05,
+      "loss": 2.9007,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.029554750863613e-05,
+      "loss": 2.9399,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.00674564226758e-05,
+      "loss": 2.9227,
+      "step": 2669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.983985644838812e-05,
+      "loss": 2.9974,
+      "step": 2670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9612747834203666e-05,
+      "loss": 3.0109,
+      "step": 2671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938613082801674e-05,
+      "loss": 3.0019,
+      "step": 2672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9160005677184875e-05,
+      "loss": 3.0058,
+      "step": 2673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8934372628528846e-05,
+      "loss": 2.9707,
+      "step": 2674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.870923192833238e-05,
+      "loss": 3.008,
+      "step": 2675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8484583822341774e-05,
+      "loss": 3.0139,
+      "step": 2676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.82604285557654e-05,
+      "loss": 3.0553,
+      "step": 2677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.803676637327392e-05,
+      "loss": 2.896,
+      "step": 2678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.781359751899983e-05,
+      "loss": 3.0753,
+      "step": 2679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7590922236536883e-05,
+      "loss": 3.0115,
+      "step": 2680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.736874076894032e-05,
+      "loss": 3.0409,
+      "step": 2681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7147053358726266e-05,
+      "loss": 2.9013,
+      "step": 2682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.692586024787182e-05,
+      "loss": 2.9493,
+      "step": 2683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.670516167781402e-05,
+      "loss": 2.8489,
+      "step": 2684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.648495788945051e-05,
+      "loss": 3.0374,
+      "step": 2685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6265249123138844e-05,
+      "loss": 3.0525,
+      "step": 2686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6046035618696136e-05,
+      "loss": 3.1137,
+      "step": 2687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.582731761539888e-05,
+      "loss": 2.992,
+      "step": 2688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5609095351982986e-05,
+      "loss": 2.9949,
+      "step": 2689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5391369066642766e-05,
+      "loss": 3.0103,
+      "step": 2690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.51741389970316e-05,
+      "loss": 3.0365,
+      "step": 2691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4957405380261054e-05,
+      "loss": 2.9117,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4741168452900884e-05,
+      "loss": 2.933,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.452542845097862e-05,
+      "loss": 2.8909,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.43101856099794e-05,
+      "loss": 2.9785,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.409544016484562e-05,
+      "loss": 3.0106,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3881192349977006e-05,
+      "loss": 2.9871,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.366744239922998e-05,
+      "loss": 2.982,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3454190545917316e-05,
+      "loss": 2.9328,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.324143702280836e-05,
+      "loss": 2.9711,
+      "step": 2700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3029182062128635e-05,
+      "loss": 2.922,
+      "step": 2701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2817425895559034e-05,
+      "loss": 3.037,
+      "step": 2702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.260616875423634e-05,
+      "loss": 2.9058,
+      "step": 2703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2395410868752645e-05,
+      "loss": 3.0743,
+      "step": 2704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2185152469154994e-05,
+      "loss": 2.9341,
+      "step": 2705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.197539378494514e-05,
+      "loss": 3.0223,
+      "step": 2706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1766135045079536e-05,
+      "loss": 2.9417,
+      "step": 2707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.155737647796892e-05,
+      "loss": 2.9853,
+      "step": 2708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134911831147797e-05,
+      "loss": 2.8863,
+      "step": 2709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.114136077292541e-05,
+      "loss": 3.0264,
+      "step": 2710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.093410408908327e-05,
+      "loss": 2.9062,
+      "step": 2711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.072734848617682e-05,
+      "loss": 2.8829,
+      "step": 2712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.05210941898847e-05,
+      "loss": 2.8905,
+      "step": 2713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.031534142533816e-05,
+      "loss": 2.9733,
+      "step": 2714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0110090417120995e-05,
+      "loss": 3.0756,
+      "step": 2715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.990534138926955e-05,
+      "loss": 2.9858,
+      "step": 2716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9701094565271824e-05,
+      "loss": 3.0573,
+      "step": 2717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9497350168068103e-05,
+      "loss": 3.0524,
+      "step": 2718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9294108420049936e-05,
+      "loss": 3.0259,
+      "step": 2719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.909136954306042e-05,
+      "loss": 2.9635,
+      "step": 2720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.888913375839376e-05,
+      "loss": 2.9811,
+      "step": 2721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.868740128679471e-05,
+      "loss": 3.025,
+      "step": 2722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.848617234845914e-05,
+      "loss": 2.961,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8285447163032836e-05,
+      "loss": 3.0398,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8085225949612e-05,
+      "loss": 3.089,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7885508926742695e-05,
+      "loss": 2.9478,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.768629631242071e-05,
+      "loss": 2.9308,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.748758832409099e-05,
+      "loss": 2.9544,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.728938517864794e-05,
+      "loss": 2.9837,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7091687092434876e-05,
+      "loss": 2.843,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.68944942812437e-05,
+      "loss": 2.9241,
+      "step": 2731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6697806960314994e-05,
+      "loss": 2.8604,
+      "step": 2732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.650162534433738e-05,
+      "loss": 2.9783,
+      "step": 2733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6305949647447544e-05,
+      "loss": 2.9882,
+      "step": 2734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.61107800832301e-05,
+      "loss": 2.963,
+      "step": 2735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.591611686471702e-05,
+      "loss": 3.0134,
+      "step": 2736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.572196020438773e-05,
+      "loss": 2.9752,
+      "step": 2737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55283103141687e-05,
+      "loss": 3.0407,
+      "step": 2738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533516740543302e-05,
+      "loss": 2.9967,
+      "step": 2739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.514253168900078e-05,
+      "loss": 3.0493,
+      "step": 2740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.495040337513822e-05,
+      "loss": 3.0282,
+      "step": 2741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4758782673557754e-05,
+      "loss": 2.9826,
+      "step": 2742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4567669793417964e-05,
+      "loss": 2.9434,
+      "step": 2743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.437706494332266e-05,
+      "loss": 2.9449,
+      "step": 2744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.418696833132145e-05,
+      "loss": 3.0032,
+      "step": 2745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3997380164909386e-05,
+      "loss": 2.936,
+      "step": 2746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.380830065102597e-05,
+      "loss": 2.9835,
+      "step": 2747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3619729996056006e-05,
+      "loss": 3.0381,
+      "step": 2748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.343166840582868e-05,
+      "loss": 2.9955,
+      "step": 2749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3244116085617446e-05,
+      "loss": 2.9462,
+      "step": 2750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.305707324013996e-05,
+      "loss": 2.96,
+      "step": 2751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.287054007355778e-05,
+      "loss": 2.9574,
+      "step": 2752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2684516789476204e-05,
+      "loss": 2.9331,
+      "step": 2753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2499003590943876e-05,
+      "loss": 2.9841,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.231400068045259e-05,
+      "loss": 2.9343,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2129508259937376e-05,
+      "loss": 2.9182,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.194552653077592e-05,
+      "loss": 2.9063,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1762055693788426e-05,
+      "loss": 2.9915,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.157909594923758e-05,
+      "loss": 2.9587,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1396647496828245e-05,
+      "loss": 3.0021,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.121471053570679e-05,
+      "loss": 2.8276,
+      "step": 2761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.103328526446175e-05,
+      "loss": 2.9389,
+      "step": 2762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0852371881123e-05,
+      "loss": 2.982,
+      "step": 2763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.067197058316157e-05,
+      "loss": 3.008,
+      "step": 2764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0492081567489673e-05,
+      "loss": 3.0634,
+      "step": 2765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0312705030460185e-05,
+      "loss": 2.8997,
+      "step": 2766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0133841167866716e-05,
+      "loss": 2.8482,
+      "step": 2767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9955490174943424e-05,
+      "loss": 2.9559,
+      "step": 2768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9777652246364306e-05,
+      "loss": 3.0353,
+      "step": 2769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960032757624359e-05,
+      "loss": 2.9414,
+      "step": 2770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.942351635813534e-05,
+      "loss": 2.9391,
+      "step": 2771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9247218785032893e-05,
+      "loss": 2.924,
+      "step": 2772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.90714350493691e-05,
+      "loss": 2.9871,
+      "step": 2773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8896165343015977e-05,
+      "loss": 3.0478,
+      "step": 2774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8721409857284442e-05,
+      "loss": 3.0229,
+      "step": 2775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8547168782924126e-05,
+      "loss": 2.8535,
+      "step": 2776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.837344231012301e-05,
+      "loss": 2.953,
+      "step": 2777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.820023062850763e-05,
+      "loss": 2.8037,
+      "step": 2778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8027533927142524e-05,
+      "loss": 2.9071,
+      "step": 2779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.785535239453002e-05,
+      "loss": 2.9785,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.768368621861028e-05,
+      "loss": 2.98,
+      "step": 2781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7512535586760868e-05,
+      "loss": 2.9828,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7341900685796505e-05,
+      "loss": 2.8558,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7171781701969157e-05,
+      "loss": 2.9931,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7002178820967616e-05,
+      "loss": 3.0215,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6833092227917244e-05,
+      "loss": 3.0068,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6664522107380073e-05,
+      "loss": 2.7821,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.649646864335409e-05,
+      "loss": 2.8857,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6328932019273556e-05,
+      "loss": 2.8896,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6161912418008527e-05,
+      "loss": 2.9762,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5995410021864786e-05,
+      "loss": 2.9789,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5829425012583395e-05,
+      "loss": 2.9157,
+      "step": 2792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5663957571340934e-05,
+      "loss": 3.0848,
+      "step": 2793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5499007878748758e-05,
+      "loss": 2.9992,
+      "step": 2794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5334576114853237e-05,
+      "loss": 3.0426,
+      "step": 2795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.517066245913552e-05,
+      "loss": 2.9613,
+      "step": 2796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5007267090511044e-05,
+      "loss": 3.0344,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4844390187329645e-05,
+      "loss": 3.0359,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4682031927375113e-05,
+      "loss": 2.9144,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.45201924878653e-05,
+      "loss": 2.9882,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4358872045451563e-05,
+      "loss": 2.9407,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4198070776218885e-05,
+      "loss": 2.8941,
+      "step": 2802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.403778885568558e-05,
+      "loss": 2.8768,
+      "step": 2803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3878026458803048e-05,
+      "loss": 2.9341,
+      "step": 2804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3718783759955508e-05,
+      "loss": 2.933,
+      "step": 2805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.356006093296009e-05,
+      "loss": 3.0058,
+      "step": 2806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3401858151066323e-05,
+      "loss": 3.0353,
+      "step": 2807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.324417558695624e-05,
+      "loss": 3.0358,
+      "step": 2808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3087013412743997e-05,
+      "loss": 2.9864,
+      "step": 2809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2930371799975592e-05,
+      "loss": 2.9777,
+      "step": 2810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.277425091962898e-05,
+      "loss": 3.0216,
+      "step": 2811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2618650942113684e-05,
+      "loss": 3.1545,
+      "step": 2812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.246357203727073e-05,
+      "loss": 2.9915,
+      "step": 2813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2309014374372104e-05,
+      "loss": 2.9338,
+      "step": 2814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2154978122121196e-05,
+      "loss": 3.0041,
+      "step": 2815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.200146344865206e-05,
+      "loss": 3.0027,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1848470521529384e-05,
+      "loss": 2.9342,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.169599950774853e-05,
+      "loss": 2.9876,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1544050573735154e-05,
+      "loss": 3.0206,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1392623885344964e-05,
+      "loss": 2.9031,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.124171960786353e-05,
+      "loss": 3.0181,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.109133790600648e-05,
+      "loss": 2.9515,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.094147894391879e-05,
+      "loss": 2.9555,
+      "step": 2823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.079214288517506e-05,
+      "loss": 2.9592,
+      "step": 2824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.064332989277895e-05,
+      "loss": 3.0156,
+      "step": 2825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.04950401291632e-05,
+      "loss": 3.085,
+      "step": 2826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0347273756189555e-05,
+      "loss": 3.0343,
+      "step": 2827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.020003093514833e-05,
+      "loss": 3.0048,
+      "step": 2828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0053311826758458e-05,
+      "loss": 2.8983,
+      "step": 2829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9907116591167284e-05,
+      "loss": 2.8957,
+      "step": 2830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9761445387950217e-05,
+      "loss": 2.993,
+      "step": 2831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.961629837611062e-05,
+      "loss": 2.8963,
+      "step": 2832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9471675714079817e-05,
+      "loss": 2.844,
+      "step": 2833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9327577559716815e-05,
+      "loss": 2.9456,
+      "step": 2834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9184004070308013e-05,
+      "loss": 2.9549,
+      "step": 2835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9040955402567228e-05,
+      "loss": 2.917,
+      "step": 2836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8898431712635222e-05,
+      "loss": 3.0309,
+      "step": 2837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.875643315608e-05,
+      "loss": 3.05,
+      "step": 2838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8614959887896078e-05,
+      "loss": 3.0395,
+      "step": 2839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.847401206250493e-05,
+      "loss": 2.915,
+      "step": 2840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8333589833754215e-05,
+      "loss": 2.9758,
+      "step": 2841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.819369335491816e-05,
+      "loss": 3.0009,
+      "step": 2842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8054322778696787e-05,
+      "loss": 3.0398,
+      "step": 2843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.79154782572164e-05,
+      "loss": 3.043,
+      "step": 2844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7777159942028886e-05,
+      "loss": 3.0001,
+      "step": 2845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.763936798411192e-05,
+      "loss": 3.0345,
+      "step": 2846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7502102533868636e-05,
+      "loss": 2.9854,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.736536374112724e-05,
+      "loss": 2.9798,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7229151755141392e-05,
+      "loss": 3.0389,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7093466724589556e-05,
+      "loss": 3.0084,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.695830879757504e-05,
+      "loss": 2.8879,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.682367812162583e-05,
+      "loss": 2.95,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6689574843694434e-05,
+      "loss": 3.011,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.655599911015754e-05,
+      "loss": 3.0414,
+      "step": 2854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6422951066816183e-05,
+      "loss": 3.0164,
+      "step": 2855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6290430858895367e-05,
+      "loss": 2.9347,
+      "step": 2856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.615843863104388e-05,
+      "loss": 2.9208,
+      "step": 2857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.602697452733437e-05,
+      "loss": 2.9071,
+      "step": 2858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.589603869126277e-05,
+      "loss": 2.9435,
+      "step": 2859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5765631265748647e-05,
+      "loss": 2.9987,
+      "step": 2860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5635752393134695e-05,
+      "loss": 2.9821,
+      "step": 2861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5506402215186565e-05,
+      "loss": 2.933,
+      "step": 2862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5377580873093046e-05,
+      "loss": 3.0069,
+      "step": 2863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5249288507465598e-05,
+      "loss": 3.0835,
+      "step": 2864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5121525258338153e-05,
+      "loss": 2.8692,
+      "step": 2865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4994291265167325e-05,
+      "loss": 2.9584,
+      "step": 2866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4867586666831856e-05,
+      "loss": 3.0219,
+      "step": 2867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4741411601632782e-05,
+      "loss": 2.9269,
+      "step": 2868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4615766207293157e-05,
+      "loss": 2.927,
+      "step": 2869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4490650620957558e-05,
+      "loss": 2.9353,
+      "step": 2870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.436606497919274e-05,
+      "loss": 3.0152,
+      "step": 2871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4242009417986646e-05,
+      "loss": 2.9636,
+      "step": 2872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.411848407274885e-05,
+      "loss": 3.0096,
+      "step": 2873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3995489078310053e-05,
+      "loss": 2.9636,
+      "step": 2874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3873024568922144e-05,
+      "loss": 3.008,
+      "step": 2875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.375109067825786e-05,
+      "loss": 2.9365,
+      "step": 2876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3629687539410907e-05,
+      "loss": 3.0268,
+      "step": 2877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3508815284895559e-05,
+      "loss": 2.8874,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.338847404664667e-05,
+      "loss": 2.9797,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.32686639560195e-05,
+      "loss": 2.9249,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3149385143789383e-05,
+      "loss": 2.9447,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3030637740151951e-05,
+      "loss": 2.9772,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2912421874722747e-05,
+      "loss": 2.9938,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2794737676536993e-05,
+      "loss": 3.053,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2677585274049774e-05,
+      "loss": 2.9174,
+      "step": 2885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2560964795135632e-05,
+      "loss": 2.988,
+      "step": 2886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2444876367088354e-05,
+      "loss": 2.9329,
+      "step": 2887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2329320116621245e-05,
+      "loss": 2.9624,
+      "step": 2888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2214296169866579e-05,
+      "loss": 2.9761,
+      "step": 2889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.209980465237559e-05,
+      "loss": 3.0035,
+      "step": 2890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.198584568911848e-05,
+      "loss": 3.0056,
+      "step": 2891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1872419404483915e-05,
+      "loss": 2.9445,
+      "step": 2892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1759525922279412e-05,
+      "loss": 2.9952,
+      "step": 2893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.164716536573074e-05,
+      "loss": 3.0598,
+      "step": 2894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1535337857482009e-05,
+      "loss": 2.9805,
+      "step": 2895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1424043519595528e-05,
+      "loss": 2.8535,
+      "step": 2896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1313282473551623e-05,
+      "loss": 2.8443,
+      "step": 2897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1203054840248473e-05,
+      "loss": 2.9117,
+      "step": 2898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1093360740002056e-05,
+      "loss": 2.8759,
+      "step": 2899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0984200292546099e-05,
+      "loss": 2.8919,
+      "step": 2900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0875573617031565e-05,
+      "loss": 2.9113,
+      "step": 2901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0767480832027166e-05,
+      "loss": 3.0646,
+      "step": 2902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0659922055518467e-05,
+      "loss": 3.0283,
+      "step": 2903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.055289740490839e-05,
+      "loss": 2.8423,
+      "step": 2904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0446406997016822e-05,
+      "loss": 3.1122,
+      "step": 2905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.034045094808045e-05,
+      "loss": 2.9487,
+      "step": 2906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0235029373752758e-05,
+      "loss": 2.9625,
+      "step": 2907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0130142389103813e-05,
+      "loss": 2.992,
+      "step": 2908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0025790108620092e-05,
+      "loss": 2.9302,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.921972646204535e-06,
+      "loss": 2.8519,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.818690115176276e-06,
+      "loss": 2.9158,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.715942628270635e-06,
+      "loss": 3.0094,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.61373029763879e-06,
+      "loss": 2.9175,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.512053234847774e-06,
+      "loss": 3.0496,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.410911550880474e-06,
+      "loss": 3.0325,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.310305356135418e-06,
+      "loss": 2.9693,
+      "step": 2916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.210234760426483e-06,
+      "loss": 2.9899,
+      "step": 2917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.110699872983131e-06,
+      "loss": 3.0623,
+      "step": 2918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.011700802449841e-06,
+      "loss": 2.9553,
+      "step": 2919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.913237656886398e-06,
+      "loss": 2.9209,
+      "step": 2920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.815310543767607e-06,
+      "loss": 3.0354,
+      "step": 2921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.71791956998319e-06,
+      "loss": 3.0742,
+      "step": 2922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.6210648418375e-06,
+      "loss": 2.9825,
+      "step": 2923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.524746465049859e-06,
+      "loss": 2.9807,
+      "step": 2924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.428964544753892e-06,
+      "loss": 2.9209,
+      "step": 2925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.33371918549769e-06,
+      "loss": 2.9916,
+      "step": 2926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.23901049124387e-06,
+      "loss": 2.8611,
+      "step": 2927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.144838565369072e-06,
+      "loss": 3.0244,
+      "step": 2928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.05120351066413e-06,
+      "loss": 2.9548,
+      "step": 2929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.958105429333784e-06,
+      "loss": 3.0994,
+      "step": 2930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.865544422996863e-06,
+      "loss": 3.0703,
+      "step": 2931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.773520592685657e-06,
+      "loss": 2.9131,
+      "step": 2932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.68203403884643e-06,
+      "loss": 2.9855,
+      "step": 2933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.591084861338748e-06,
+      "loss": 2.9434,
+      "step": 2934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.500673159435756e-06,
+      "loss": 2.9152,
+      "step": 2935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4107990318239e-06,
+      "loss": 2.9683,
+      "step": 2936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.321462576602877e-06,
+      "loss": 2.9787,
+      "step": 2937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.232663891285407e-06,
+      "loss": 3.0803,
+      "step": 2938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.144403072797345e-06,
+      "loss": 2.9853,
+      "step": 2939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.056680217477407e-06,
+      "loss": 2.982,
+      "step": 2940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.969495421076999e-06,
+      "loss": 2.9903,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.882848778760387e-06,
+      "loss": 2.9062,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.796740385104416e-06,
+      "loss": 2.8146,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.7111703340982935e-06,
+      "loss": 3.0582,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.626138719143693e-06,
+      "loss": 3.0897,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.541645633054649e-06,
+      "loss": 2.9442,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.457691168057222e-06,
+      "loss": 2.9167,
+      "step": 2947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3742754157896655e-06,
+      "loss": 2.8642,
+      "step": 2948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.291398467302145e-06,
+      "loss": 2.7971,
+      "step": 2949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.209060413056799e-06,
+      "loss": 2.99,
+      "step": 2950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.127261342927404e-06,
+      "loss": 2.8999,
+      "step": 2951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.046001346199648e-06,
+      "loss": 3.0001,
+      "step": 2952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.965280511570636e-06,
+      "loss": 2.9963,
+      "step": 2953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.885098927148946e-06,
+      "loss": 3.0147,
+      "step": 2954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.805456680454624e-06,
+      "loss": 3.0048,
+      "step": 2955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7263538584190755e-06,
+      "loss": 2.9379,
+      "step": 2956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.647790547384845e-06,
+      "loss": 2.9875,
+      "step": 2957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5697668331054496e-06,
+      "loss": 2.9582,
+      "step": 2958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.492282800745707e-06,
+      "loss": 2.9387,
+      "step": 2959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.415338534881131e-06,
+      "loss": 3.0862,
+      "step": 2960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.338934119498207e-06,
+      "loss": 3.0562,
+      "step": 2961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.263069637994111e-06,
+      "loss": 2.9947,
+      "step": 2962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.18774517317655e-06,
+      "loss": 2.9725,
+      "step": 2963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.112960807263978e-06,
+      "loss": 2.9932,
+      "step": 2964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.038716621885264e-06,
+      "loss": 3.0604,
+      "step": 2965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.965012698079641e-06,
+      "loss": 2.9706,
+      "step": 2966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.891849116296532e-06,
+      "loss": 2.8735,
+      "step": 2967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.819225956395834e-06,
+      "loss": 2.9544,
+      "step": 2968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.747143297647194e-06,
+      "loss": 2.9994,
+      "step": 2969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.675601218730618e-06,
+      "loss": 3.035,
+      "step": 2970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.604599797735864e-06,
+      "loss": 2.9517,
+      "step": 2971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.534139112162605e-06,
+      "loss": 2.9484,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.464219238920375e-06,
+      "loss": 2.9975,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.394840254328236e-06,
+      "loss": 2.9462,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.326002234114945e-06,
+      "loss": 3.0088,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.257705253418787e-06,
+      "loss": 3.0166,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.189949386787462e-06,
+      "loss": 2.9709,
+      "step": 2977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.122734708178088e-06,
+      "loss": 3.0078,
+      "step": 2978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.05606129095698e-06,
+      "loss": 3.0701,
+      "step": 2979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.989929207899756e-06,
+      "loss": 2.9283,
+      "step": 2980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.924338531191063e-06,
+      "loss": 3.0114,
+      "step": 2981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8592893324246315e-06,
+      "loss": 2.8951,
+      "step": 2982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7947816826031657e-06,
+      "loss": 3.0315,
+      "step": 2983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7308156521382306e-06,
+      "loss": 2.9825,
+      "step": 2984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.667391310850199e-06,
+      "loss": 2.9646,
+      "step": 2985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.604508727968192e-06,
+      "loss": 2.9898,
+      "step": 2986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5421679721299726e-06,
+      "loss": 2.9788,
+      "step": 2987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.480369111381998e-06,
+      "loss": 2.996,
+      "step": 2988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.419112213179032e-06,
+      "loss": 2.8949,
+      "step": 2989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3583973443844784e-06,
+      "loss": 2.9663,
+      "step": 2990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.298224571269881e-06,
+      "loss": 2.9821,
+      "step": 2991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2385939595153124e-06,
+      "loss": 3.0838,
+      "step": 2992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.179505574208874e-06,
+      "loss": 3.0757,
+      "step": 2993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.120959479846919e-06,
+      "loss": 2.9847,
+      "step": 2994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0629557403338836e-06,
+      "loss": 2.9186,
+      "step": 2995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0054944189821243e-06,
+      "loss": 2.9156,
+      "step": 2996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.948575578512025e-06,
+      "loss": 3.0902,
+      "step": 2997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8921992810518328e-06,
+      "loss": 2.8893,
+      "step": 2998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8363655881374906e-06,
+      "loss": 2.8615,
+      "step": 2999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7810745607127486e-06,
+      "loss": 2.9979,
+      "step": 3000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7263262591291638e-06,
+      "loss": 2.8662,
+      "step": 3001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.672120743145601e-06,
+      "loss": 2.9249,
+      "step": 3002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6184580719286755e-06,
+      "loss": 2.969,
+      "step": 3003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5653383040524227e-06,
+      "loss": 3.0145,
+      "step": 3004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5127614974983504e-06,
+      "loss": 2.8966,
+      "step": 3005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.460727709655164e-06,
+      "loss": 3.0293,
+      "step": 3006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4092369973189308e-06,
+      "loss": 3.0221,
+      "step": 3007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3582894166930268e-06,
+      "loss": 2.9962,
+      "step": 3008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3078850233878013e-06,
+      "loss": 2.9829,
+      "step": 3009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2580238724208555e-06,
+      "loss": 2.8966,
+      "step": 3010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2087060182168215e-06,
+      "loss": 3.0617,
+      "step": 3011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.159931514607194e-06,
+      "loss": 2.9882,
+      "step": 3012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1117004148304418e-06,
+      "loss": 2.9314,
+      "step": 3013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.064012771532009e-06,
+      "loss": 3.0016,
+      "step": 3014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.01686863676398e-06,
+      "loss": 3.0379,
+      "step": 3015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.970268061985303e-06,
+      "loss": 2.8857,
+      "step": 3016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9242110980615123e-06,
+      "loss": 3.0664,
+      "step": 3017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8786977952648944e-06,
+      "loss": 2.9524,
+      "step": 3018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8337282032742098e-06,
+      "loss": 2.9744,
+      "step": 3019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7893023711748613e-06,
+      "loss": 2.8596,
+      "step": 3020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7454203474585596e-06,
+      "loss": 2.913,
+      "step": 3021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7020821800236563e-06,
+      "loss": 2.9144,
+      "step": 3022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.659287916174701e-06,
+      "loss": 3.1117,
+      "step": 3023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6170376026226064e-06,
+      "loss": 2.8773,
+      "step": 3024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753312854845935e-06,
+      "loss": 3.0182,
+      "step": 3025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.53416901028397e-06,
+      "loss": 3.0532,
+      "step": 3026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4935508219504067e-06,
+      "loss": 2.9159,
+      "step": 3027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4534767648195503e-06,
+      "loss": 2.9009,
+      "step": 3028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4139468826331325e-06,
+      "loss": 2.9993,
+      "step": 3029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3749612185389172e-06,
+      "loss": 2.8559,
+      "step": 3030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3365198150906976e-06,
+      "loss": 2.9495,
+      "step": 3031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2986227142480766e-06,
+      "loss": 3.0426,
+      "step": 3032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2612699573766873e-06,
+      "loss": 2.9086,
+      "step": 3033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2244615852479157e-06,
+      "loss": 3.0583,
+      "step": 3034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.188197638038846e-06,
+      "loss": 3.0339,
+      "step": 3035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1524781553324814e-06,
+      "loss": 2.9293,
+      "step": 3036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.117303176117468e-06,
+      "loss": 2.9114,
+      "step": 3037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0826727387880376e-06,
+      "loss": 3.0579,
+      "step": 3038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0485868811441758e-06,
+      "loss": 2.9717,
+      "step": 3039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0150456403913432e-06,
+      "loss": 2.987,
+      "step": 3040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.820490531405323e-07,
+      "loss": 3.1101,
+      "step": 3041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.495971554083216e-07,
+      "loss": 3.0442,
+      "step": 3042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.17689982616765e-07,
+      "loss": 2.9431,
+      "step": 3043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.86327569593115e-07,
+      "loss": 2.9751,
+      "step": 3044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.555099505703213e-07,
+      "loss": 2.8528,
+      "step": 3045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.25237159186365e-07,
+      "loss": 2.9906,
+      "step": 3046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.955092284848142e-07,
+      "loss": 2.8804,
+      "step": 3047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.663261909143793e-07,
+      "loss": 2.8058,
+      "step": 3048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.376880783289131e-07,
+      "loss": 2.9914,
+      "step": 3049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.095949219876885e-07,
+      "loss": 2.9278,
+      "step": 3050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.820467525549546e-07,
+      "loss": 2.873,
+      "step": 3051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.550436001002136e-07,
+      "loss": 3.0636,
+      "step": 3052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.285854940979996e-07,
+      "loss": 2.9178,
+      "step": 3053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.026724634279335e-07,
+      "loss": 3.0012,
+      "step": 3054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.773045363746677e-07,
+      "loss": 2.8893,
+      "step": 3055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.524817406279415e-07,
+      "loss": 2.9697,
+      "step": 3056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.282041032823592e-07,
+      "loss": 3.0442,
+      "step": 3057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.044716508375014e-07,
+      "loss": 2.9752,
+      "step": 3058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.81284409197924e-07,
+      "loss": 3.002,
+      "step": 3059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5864240367299305e-07,
+      "loss": 2.9447,
+      "step": 3060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.365456589769945e-07,
+      "loss": 2.9007,
+      "step": 3061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1499419922907956e-07,
+      "loss": 2.9498,
+      "step": 3062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.939880479530422e-07,
+      "loss": 2.783,
+      "step": 3063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.73527228077708e-07,
+      "loss": 2.9055,
+      "step": 3064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.536117619364343e-07,
+      "loss": 2.7818,
+      "step": 3065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.342416712674989e-07,
+      "loss": 3.0663,
+      "step": 3066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1541697721371165e-07,
+      "loss": 2.9317,
+      "step": 3067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.971377003226916e-07,
+      "loss": 3.0031,
+      "step": 3068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7940386054664537e-07,
+      "loss": 3.0544,
+      "step": 3069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6221547724253336e-07,
+      "loss": 2.9911,
+      "step": 3070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.45572569171848e-07,
+      "loss": 3.0079,
+      "step": 3071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2947515450072454e-07,
+      "loss": 2.9155,
+      "step": 3072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1392325079983011e-07,
+      "loss": 2.9659,
+      "step": 3073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9891687504436372e-07,
+      "loss": 2.8954,
+      "step": 3074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8445604361427837e-07,
+      "loss": 2.9006,
+      "step": 3075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7054077229372579e-07,
+      "loss": 3.0201,
+      "step": 3076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5717107627172267e-07,
+      "loss": 3.0236,
+      "step": 3077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4434697014148456e-07,
+      "loss": 3.0125,
+      "step": 3078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.320684679008144e-07,
+      "loss": 2.9281,
+      "step": 3079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2033558295199144e-07,
+      "loss": 2.9419,
+      "step": 3080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0914832810177133e-07,
+      "loss": 3.0009,
+      "step": 3081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.85067155612751e-08,
+      "loss": 2.9223,
+      "step": 3082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.841075694604461e-08,
+      "loss": 3.003,
+      "step": 3083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.886046327609808e-08,
+      "loss": 3.0325,
+      "step": 3084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.985584497576359e-08,
+      "loss": 2.9955,
+      "step": 3085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.139691187373453e-08,
+      "loss": 2.9892,
+      "step": 3086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3483673203236215e-08,
+      "loss": 2.9334,
+      "step": 3087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6116137601692754e-08,
+      "loss": 2.9557,
+      "step": 3088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.929431311094911e-08,
+      "loss": 2.9163,
+      "step": 3089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.30182071772156e-08,
+      "loss": 2.8729,
+      "step": 3090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.728782665095686e-08,
+      "loss": 2.8521,
+      "step": 3091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.210317778700288e-08,
+      "loss": 2.9092,
+      "step": 3092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7464266244549e-08,
+      "loss": 3.0195,
+      "step": 3093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.337109708704487e-08,
+      "loss": 2.9739,
+      "step": 3094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.823674782305503e-09,
+      "loss": 2.9205,
+      "step": 3095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.82200320245574e-09,
+      "loss": 2.9763,
+      "step": 3096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.366085623819238e-09,
+      "loss": 2.8237,
+      "step": 3097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.455924727140513e-09,
+      "loss": 2.9198,
+      "step": 3098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0915225973628928e-09,
+      "loss": 2.9677,
+      "step": 3099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.728807237950548e-10,
+      "loss": 2.9228,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.8494,
+      "step": 3101
+    },
+    {
+      "epoch": 1.0,
+      "step": 3101,
+      "total_flos": 3.908356201733161e+17,
+      "train_loss": 3.100002513320859,
+      "train_runtime": 8586.363,
+      "train_samples_per_second": 69.34,
+      "train_steps_per_second": 0.361
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3101,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 3.908356201733161e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}