diff --git "a/checkpoints_ft/llava_raw_200k/llava-v1.5-13b-pretrain/trainer_state.json" "b/checkpoints_ft/llava_raw_200k/llava-v1.5-13b-pretrain/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints_ft/llava_raw_200k/llava-v1.5-13b-pretrain/trainer_state.json"
@@ -0,0 +1,9408 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1563,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-05,
+      "loss": 7.108,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-05,
+      "loss": 7.1953,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.382978723404255e-05,
+      "loss": 6.5838,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-05,
+      "loss": 6.0248,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010638297872340425,
+      "loss": 5.2211,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001276595744680851,
+      "loss": 4.8425,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014893617021276596,
+      "loss": 4.6524,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017021276595744682,
+      "loss": 4.813,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019148936170212765,
+      "loss": 4.6359,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002127659574468085,
+      "loss": 4.5986,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023404255319148937,
+      "loss": 4.2365,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002553191489361702,
+      "loss": 4.3282,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002765957446808511,
+      "loss": 4.2216,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002978723404255319,
+      "loss": 4.2701,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003191489361702128,
+      "loss": 4.2888,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034042553191489364,
+      "loss": 4.0663,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003617021276595745,
+      "loss": 4.0237,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003829787234042553,
+      "loss": 3.9743,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040425531914893613,
+      "loss": 3.8384,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000425531914893617,
+      "loss": 3.8563,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044680851063829785,
+      "loss": 3.9171,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046808510638297874,
+      "loss": 3.7824,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004893617021276596,
+      "loss": 3.8326,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005106382978723404,
+      "loss": 3.7543,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005319148936170213,
+      "loss": 3.6159,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005531914893617022,
+      "loss": 3.8324,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005744680851063831,
+      "loss": 3.7801,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005957446808510638,
+      "loss": 3.6986,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006170212765957447,
+      "loss": 3.514,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006382978723404256,
+      "loss": 3.5191,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006595744680851064,
+      "loss": 3.6786,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006808510638297873,
+      "loss": 3.4974,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007021276595744682,
+      "loss": 3.48,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000723404255319149,
+      "loss": 3.503,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007446808510638298,
+      "loss": 3.5276,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007659574468085106,
+      "loss": 3.4442,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007872340425531915,
+      "loss": 3.4631,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008085106382978723,
+      "loss": 3.4147,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008297872340425531,
+      "loss": 3.4706,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000851063829787234,
+      "loss": 3.5004,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008723404255319149,
+      "loss": 3.4641,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008936170212765957,
+      "loss": 3.4686,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009148936170212766,
+      "loss": 3.3098,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009361702127659575,
+      "loss": 3.1808,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009574468085106384,
+      "loss": 3.1921,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009787234042553192,
+      "loss": 3.3104,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 3.3155,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999989264032304,
+      "loss": 3.2819,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999957056175321,
+      "loss": 3.2802,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999903376567364,
+      "loss": 3.0536,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999828225438954,
+      "loss": 3.101,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999731603112817,
+      "loss": 3.1781,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999961351000389,
+      "loss": 3.0154,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999947394661931,
+      "loss": 3.1807,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999312913558413,
+      "loss": 3.1276,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999913041151274,
+      "loss": 3.1217,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998926441266026,
+      "loss": 3.0376,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998701003694195,
+      "loss": 3.1796,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998454099765367,
+      "loss": 3.1062,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998185730539842,
+      "loss": 3.0824,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997895897170099,
+      "loss": 3.0683,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997584600900799,
+      "loss": 2.8955,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999725184306876,
+      "loss": 3.1632,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996897625102986,
+      "loss": 2.878,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996521948524616,
+      "loss": 2.9642,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996124814946952,
+      "loss": 3.0548,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995706226075445,
+      "loss": 2.9631,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999526618370767,
+      "loss": 2.9474,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994804689733344,
+      "loss": 2.9853,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00099943217461343,
+      "loss": 2.9542,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993817354984485,
+      "loss": 2.984,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993291518449947,
+      "loss": 3.0298,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992744238788835,
+      "loss": 3.0919,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992175518351379,
+      "loss": 2.9055,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991585359579885,
+      "loss": 2.993,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999097376500872,
+      "loss": 3.0122,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990340737264312,
+      "loss": 2.9489,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989686279065124,
+      "loss": 2.8933,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989010393221655,
+      "loss": 2.9019,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988313082636422,
+      "loss": 2.9815,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987594350303942,
+      "loss": 2.8513,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998685419931073,
+      "loss": 2.9635,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986092632835286,
+      "loss": 2.9016,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985309654148068,
+      "loss": 2.881,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998450526661149,
+      "loss": 2.8898,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983679473679903,
+      "loss": 2.709,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998283227889958,
+      "loss": 2.8043,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981963685908705,
+      "loss": 2.9313,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981073698437355,
+      "loss": 2.7111,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980162320307477,
+      "loss": 2.8254,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979229555432882,
+      "loss": 2.8376,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978275407819225,
+      "loss": 2.6649,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977299881563984,
+      "loss": 2.8524,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976302980856447,
+      "loss": 2.7397,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997528470997769,
+      "loss": 2.6912,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974245073300564,
+      "loss": 2.8565,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997318407528967,
+      "loss": 2.789,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972101720501347,
+      "loss": 2.84,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970998013583642,
+      "loss": 2.774,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099698729592763,
+      "loss": 2.6355,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968726562410743,
+      "loss": 2.8173,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996755882791004,
+      "loss": 2.6143,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966369760788896,
+      "loss": 2.6691,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965159366153622,
+      "loss": 2.6734,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963927649202127,
+      "loss": 2.7468,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962674615223877,
+      "loss": 2.6971,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961400269599885,
+      "loss": 2.7469,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960104617802686,
+      "loss": 2.7382,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958787665396308,
+      "loss": 2.7565,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957449418036256,
+      "loss": 2.6273,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956089881469482,
+      "loss": 2.7472,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995470906153436,
+      "loss": 2.588,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995330696416067,
+      "loss": 2.6636,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951883595369555,
+      "loss": 2.5938,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950438961273517,
+      "loss": 2.7012,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994897306807637,
+      "loss": 2.5484,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994748592207323,
+      "loss": 2.5682,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945977529650476,
+      "loss": 2.583,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944447897285728,
+      "loss": 2.668,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942897031547822,
+      "loss": 2.5936,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941324939096773,
+      "loss": 2.514,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939731626683755,
+      "loss": 2.6285,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993811710115107,
+      "loss": 2.5116,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936481369432115,
+      "loss": 2.4464,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934824438551353,
+      "loss": 2.5799,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933146315624288,
+      "loss": 2.5832,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931447007857433,
+      "loss": 2.5727,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929726522548267,
+      "loss": 2.6191,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927984867085223,
+      "loss": 2.6684,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926222048947644,
+      "loss": 2.4844,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924438075705755,
+      "loss": 2.6227,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922632955020622,
+      "loss": 2.6172,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920806694644138,
+      "loss": 2.5165,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991895930241897,
+      "loss": 2.5375,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917090786278533,
+      "loss": 2.6482,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915201154246961,
+      "loss": 2.6062,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913290414439067,
+      "loss": 2.5604,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911358575060304,
+      "loss": 2.5019,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909405644406737,
+      "loss": 2.5783,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990743163086501,
+      "loss": 2.643,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905436542912301,
+      "loss": 2.4923,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903420389116287,
+      "loss": 2.5403,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901383178135112,
+      "loss": 2.526,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899324918717354,
+      "loss": 2.3946,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897245619701971,
+      "loss": 2.4762,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895145290018282,
+      "loss": 2.5249,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989302393868591,
+      "loss": 2.5589,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890881574814764,
+      "loss": 2.5168,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888718207604979,
+      "loss": 2.5024,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886533846346897,
+      "loss": 2.4461,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884328500421005,
+      "loss": 2.5174,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882102179297914,
+      "loss": 2.5289,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987985489253831,
+      "loss": 2.4729,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987758664979291,
+      "loss": 2.4637,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987529746080243,
+      "loss": 2.5186,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872987335397532,
+      "loss": 2.4407,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870656283498785,
+      "loss": 2.4552,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868304315116634,
+      "loss": 2.4524,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865931440351337,
+      "loss": 2.4382,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986353766939294,
+      "loss": 2.5236,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986112301252122,
+      "loss": 2.3765,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858687480105645,
+      "loss": 2.3683,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856231082605344,
+      "loss": 2.3164,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853753830569027,
+      "loss": 2.3975,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851255734634981,
+      "loss": 2.4327,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848736805530996,
+      "loss": 2.3305,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846197054074325,
+      "loss": 2.4151,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843636491171648,
+      "loss": 2.3239,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841055127819008,
+      "loss": 2.3778,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838452975101783,
+      "loss": 2.4159,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835830044194625,
+      "loss": 2.4867,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983318634636141,
+      "loss": 2.2631,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830521892955201,
+      "loss": 2.3465,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827836695418194,
+      "loss": 2.2615,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825130765281668,
+      "loss": 2.3477,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982240411416593,
+      "loss": 2.3642,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981965675378028,
+      "loss": 2.3737,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981688869592294,
+      "loss": 2.3817,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981409995248103,
+      "loss": 2.3798,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000981129053543049,
+      "loss": 2.2757,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808460456836045,
+      "loss": 2.4109,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805609728851148,
+      "loss": 2.3866,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802738363717926,
+      "loss": 2.2564,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799846373767134,
+      "loss": 2.2494,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796933771418098,
+      "loss": 2.349,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794000569178659,
+      "loss": 2.4194,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791046779645122,
+      "loss": 2.3216,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097880724155022,
+      "loss": 2.3861,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785077489522969,
+      "loss": 2.2825,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782062014568797,
+      "loss": 2.3263,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779026003589304,
+      "loss": 2.2013,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775969469622292,
+      "loss": 2.2716,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772892425793705,
+      "loss": 2.4497,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769794885317558,
+      "loss": 2.2882,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766676861495888,
+      "loss": 2.2883,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763538367718698,
+      "loss": 2.3532,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760379417463893,
+      "loss": 2.3784,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757200024297232,
+      "loss": 2.3293,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754000201872258,
+      "loss": 2.3501,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750779963930245,
+      "loss": 2.3574,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747539324300144,
+      "loss": 2.2178,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744278296898515,
+      "loss": 2.3093,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974099689572947,
+      "loss": 2.4095,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737695134884619,
+      "loss": 2.3221,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734373028543,
+      "loss": 2.3116,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731030590971022,
+      "loss": 2.3332,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727667836522407,
+      "loss": 2.251,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724284779638126,
+      "loss": 2.3526,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720881434846331,
+      "loss": 2.2599,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717457816762305,
+      "loss": 2.2631,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009714013940088388,
+      "loss": 2.4821,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710549819613918,
+      "loss": 2.2771,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707065470215173,
+      "loss": 2.335,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703560906855295,
+      "loss": 2.2848,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700036144584237,
+      "loss": 2.2959,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696491198538691,
+      "loss": 2.3476,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692926083942028,
+      "loss": 2.3183,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689340816104232,
+      "loss": 2.3273,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968573541042183,
+      "loss": 2.2907,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682109882377828,
+      "loss": 2.2288,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678464247541647,
+      "loss": 2.3493,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674798521569056,
+      "loss": 2.2174,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00096711127202021,
+      "loss": 2.2263,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667406859269037,
+      "loss": 2.1875,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663680954684268,
+      "loss": 2.1721,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000965993502244827,
+      "loss": 2.2401,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656169078647525,
+      "loss": 2.2895,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652383139454455,
+      "loss": 2.2994,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648577221127346,
+      "loss": 2.304,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644751340010287,
+      "loss": 2.3248,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640905512533091,
+      "loss": 2.1651,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637039755211232,
+      "loss": 2.4588,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633154084645766,
+      "loss": 2.3051,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629248517523265,
+      "loss": 2.201,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000962532307061575,
+      "loss": 2.2204,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621377760780606,
+      "loss": 2.2385,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617412604960523,
+      "loss": 2.2268,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613427620183412,
+      "loss": 2.1924,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609422823562345,
+      "loss": 2.1914,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605398232295464,
+      "loss": 2.2397,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601353863665925,
+      "loss": 2.2382,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000959728973504181,
+      "loss": 2.1397,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593205863876062,
+      "loss": 2.3034,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589102267706405,
+      "loss": 2.2742,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584978964155267,
+      "loss": 2.1876,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958083597092971,
+      "loss": 2.1879,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576673305821353,
+      "loss": 2.2453,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572490986706289,
+      "loss": 2.1252,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568289031545017,
+      "loss": 2.1716,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564067458382357,
+      "loss": 2.226,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559826285347379,
+      "loss": 2.193,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555565530653322,
+      "loss": 2.2524,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551285212597516,
+      "loss": 2.2862,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546985349561302,
+      "loss": 2.2861,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542665960009959,
+      "loss": 2.2567,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538327062492616,
+      "loss": 2.2102,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533968675642177,
+      "loss": 2.1723,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529590818175245,
+      "loss": 2.3281,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525193508892034,
+      "loss": 2.148,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952077676667629,
+      "loss": 2.3173,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516340610495216,
+      "loss": 2.1974,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511885059399383,
+      "loss": 2.1511,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009507410132522651,
+      "loss": 2.2892,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950291584908209,
+      "loss": 2.266,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498402228377891,
+      "loss": 2.2416,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493869289793291,
+      "loss": 2.14,
+      "step": 266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009489317052794481,
+      "loss": 2.3052,
+      "step": 267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484745536930528,
+      "loss": 2.2679,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480154761833292,
+      "loss": 2.2388,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475544747217337,
+      "loss": 2.3174,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470915512879852,
+      "loss": 2.2157,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466267078700561,
+      "loss": 2.3632,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461599464641637,
+      "loss": 2.0922,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456912690747624,
+      "loss": 2.2296,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452206777145342,
+      "loss": 2.1478,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447481744043808,
+      "loss": 2.1135,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000944273761173414,
+      "loss": 2.2915,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000943797440058948,
+      "loss": 2.1634,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094331921310649,
+      "loss": 2.0843,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428390823697317,
+      "loss": 2.1926,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423570499105402,
+      "loss": 2.2488,
+      "step": 281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418731177989497,
+      "loss": 2.18,
+      "step": 282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413872881131517,
+      "loss": 2.2783,
+      "step": 283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408995629394872,
+      "loss": 2.1865,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404099443724368,
+      "loss": 2.2078,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399184345146121,
+      "loss": 2.1406,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394250354767467,
+      "loss": 2.1094,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000938929749377687,
+      "loss": 2.2685,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384325783443832,
+      "loss": 2.2398,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379335245118803,
+      "loss": 2.1341,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374325900233087,
+      "loss": 2.165,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369297770298747,
+      "loss": 2.2718,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364250876908523,
+      "loss": 2.274,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009359185241735724,
+      "loss": 2.2583,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009354100886534151,
+      "loss": 2.2744,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009348997833137995,
+      "loss": 2.1787,
+      "step": 296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000934387610346174,
+      "loss": 2.1388,
+      "step": 297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338735719500075,
+      "loss": 2.1725,
+      "step": 298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333576703327802,
+      "loss": 2.2194,
+      "step": 299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009328399077099729,
+      "loss": 2.1777,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932320286305059,
+      "loss": 2.2725,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317988083494942,
+      "loss": 2.3469,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931275476082706,
+      "loss": 2.2439,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009307502917520863,
+      "loss": 2.2337,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009302232576129797,
+      "loss": 2.2943,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296943759286748,
+      "loss": 2.1714,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009291636489703941,
+      "loss": 2.243,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009286310790172851,
+      "loss": 2.2194,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280966683564087,
+      "loss": 2.2637,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275604192827316,
+      "loss": 2.144,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270223340991146,
+      "loss": 2.2584,
+      "step": 311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926482415116304,
+      "loss": 2.2547,
+      "step": 312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259406646529209,
+      "loss": 2.2587,
+      "step": 313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253970850354511,
+      "loss": 2.2832,
+      "step": 314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009248516785982364,
+      "loss": 2.201,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243044476834628,
+      "loss": 2.268,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237553946411519,
+      "loss": 2.1911,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232045218291498,
+      "loss": 2.3109,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009226518316131176,
+      "loss": 2.3246,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220973263665211,
+      "loss": 2.1487,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215410084706206,
+      "loss": 2.1668,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209828803144602,
+      "loss": 2.1548,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009204229442948584,
+      "loss": 2.1904,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198612028163973,
+      "loss": 2.0854,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919297658291412,
+      "loss": 2.2675,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000918732313139981,
+      "loss": 2.102,
+      "step": 326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181651697899152,
+      "loss": 2.199,
+      "step": 327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175962306767476,
+      "loss": 2.184,
+      "step": 328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917025498243723,
+      "loss": 2.2724,
+      "step": 329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009164529749417875,
+      "loss": 2.116,
+      "step": 330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009158786632295776,
+      "loss": 2.229,
+      "step": 331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153025655734102,
+      "loss": 2.2581,
+      "step": 332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009147246844472717,
+      "loss": 2.1557,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914145022332807,
+      "loss": 2.2587,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009135635817193099,
+      "loss": 2.2502,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009129803651037117,
+      "loss": 2.1707,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009123953749905697,
+      "loss": 2.2182,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118086138920582,
+      "loss": 2.1679,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112200843279565,
+      "loss": 2.11,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009106297888256383,
+      "loss": 2.199,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100377299200611,
+      "loss": 2.1865,
+      "step": 341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009094439101537548,
+      "loss": 2.2557,
+      "step": 342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009088483320768114,
+      "loss": 2.2548,
+      "step": 343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082509982468738,
+      "loss": 2.153,
+      "step": 344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009076519112291246,
+      "loss": 2.2516,
+      "step": 345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009070510735962753,
+      "loss": 2.1287,
+      "step": 346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009064484879285555,
+      "loss": 2.2911,
+      "step": 347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000905844156813701,
+      "loss": 2.0597,
+      "step": 348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052380828469435,
+      "loss": 2.0511,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046302686309997,
+      "loss": 2.1439,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040207167760586,
+      "loss": 2.2244,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000903409429899772,
+      "loss": 2.2189,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009027964106272424,
+      "loss": 2.1356,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021816615910118,
+      "loss": 2.09,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015651854310505,
+      "loss": 2.2249,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009469847947459,
+      "loss": 2.2083,
+      "step": 356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003270623368906,
+      "loss": 2.2736,
+      "step": 357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008997054207196717,
+      "loss": 2.2922,
+      "step": 358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990820626126589,
+      "loss": 2.2362,
+      "step": 359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008984569906927933,
+      "loss": 2.1109,
+      "step": 360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008978302076443754,
+      "loss": 2.1859,
+      "step": 361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008972017161590546,
+      "loss": 2.4003,
+      "step": 362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008965715189358165,
+      "loss": 2.2753,
+      "step": 363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959396186809716,
+      "loss": 2.1252,
+      "step": 364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008953060181081448,
+      "loss": 2.2287,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008946707199382615,
+      "loss": 2.0739,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008940337268995384,
+      "loss": 2.1275,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933950417274701,
+      "loss": 2.2411,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000892754667164818,
+      "loss": 2.1115,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008921126059615983,
+      "loss": 2.1571,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008914688608750701,
+      "loss": 2.2391,
+      "step": 371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008908234346697244,
+      "loss": 2.119,
+      "step": 372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008901763301172708,
+      "loss": 2.1566,
+      "step": 373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895275499966269,
+      "loss": 2.0553,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008888770970939058,
+      "loss": 2.203,
+      "step": 375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882249742024036,
+      "loss": 2.1265,
+      "step": 376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008875711841225889,
+      "loss": 2.2856,
+      "step": 377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008869157296620892,
+      "loss": 2.0852,
+      "step": 378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008862586136356794,
+      "loss": 2.2626,
+      "step": 379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008855998388652705,
+      "loss": 2.0272,
+      "step": 380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849394081798962,
+      "loss": 2.1022,
+      "step": 381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008842773244157014,
+      "loss": 2.207,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008836135904159301,
+      "loss": 2.1352,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008829482090309131,
+      "loss": 2.1552,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008822811831180555,
+      "loss": 2.1444,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008816125155418249,
+      "loss": 2.2295,
+      "step": 386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809422091737387,
+      "loss": 2.1854,
+      "step": 387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802702668923517,
+      "loss": 2.2542,
+      "step": 388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008795966915832443,
+      "loss": 2.3112,
+      "step": 389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008789214861390095,
+      "loss": 2.2477,
+      "step": 390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008782446534592412,
+      "loss": 2.1792,
+      "step": 391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008775661964505204,
+      "loss": 2.2727,
+      "step": 392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008768861180264045,
+      "loss": 2.2526,
+      "step": 393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762044211074132,
+      "loss": 2.2027,
+      "step": 394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008755211086210172,
+      "loss": 2.1673,
+      "step": 395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748361835016246,
+      "loss": 2.1185,
+      "step": 396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008741496486905691,
+      "loss": 2.1435,
+      "step": 397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873461507136097,
+      "loss": 2.1853,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008727717617933544,
+      "loss": 2.0286,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720804156243747,
+      "loss": 2.183,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008713874715980661,
+      "loss": 2.2431,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008706929326901982,
+      "loss": 2.2307,
+      "step": 402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008699968018833904,
+      "loss": 2.3219,
+      "step": 403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692990821670974,
+      "loss": 2.1649,
+      "step": 404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008685997765375981,
+      "loss": 2.2494,
+      "step": 405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008678988879979812,
+      "loss": 2.1301,
+      "step": 406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008671964195581336,
+      "loss": 2.0965,
+      "step": 407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008664923742347266,
+      "loss": 2.2184,
+      "step": 408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008657867550512033,
+      "loss": 2.2106,
+      "step": 409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008650795650377659,
+      "loss": 2.172,
+      "step": 410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008643708072313618,
+      "loss": 2.0944,
+      "step": 411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008636604846756713,
+      "loss": 2.1609,
+      "step": 412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008629486004210945,
+      "loss": 2.1832,
+      "step": 413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862235157524738,
+      "loss": 2.0838,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008615201590504016,
+      "loss": 2.1057,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008608036080685658,
+      "loss": 2.1207,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008600855076563776,
+      "loss": 2.0985,
+      "step": 417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593658608976383,
+      "loss": 2.1249,
+      "step": 418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008586446708827895,
+      "loss": 2.1138,
+      "step": 419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579219407089003,
+      "loss": 2.1276,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857197673479654,
+      "loss": 2.1115,
+      "step": 421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008564718723053342,
+      "loss": 2.1083,
+      "step": 422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008557445403028121,
+      "loss": 2.1589,
+      "step": 423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008550156805955331,
+      "loss": 2.2647,
+      "step": 424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008542852963135028,
+      "loss": 2.3004,
+      "step": 425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008535533905932737,
+      "loss": 2.1676,
+      "step": 426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008528199665779328,
+      "loss": 2.227,
+      "step": 427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008520850274170863,
+      "loss": 2.0909,
+      "step": 428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008513485762668475,
+      "loss": 2.1056,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000850610616289823,
+      "loss": 2.1027,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008498711506550983,
+      "loss": 2.2032,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008491301825382252,
+      "loss": 2.1601,
+      "step": 432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008483877151212076,
+      "loss": 2.1407,
+      "step": 433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008476437515924877,
+      "loss": 2.078,
+      "step": 434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008468982951469333,
+      "loss": 2.1124,
+      "step": 435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008461513489858227,
+      "loss": 2.0846,
+      "step": 436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008454029163168317,
+      "loss": 2.2554,
+      "step": 437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008446530003540203,
+      "loss": 2.1727,
+      "step": 438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008439016043178176,
+      "loss": 2.1117,
+      "step": 439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431487314350091,
+      "loss": 2.2175,
+      "step": 440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008423943849387222,
+      "loss": 2.1967,
+      "step": 441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416385680684131,
+      "loss": 2.1834,
+      "step": 442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008408812840698517,
+      "loss": 2.0768,
+      "step": 443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008401225361951089,
+      "loss": 2.3348,
+      "step": 444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008393623277025415,
+      "loss": 2.1338,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008386006618567792,
+      "loss": 2.2643,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008378375419287098,
+      "loss": 2.2881,
+      "step": 447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837072971195466,
+      "loss": 2.1771,
+      "step": 448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008363069529404102,
+      "loss": 2.1407,
+      "step": 449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355394904531211,
+      "loss": 2.3564,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083477058702938,
+      "loss": 2.2557,
+      "step": 451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000834000245971156,
+      "loss": 2.1336,
+      "step": 452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008332284705865914,
+      "loss": 2.1308,
+      "step": 453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008324552641899885,
+      "loss": 2.1951,
+      "step": 454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000831680630101795,
+      "loss": 2.2223,
+      "step": 455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008309045716485895,
+      "loss": 2.114,
+      "step": 456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008301270921630673,
+      "loss": 2.1581,
+      "step": 457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008293481949840261,
+      "loss": 2.1657,
+      "step": 458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008285678834563523,
+      "loss": 2.2689,
+      "step": 459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008277861609310056,
+      "loss": 2.1744,
+      "step": 460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008270030307650047,
+      "loss": 2.0757,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826218496321414,
+      "loss": 2.1619,
+      "step": 462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008254325609693279,
+      "loss": 2.1041,
+      "step": 463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008246452280838572,
+      "loss": 2.0972,
+      "step": 464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008238565010461139,
+      "loss": 2.2143,
+      "step": 465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008230663832431974,
+      "loss": 2.0791,
+      "step": 466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822274878068179,
+      "loss": 2.1503,
+      "step": 467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008214819889200888,
+      "loss": 2.0529,
+      "step": 468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008206877192038995,
+      "loss": 2.2145,
+      "step": 469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008198920723305126,
+      "loss": 2.1894,
+      "step": 470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008190950517167437,
+      "loss": 2.2042,
+      "step": 471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818296660785308,
+      "loss": 2.13,
+      "step": 472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008174969029648052,
+      "loss": 2.2678,
+      "step": 473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008166957816897049,
+      "loss": 2.1321,
+      "step": 474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008158933004003319,
+      "loss": 2.1269,
+      "step": 475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008150894625428515,
+      "loss": 2.1511,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008142842715692548,
+      "loss": 2.1518,
+      "step": 477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134777309373434,
+      "loss": 2.0268,
+      "step": 478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126698441107145,
+      "loss": 2.2607,
+      "step": 479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118606145587477,
+      "loss": 2.1925,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008110500457565872,
+      "loss": 2.1061,
+      "step": 481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008102381411851296,
+      "loss": 2.0693,
+      "step": 482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008094249043310073,
+      "loss": 2.0749,
+      "step": 483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008086103386865742,
+      "loss": 2.1105,
+      "step": 484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077944477498905,
+      "loss": 2.3,
+      "step": 485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069772350247073,
+      "loss": 2.2033,
+      "step": 486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008061587040204528,
+      "loss": 2.1764,
+      "step": 487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008053388582522159,
+      "loss": 2.1677,
+      "step": 488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008045177012407315,
+      "loss": 2.1205,
+      "step": 489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008036952365123659,
+      "loss": 2.2182,
+      "step": 490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008028714675991006,
+      "loss": 2.1146,
+      "step": 491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008020463980385189,
+      "loss": 2.1031,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008012200313737881,
+      "loss": 2.2019,
+      "step": 493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008003923711536468,
+      "loss": 2.1299,
+      "step": 494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007995634209323885,
+      "loss": 2.0436,
+      "step": 495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007987331842698462,
+      "loss": 2.1156,
+      "step": 496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007979016647313775,
+      "loss": 1.9952,
+      "step": 497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007970688658878491,
+      "loss": 2.2441,
+      "step": 498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007962347913156217,
+      "loss": 2.1284,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007953994445965342,
+      "loss": 2.0766,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007945628293178892,
+      "loss": 2.0884,
+      "step": 501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000793724949072436,
+      "loss": 2.0743,
+      "step": 502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007928858074583569,
+      "loss": 2.1974,
+      "step": 503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792045408079251,
+      "loss": 2.1545,
+      "step": 504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007912037545441181,
+      "loss": 2.1991,
+      "step": 505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903608504673446,
+      "loss": 2.1463,
+      "step": 506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007895166994686868,
+      "loss": 2.1355,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000788671305173256,
+      "loss": 2.1087,
+      "step": 508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878246712115023,
+      "loss": 2.0589,
+      "step": 509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007869768012191997,
+      "loss": 2.2506,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007861276988374302,
+      "loss": 2.0158,
+      "step": 511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000785277367712568,
+      "loss": 2.4129,
+      "step": 512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007844258114962641,
+      "loss": 2.1517,
+      "step": 513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835730338454308,
+      "loss": 2.0532,
+      "step": 514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007827190384222249,
+      "loss": 2.0328,
+      "step": 515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818638288940338,
+      "loss": 2.0241,
+      "step": 516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000781007408933458,
+      "loss": 2.089,
+      "step": 517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007801497822182963,
+      "loss": 2.1421,
+      "step": 518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007792909524315298,
+      "loss": 2.0429,
+      "step": 519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007784309232613063,
+      "loss": 2.0567,
+      "step": 520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007775696984009236,
+      "loss": 2.0844,
+      "step": 521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007767072815488146,
+      "loss": 1.9539,
+      "step": 522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007758436764085315,
+      "loss": 2.0838,
+      "step": 523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007749788866887286,
+      "loss": 2.1944,
+      "step": 524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000774112916103148,
+      "loss": 2.0697,
+      "step": 525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007732457683706025,
+      "loss": 2.2145,
+      "step": 526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00077237744721496,
+      "loss": 2.1201,
+      "step": 527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771507956365128,
+      "loss": 2.1739,
+      "step": 528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706372995550362,
+      "loss": 2.0991,
+      "step": 529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007697654805236224,
+      "loss": 2.1051,
+      "step": 530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007688925030148149,
+      "loss": 2.1091,
+      "step": 531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007680183707775171,
+      "loss": 2.1335,
+      "step": 532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007671430875655907,
+      "loss": 2.0519,
+      "step": 533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007662666571378414,
+      "loss": 2.2235,
+      "step": 534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007653890832580002,
+      "loss": 2.3042,
+      "step": 535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007645103696947089,
+      "loss": 2.0202,
+      "step": 536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007636305202215041,
+      "loss": 2.1496,
+      "step": 537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007627495386167998,
+      "loss": 2.066,
+      "step": 538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007618674286638719,
+      "loss": 2.1071,
+      "step": 539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007609841941508422,
+      "loss": 2.1633,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007600998388706615,
+      "loss": 1.9864,
+      "step": 541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007592143666210935,
+      "loss": 1.9886,
+      "step": 542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007583277812046993,
+      "loss": 2.228,
+      "step": 543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007574400864288194,
+      "loss": 2.1525,
+      "step": 544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007565512861055588,
+      "loss": 2.1214,
+      "step": 545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007556613840517703,
+      "loss": 2.0097,
+      "step": 546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007547703840890377,
+      "loss": 2.1701,
+      "step": 547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007538782900436596,
+      "loss": 2.153,
+      "step": 548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007529851057466335,
+      "loss": 2.0473,
+      "step": 549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007520908350336381,
+      "loss": 2.1777,
+      "step": 550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511954817450181,
+      "loss": 2.0794,
+      "step": 551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007502990497257672,
+      "loss": 2.1227,
+      "step": 552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007494015428255116,
+      "loss": 2.0203,
+      "step": 553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000748502964898493,
+      "loss": 2.2284,
+      "step": 554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007476033198035531,
+      "loss": 2.1305,
+      "step": 555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007467026114041162,
+      "loss": 2.1204,
+      "step": 556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007458008435681725,
+      "loss": 2.1656,
+      "step": 557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007448980201682625,
+      "loss": 2.0631,
+      "step": 558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000743994145081459,
+      "loss": 2.1667,
+      "step": 559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007430892221893517,
+      "loss": 2.2527,
+      "step": 560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007421832553780298,
+      "loss": 2.1574,
+      "step": 561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007412762485380654,
+      "loss": 2.2368,
+      "step": 562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740368205564497,
+      "loss": 1.9816,
+      "step": 563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007394591303568125,
+      "loss": 2.0842,
+      "step": 564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007385490268189329,
+      "loss": 2.1682,
+      "step": 565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000737637898859195,
+      "loss": 2.0874,
+      "step": 566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007367257503903349,
+      "loss": 2.1096,
+      "step": 567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007358125853294713,
+      "loss": 2.0769,
+      "step": 568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007348984075980881,
+      "loss": 2.1701,
+      "step": 569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007339832211220188,
+      "loss": 2.1171,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007330670298314281,
+      "loss": 2.0259,
+      "step": 571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321498376607962,
+      "loss": 2.3257,
+      "step": 572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000731231648548901,
+      "loss": 2.1601,
+      "step": 573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007303124664388023,
+      "loss": 2.144,
+      "step": 574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293922952778238,
+      "loss": 2.1255,
+      "step": 575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007284711390175365,
+      "loss": 2.1666,
+      "step": 576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007275490016137421,
+      "loss": 2.169,
+      "step": 577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007266258870264554,
+      "loss": 2.1561,
+      "step": 578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007257017992198879,
+      "loss": 2.1957,
+      "step": 579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007247767421624303,
+      "loss": 1.9553,
+      "step": 580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007238507198266355,
+      "loss": 2.2364,
+      "step": 581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007229237361892023,
+      "loss": 2.0767,
+      "step": 582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007219957952309567,
+      "loss": 2.0499,
+      "step": 583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007210669009368366,
+      "loss": 2.0317,
+      "step": 584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007201370572958737,
+      "loss": 2.2624,
+      "step": 585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007192062683011766,
+      "loss": 2.2565,
+      "step": 586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007182745379499133,
+      "loss": 2.0744,
+      "step": 587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007173418702432948,
+      "loss": 2.1936,
+      "step": 588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007164082691865572,
+      "loss": 2.1317,
+      "step": 589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007154737387889445,
+      "loss": 2.1379,
+      "step": 590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007145382830636924,
+      "loss": 2.1355,
+      "step": 591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007136019060280099,
+      "loss": 2.2058,
+      "step": 592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007126646117030618,
+      "loss": 2.1335,
+      "step": 593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007117264041139535,
+      "loss": 2.0544,
+      "step": 594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007107872872897112,
+      "loss": 2.045,
+      "step": 595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000709847265263266,
+      "loss": 2.1062,
+      "step": 596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007089063420714365,
+      "loss": 2.0644,
+      "step": 597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000707964521754911,
+      "loss": 2.1581,
+      "step": 598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007070218083582305,
+      "loss": 2.0723,
+      "step": 599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007060782059297712,
+      "loss": 2.191,
+      "step": 600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007051337185217274,
+      "loss": 2.0073,
+      "step": 601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007041883501900933,
+      "loss": 2.1849,
+      "step": 602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007032421049946467,
+      "loss": 2.1022,
+      "step": 603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007022949869989306,
+      "loss": 2.0704,
+      "step": 604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007013470002702363,
+      "loss": 2.1019,
+      "step": 605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007003981488795858,
+      "loss": 2.1711,
+      "step": 606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006994484369017142,
+      "loss": 1.9983,
+      "step": 607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006984978684150524,
+      "loss": 2.1143,
+      "step": 608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006975464475017094,
+      "loss": 2.102,
+      "step": 609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965941782474548,
+      "loss": 1.9986,
+      "step": 610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006956410647417017,
+      "loss": 2.1441,
+      "step": 611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000694687111077488,
+      "loss": 2.0354,
+      "step": 612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006937323213514602,
+      "loss": 2.0708,
+      "step": 613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006927766996638548,
+      "loss": 1.9991,
+      "step": 614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006918202501184815,
+      "loss": 2.1745,
+      "step": 615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006908629768227048,
+      "loss": 2.2681,
+      "step": 616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006899048838874267,
+      "loss": 2.2005,
+      "step": 617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006889459754270689,
+      "loss": 2.1742,
+      "step": 618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006879862555595559,
+      "loss": 2.003,
+      "step": 619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006870257284062961,
+      "loss": 1.914,
+      "step": 620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006860643980921648,
+      "loss": 2.0526,
+      "step": 621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006851022687454867,
+      "loss": 2.1416,
+      "step": 622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006841393444980176,
+      "loss": 2.0452,
+      "step": 623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006831756294849267,
+      "loss": 2.081,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006822111278447795,
+      "loss": 2.1476,
+      "step": 625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006812458437195197,
+      "loss": 2.138,
+      "step": 626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006802797812544502,
+      "loss": 2.066,
+      "step": 627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006793129445982178,
+      "loss": 2.0665,
+      "step": 628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006783453379027931,
+      "loss": 2.1124,
+      "step": 629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773769653234538,
+      "loss": 2.1271,
+      "step": 630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006764078310187667,
+      "loss": 2.1177,
+      "step": 631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006754379391505695,
+      "loss": 2.1263,
+      "step": 632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006744672938839533,
+      "loss": 2.1893,
+      "step": 633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006734958993872447,
+      "loss": 2.0714,
+      "step": 634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006725237598319877,
+      "loss": 2.1461,
+      "step": 635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006715508793929257,
+      "loss": 2.0332,
+      "step": 636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000670577262247984,
+      "loss": 2.2587,
+      "step": 637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006696029125782515,
+      "loss": 2.0497,
+      "step": 638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006686278345679625,
+      "loss": 2.1505,
+      "step": 639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006676520324044801,
+      "loss": 2.2021,
+      "step": 640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006666755102782757,
+      "loss": 2.0617,
+      "step": 641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006656982723829139,
+      "loss": 2.0975,
+      "step": 642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006647203229150322,
+      "loss": 2.0867,
+      "step": 643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006637416660743243,
+      "loss": 2.1205,
+      "step": 644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006627623060635214,
+      "loss": 2.1765,
+      "step": 645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006617822470883746,
+      "loss": 2.1082,
+      "step": 646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006608014933576361,
+      "loss": 2.1609,
+      "step": 647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006598200490830426,
+      "loss": 2.1197,
+      "step": 648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006588379184792954,
+      "loss": 2.1256,
+      "step": 649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006578551057640436,
+      "loss": 2.073,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006568716151578654,
+      "loss": 2.0063,
+      "step": 651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006558874508842498,
+      "loss": 2.0374,
+      "step": 652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006549026171695797,
+      "loss": 2.1831,
+      "step": 653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006539171182431121,
+      "loss": 2.1439,
+      "step": 654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006529309583369606,
+      "loss": 2.011,
+      "step": 655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006519441416860778,
+      "loss": 2.0212,
+      "step": 656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006509566725282362,
+      "loss": 2.1151,
+      "step": 657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006499685551040108,
+      "loss": 2.0407,
+      "step": 658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006489797936567603,
+      "loss": 2.2286,
+      "step": 659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000647990392432609,
+      "loss": 2.0817,
+      "step": 660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006470003556804286,
+      "loss": 2.0324,
+      "step": 661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006460096876518204,
+      "loss": 2.0501,
+      "step": 662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000645018392601096,
+      "loss": 2.0481,
+      "step": 663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006440264747852605,
+      "loss": 2.0808,
+      "step": 664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006430339384639927,
+      "loss": 2.0095,
+      "step": 665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006420407878996277,
+      "loss": 2.144,
+      "step": 666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006410470273571386,
+      "loss": 1.9558,
+      "step": 667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006400526611041177,
+      "loss": 2.1207,
+      "step": 668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006390576934107589,
+      "loss": 2.0454,
+      "step": 669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380621285498382,
+      "loss": 2.1451,
+      "step": 670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006370659707966967,
+      "loss": 2.0708,
+      "step": 671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006360692244292213,
+      "loss": 2.0124,
+      "step": 672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006350718937278268,
+      "loss": 2.0163,
+      "step": 673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006340739829754372,
+      "loss": 2.1376,
+      "step": 674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006330754964574676,
+      "loss": 2.0892,
+      "step": 675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006320764384618055,
+      "loss": 2.2149,
+      "step": 676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006310768132787928,
+      "loss": 1.946,
+      "step": 677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006300766252012069,
+      "loss": 2.0116,
+      "step": 678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006290758785242424,
+      "loss": 2.1479,
+      "step": 679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006280745775454933,
+      "loss": 2.2129,
+      "step": 680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000627072726564933,
+      "loss": 2.0555,
+      "step": 681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006260703298848979,
+      "loss": 2.177,
+      "step": 682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006250673918100671,
+      "loss": 1.9583,
+      "step": 683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006240639166474451,
+      "loss": 2.2028,
+      "step": 684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006230599087063425,
+      "loss": 2.1802,
+      "step": 685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006220553722983582,
+      "loss": 2.1482,
+      "step": 686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00062105031173736,
+      "loss": 2.0283,
+      "step": 687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006200447313394675,
+      "loss": 2.2278,
+      "step": 688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006190386354230319,
+      "loss": 2.0406,
+      "step": 689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006180320283086184,
+      "loss": 2.135,
+      "step": 690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006170249143189878,
+      "loss": 2.1924,
+      "step": 691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006160172977790772,
+      "loss": 2.0691,
+      "step": 692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006150091830159823,
+      "loss": 2.0878,
+      "step": 693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006140005743589379,
+      "loss": 2.2254,
+      "step": 694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006129914761393001,
+      "loss": 2.0358,
+      "step": 695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006119818926905269,
+      "loss": 2.1806,
+      "step": 696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000610971828348161,
+      "loss": 2.0931,
+      "step": 697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099612874498095,
+      "loss": 2.1897,
+      "step": 698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006089502743351258,
+      "loss": 2.2158,
+      "step": 699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000607938793345792,
+      "loss": 2.1099,
+      "step": 700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006069268488254986,
+      "loss": 2.1447,
+      "step": 701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006059144451199274,
+      "loss": 2.088,
+      "step": 702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006049015865767318,
+      "loss": 2.0913,
+      "step": 703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006038882775455182,
+      "loss": 1.9968,
+      "step": 704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006028745223778278,
+      "loss": 2.1489,
+      "step": 705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006018603254271178,
+      "loss": 2.1627,
+      "step": 706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006008456910487428,
+      "loss": 2.01,
+      "step": 707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000599830623599935,
+      "loss": 2.1357,
+      "step": 708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005988151274397873,
+      "loss": 1.9943,
+      "step": 709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005977992069292331,
+      "loss": 2.1386,
+      "step": 710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005967828664310283,
+      "loss": 2.085,
+      "step": 711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005957661103097327,
+      "loss": 1.9988,
+      "step": 712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005947489429316904,
+      "loss": 2.0894,
+      "step": 713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005937313686650117,
+      "loss": 2.1503,
+      "step": 714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005927133918795548,
+      "loss": 2.0145,
+      "step": 715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005916950169469057,
+      "loss": 2.1422,
+      "step": 716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005906762482403607,
+      "loss": 2.076,
+      "step": 717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000589657090134907,
+      "loss": 2.0052,
+      "step": 718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000588637547007204,
+      "loss": 2.1047,
+      "step": 719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005876176232355642,
+      "loss": 2.0958,
+      "step": 720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005865973231999356,
+      "loss": 2.0371,
+      "step": 721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005855766512818812,
+      "loss": 2.0751,
+      "step": 722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005845556118645611,
+      "loss": 2.1058,
+      "step": 723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005835342093327143,
+      "loss": 2.0289,
+      "step": 724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005825124480726382,
+      "loss": 2.0813,
+      "step": 725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005814903324721715,
+      "loss": 2.0131,
+      "step": 726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005804678669206738,
+      "loss": 2.1488,
+      "step": 727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005794450558090082,
+      "loss": 2.1403,
+      "step": 728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005784219035295214,
+      "loss": 2.0519,
+      "step": 729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005773984144760256,
+      "loss": 2.0766,
+      "step": 730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005763745930437787,
+      "loss": 2.1381,
+      "step": 731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005753504436294665,
+      "loss": 2.183,
+      "step": 732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005743259706311826,
+      "loss": 2.0202,
+      "step": 733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000573301178448411,
+      "loss": 1.9394,
+      "step": 734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005722760714820057,
+      "loss": 2.1314,
+      "step": 735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005712506541341731,
+      "loss": 2.0171,
+      "step": 736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005702249308084519,
+      "loss": 2.1841,
+      "step": 737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005691989059096954,
+      "loss": 2.1028,
+      "step": 738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005681725838440515,
+      "loss": 2.1786,
+      "step": 739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005671459690189443,
+      "loss": 1.9829,
+      "step": 740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005661190658430556,
+      "loss": 2.0996,
+      "step": 741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005650918787263048,
+      "loss": 2.1729,
+      "step": 742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005640644120798312,
+      "loss": 2.0628,
+      "step": 743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005630366703159739,
+      "loss": 2.0293,
+      "step": 744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005620086578482544,
+      "loss": 2.1123,
+      "step": 745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000560980379091356,
+      "loss": 2.0685,
+      "step": 746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005599518384611052,
+      "loss": 2.1878,
+      "step": 747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005589230403744543,
+      "loss": 2.1039,
+      "step": 748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005578939892494601,
+      "loss": 2.0256,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005568646895052666,
+      "loss": 2.1454,
+      "step": 750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005558351455620851,
+      "loss": 2.1523,
+      "step": 751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005548053618411763,
+      "loss": 2.1313,
+      "step": 752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005537753427648295,
+      "loss": 2.0054,
+      "step": 753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005527450927563457,
+      "loss": 2.0739,
+      "step": 754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000551714616240017,
+      "loss": 2.0825,
+      "step": 755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005506839176411088,
+      "loss": 2.1276,
+      "step": 756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005496530013858395,
+      "loss": 2.003,
+      "step": 757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005486218719013626,
+      "loss": 2.1952,
+      "step": 758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005475905336157473,
+      "loss": 2.0608,
+      "step": 759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005465589909579595,
+      "loss": 2.1367,
+      "step": 760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005455272483578424,
+      "loss": 2.1141,
+      "step": 761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005444953102460982,
+      "loss": 2.1984,
+      "step": 762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005434631810542687,
+      "loss": 2.0545,
+      "step": 763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005424308652147162,
+      "loss": 2.2969,
+      "step": 764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005413983671606043,
+      "loss": 2.0684,
+      "step": 765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005403656913258796,
+      "loss": 2.1263,
+      "step": 766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005393328421452513,
+      "loss": 2.0416,
+      "step": 767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005382998240541741,
+      "loss": 2.0437,
+      "step": 768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005372666414888274,
+      "loss": 2.0819,
+      "step": 769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005362332988860972,
+      "loss": 2.1703,
+      "step": 770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005351998006835562,
+      "loss": 2.0741,
+      "step": 771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005341661513194461,
+      "loss": 2.0513,
+      "step": 772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005331323552326573,
+      "loss": 2.076,
+      "step": 773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005320984168627102,
+      "loss": 2.1185,
+      "step": 774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005310643406497366,
+      "loss": 2.0478,
+      "step": 775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005300301310344599,
+      "loss": 2.1537,
+      "step": 776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005289957924581764,
+      "loss": 2.1505,
+      "step": 777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005279613293627366,
+      "loss": 2.0568,
+      "step": 778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005269267461905253,
+      "loss": 2.2114,
+      "step": 779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005258920473844429,
+      "loss": 2.1625,
+      "step": 780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005248572373878869,
+      "loss": 2.0115,
+      "step": 781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005238223206447319,
+      "loss": 2.1034,
+      "step": 782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005227873015993108,
+      "loss": 2.1129,
+      "step": 783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005217521846963964,
+      "loss": 2.0187,
+      "step": 784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005207169743811809,
+      "loss": 2.1779,
+      "step": 785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005196816750992583,
+      "loss": 2.0684,
+      "step": 786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005186462912966047,
+      "loss": 2.1183,
+      "step": 787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005176108274195586,
+      "loss": 2.1561,
+      "step": 788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005165752879148027,
+      "loss": 2.1467,
+      "step": 789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005155396772293447,
+      "loss": 1.9988,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005145039998104974,
+      "loss": 2.0457,
+      "step": 791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005134682601058609,
+      "loss": 2.0758,
+      "step": 792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005124324625633021,
+      "loss": 2.1233,
+      "step": 793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000511396611630937,
+      "loss": 2.064,
+      "step": 794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005103607117571098,
+      "loss": 2.1289,
+      "step": 795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005093247673903761,
+      "loss": 2.108,
+      "step": 796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005082887829794819,
+      "loss": 2.2605,
+      "step": 797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005072527629733451,
+      "loss": 2.2369,
+      "step": 798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005062167118210367,
+      "loss": 2.1236,
+      "step": 799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005051806339717613,
+      "loss": 1.9941,
+      "step": 800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005041445338748383,
+      "loss": 2.0634,
+      "step": 801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005031084159796827,
+      "loss": 2.124,
+      "step": 802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005020722847357858,
+      "loss": 1.9971,
+      "step": 803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005010361445926958,
+      "loss": 2.146,
+      "step": 804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 2.1,
+      "step": 805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004989638554073043,
+      "loss": 2.0237,
+      "step": 806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004979277152642144,
+      "loss": 2.0851,
+      "step": 807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004968915840203174,
+      "loss": 2.117,
+      "step": 808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004958554661251616,
+      "loss": 2.1807,
+      "step": 809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004948193660282387,
+      "loss": 2.0706,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004937832881789635,
+      "loss": 1.9671,
+      "step": 811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000492747237026655,
+      "loss": 2.2476,
+      "step": 812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004917112170205183,
+      "loss": 2.1059,
+      "step": 813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000490675232609624,
+      "loss": 1.9637,
+      "step": 814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004896392882428901,
+      "loss": 2.1806,
+      "step": 815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004886033883690631,
+      "loss": 2.231,
+      "step": 816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004875675374366979,
+      "loss": 2.0196,
+      "step": 817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048653173989413924,
+      "loss": 2.0711,
+      "step": 818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004854960001895027,
+      "loss": 2.178,
+      "step": 819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048446032277065556,
+      "loss": 2.1598,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004834247120851972,
+      "loss": 2.0319,
+      "step": 821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048238917258044153,
+      "loss": 1.9075,
+      "step": 822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004813537087033954,
+      "loss": 2.1124,
+      "step": 823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048031832490074165,
+      "loss": 2.0723,
+      "step": 824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004792830256188192,
+      "loss": 2.1403,
+      "step": 825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004782478153036038,
+      "loss": 2.0108,
+      "step": 826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004772126984006892,
+      "loss": 2.1216,
+      "step": 827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004761776793552682,
+      "loss": 2.0904,
+      "step": 828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004751427626121132,
+      "loss": 2.0011,
+      "step": 829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047410795261555716,
+      "loss": 2.118,
+      "step": 830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004730732538094749,
+      "loss": 2.0842,
+      "step": 831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004720386706372636,
+      "loss": 2.0677,
+      "step": 832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004710042075418236,
+      "loss": 2.0989,
+      "step": 833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046996986896554024,
+      "loss": 2.0864,
+      "step": 834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004689356593502635,
+      "loss": 2.0603,
+      "step": 835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004679015831372898,
+      "loss": 2.074,
+      "step": 836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004668676447673429,
+      "loss": 2.0325,
+      "step": 837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004658338486805541,
+      "loss": 2.0772,
+      "step": 838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046480019931644374,
+      "loss": 2.0146,
+      "step": 839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046376670111390283,
+      "loss": 2.0714,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046273335851117263,
+      "loss": 2.0015,
+      "step": 841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000461700175945826,
+      "loss": 2.0755,
+      "step": 842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046066715785474875,
+      "loss": 2.1154,
+      "step": 843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004596343086741207,
+      "loss": 2.1053,
+      "step": 844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004586016328393956,
+      "loss": 2.2043,
+      "step": 845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004575691347852839,
+      "loss": 2.052,
+      "step": 846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004565368189457313,
+      "loss": 2.0104,
+      "step": 847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045550468975390186,
+      "loss": 2.0294,
+      "step": 848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045447275164215773,
+      "loss": 2.172,
+      "step": 849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045344100904204073,
+      "loss": 2.0919,
+      "step": 850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045240946638425275,
+      "loss": 2.1253,
+      "step": 851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004513781280986375,
+      "loss": 2.097,
+      "step": 852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004503469986141606,
+      "loss": 1.9504,
+      "step": 853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004493160823588913,
+      "loss": 2.0434,
+      "step": 854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044828538375998303,
+      "loss": 2.0219,
+      "step": 855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004472549072436544,
+      "loss": 2.0524,
+      "step": 856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044622465723517053,
+      "loss": 1.9099,
+      "step": 857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004451946381588239,
+      "loss": 2.1645,
+      "step": 858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004441648544379149,
+      "loss": 2.0799,
+      "step": 859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044313531049473354,
+      "loss": 2.0362,
+      "step": 860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044210601075054016,
+      "loss": 2.1315,
+      "step": 861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044107695962554595,
+      "loss": 2.0403,
+      "step": 862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044004816153889474,
+      "loss": 2.0709,
+      "step": 863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004390196209086442,
+      "loss": 1.9408,
+      "step": 864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004379913421517457,
+      "loss": 2.1104,
+      "step": 865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004369633296840262,
+      "loss": 2.0904,
+      "step": 866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043593558792016907,
+      "loss": 2.1467,
+      "step": 867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043490812127369537,
+      "loss": 1.9398,
+      "step": 868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004338809341569444,
+      "loss": 2.0858,
+      "step": 869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043285403098105576,
+      "loss": 2.0431,
+      "step": 870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043182741615594865,
+      "loss": 2.1525,
+      "step": 871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043080109409030475,
+      "loss": 2.0767,
+      "step": 872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004297750691915482,
+      "loss": 2.0539,
+      "step": 873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042874934586582716,
+      "loss": 2.0088,
+      "step": 874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042772392851799434,
+      "loss": 2.0845,
+      "step": 875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004266988215515891,
+      "loss": 2.0948,
+      "step": 876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004256740293688175,
+      "loss": 1.9981,
+      "step": 877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004246495563705337,
+      "loss": 2.1624,
+      "step": 878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042362540695622134,
+      "loss": 2.0171,
+      "step": 879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004226015855239745,
+      "loss": 2.0464,
+      "step": 880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042157809647047856,
+      "loss": 1.9871,
+      "step": 881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004205549441909919,
+      "loss": 2.0953,
+      "step": 882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041953213307932633,
+      "loss": 2.2027,
+      "step": 883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004185096675278287,
+      "loss": 1.9868,
+      "step": 884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004174875519273619,
+      "loss": 2.2277,
+      "step": 885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004164657906672859,
+      "loss": 2.0673,
+      "step": 886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004154443881354388,
+      "loss": 2.0855,
+      "step": 887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041442334871811886,
+      "loss": 2.0787,
+      "step": 888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041340267680006456,
+      "loss": 2.0319,
+      "step": 889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041238237676443593,
+      "loss": 2.0753,
+      "step": 890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004113624529927963,
+      "loss": 2.1217,
+      "step": 891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004103429098650932,
+      "loss": 1.9521,
+      "step": 892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040932375175963936,
+      "loss": 1.9819,
+      "step": 893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004083049830530944,
+      "loss": 1.9791,
+      "step": 894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004072866081204453,
+      "loss": 2.1413,
+      "step": 895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040626863133498835,
+      "loss": 2.0922,
+      "step": 896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004052510570683098,
+      "loss": 2.1612,
+      "step": 897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040423388969026747,
+      "loss": 2.0848,
+      "step": 898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040321713356897173,
+      "loss": 2.2011,
+      "step": 899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000402200793070767,
+      "loss": 2.064,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040118487256021284,
+      "loss": 2.1055,
+      "step": 901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004001693764000651,
+      "loss": 2.0265,
+      "step": 902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039915430895125735,
+      "loss": 2.17,
+      "step": 903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003981396745728822,
+      "loss": 2.0551,
+      "step": 904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003971254776221722,
+      "loss": 2.157,
+      "step": 905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003961117224544819,
+      "loss": 2.1378,
+      "step": 906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003950984134232683,
+      "loss": 2.0609,
+      "step": 907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039408555488007257,
+      "loss": 1.9766,
+      "step": 908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003930731511745015,
+      "loss": 2.0318,
+      "step": 909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039206120665420826,
+      "loss": 2.2185,
+      "step": 910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003910497256648742,
+      "loss": 2.2372,
+      "step": 911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039003871255019056,
+      "loss": 1.9921,
+      "step": 912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038902817165183904,
+      "loss": 2.1184,
+      "step": 913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003880181073094731,
+      "loss": 2.0634,
+      "step": 914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003870085238607002,
+      "loss": 2.0882,
+      "step": 915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003859994256410623,
+      "loss": 2.0986,
+      "step": 916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038499081698401786,
+      "loss": 1.994,
+      "step": 917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003839827022209228,
+      "loss": 2.0288,
+      "step": 918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003829750856810123,
+      "loss": 2.0549,
+      "step": 919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038196797169138164,
+      "loss": 2.0223,
+      "step": 920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038096136457696825,
+      "loss": 2.043,
+      "step": 921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003799552686605326,
+      "loss": 2.1654,
+      "step": 922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003789496882626399,
+      "loss": 1.9595,
+      "step": 923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037794462770164187,
+      "loss": 2.1133,
+      "step": 924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003769400912936575,
+      "loss": 2.109,
+      "step": 925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037593608335255495,
+      "loss": 1.9783,
+      "step": 926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003749326081899329,
+      "loss": 2.1163,
+      "step": 927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037392967011510225,
+      "loss": 2.039,
+      "step": 928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037292727343506707,
+      "loss": 2.1645,
+      "step": 929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037192542245450687,
+      "loss": 2.1381,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003709241214757576,
+      "loss": 1.9833,
+      "step": 931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003699233747987932,
+      "loss": 2.0596,
+      "step": 932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003689231867212074,
+      "loss": 2.1023,
+      "step": 933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036792356153819467,
+      "loss": 2.2186,
+      "step": 934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003669245035425324,
+      "loss": 2.0252,
+      "step": 935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003659260170245628,
+      "loss": 2.0106,
+      "step": 936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036492810627217334,
+      "loss": 2.0541,
+      "step": 937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003639307755707788,
+      "loss": 2.0236,
+      "step": 938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036293402920330344,
+      "loss": 2.0874,
+      "step": 939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003619378714501619,
+      "loss": 2.1146,
+      "step": 940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003609423065892412,
+      "loss": 2.0,
+      "step": 941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003599473388958823,
+      "loss": 2.0712,
+      "step": 942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003589529726428615,
+      "loss": 2.0856,
+      "step": 943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003579592121003724,
+      "loss": 2.034,
+      "step": 944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035696606153600753,
+      "loss": 2.0107,
+      "step": 945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003559735252147397,
+      "loss": 2.0399,
+      "step": 946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000354981607398904,
+      "loss": 1.9285,
+      "step": 947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003539903123481798,
+      "loss": 1.9658,
+      "step": 948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035299964431957144,
+      "loss": 2.0946,
+      "step": 949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035200960756739106,
+      "loss": 1.9822,
+      "step": 950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003510202063432397,
+      "loss": 1.8863,
+      "step": 951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003500314448959892,
+      "loss": 2.0551,
+      "step": 952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003490433274717637,
+      "loss": 1.9943,
+      "step": 953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034805585831392227,
+      "loss": 2.0048,
+      "step": 954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003470690416630395,
+      "loss": 2.0481,
+      "step": 955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034608288175688804,
+      "loss": 2.0627,
+      "step": 956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034509738283042046,
+      "loss": 2.0432,
+      "step": 957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003441125491157503,
+      "loss": 2.0715,
+      "step": 958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003431283848421347,
+      "loss": 2.1223,
+      "step": 959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003421448942359564,
+      "loss": 2.0992,
+      "step": 960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034116208152070473,
+      "loss": 1.9942,
+      "step": 961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003401799509169575,
+      "loss": 2.024,
+      "step": 962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033919850664236394,
+      "loss": 2.0626,
+      "step": 963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003382177529116257,
+      "loss": 2.1118,
+      "step": 964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003372376939364787,
+      "loss": 2.0552,
+      "step": 965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033625833392567577,
+      "loss": 1.9632,
+      "step": 966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033527967708496786,
+      "loss": 2.0454,
+      "step": 967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003343017276170862,
+      "loss": 2.1536,
+      "step": 968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033332448972172435,
+      "loss": 2.2057,
+      "step": 969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003323479675955202,
+      "loss": 2.0075,
+      "step": 970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033137216543203745,
+      "loss": 2.0614,
+      "step": 971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003303970874217487,
+      "loss": 2.1614,
+      "step": 972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003294227377520161,
+      "loss": 2.1203,
+      "step": 973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003284491206070744,
+      "loss": 2.0786,
+      "step": 974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003274762401680124,
+      "loss": 2.0887,
+      "step": 975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032650410061275535,
+      "loss": 2.0989,
+      "step": 976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032553270611604667,
+      "loss": 2.1184,
+      "step": 977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032456206084943056,
+      "loss": 2.1357,
+      "step": 978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003235921689812334,
+      "loss": 2.099,
+      "step": 979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032262303467654627,
+      "loss": 2.0439,
+      "step": 980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003216546620972071,
+      "loss": 2.2389,
+      "step": 981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003206870554017824,
+      "loss": 2.0241,
+      "step": 982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003197202187455498,
+      "loss": 1.9409,
+      "step": 983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031875415628048045,
+      "loss": 2.018,
+      "step": 984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031778887215522044,
+      "loss": 2.0431,
+      "step": 985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003168243705150733,
+      "loss": 2.0308,
+      "step": 986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031586065550198256,
+      "loss": 2.0323,
+      "step": 987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003148977312545134,
+      "loss": 2.0391,
+      "step": 988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003139356019078352,
+      "loss": 2.0787,
+      "step": 989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000312974271593704,
+      "loss": 2.0737,
+      "step": 990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031201374444044417,
+      "loss": 1.9624,
+      "step": 991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003110540245729312,
+      "loss": 2.1494,
+      "step": 992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003100951161125735,
+      "loss": 1.9804,
+      "step": 993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003091370231772954,
+      "loss": 2.1109,
+      "step": 994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003081797498815185,
+      "loss": 2.0743,
+      "step": 995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003072233003361452,
+      "loss": 2.1092,
+      "step": 996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030626767864853986,
+      "loss": 2.1497,
+      "step": 997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030531288892251206,
+      "loss": 2.0427,
+      "step": 998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030435893525829845,
+      "loss": 2.0913,
+      "step": 999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030340582175254516,
+      "loss": 1.9625,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030245355249829055,
+      "loss": 2.0429,
+      "step": 1001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003015021315849476,
+      "loss": 2.0236,
+      "step": 1002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003005515630982858,
+      "loss": 2.061,
+      "step": 1003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029960185112041425,
+      "loss": 2.0687,
+      "step": 1004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002986529997297638,
+      "loss": 2.1061,
+      "step": 1005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029770501300106956,
+      "loss": 2.0088,
+      "step": 1006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002967578950053533,
+      "loss": 1.9895,
+      "step": 1007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002958116498099066,
+      "loss": 2.1139,
+      "step": 1008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029486628147827273,
+      "loss": 1.9323,
+      "step": 1009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002939217940702288,
+      "loss": 2.0678,
+      "step": 1010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002929781916417696,
+      "loss": 2.0156,
+      "step": 1011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029203547824508915,
+      "loss": 2.0424,
+      "step": 1012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029109365792856354,
+      "loss": 2.0264,
+      "step": 1013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000290152734736734,
+      "loss": 2.2246,
+      "step": 1014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892127127102889,
+      "loss": 2.208,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002882735958860465,
+      "loss": 2.085,
+      "step": 1016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028733538829693817,
+      "loss": 2.056,
+      "step": 1017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028639809397199045,
+      "loss": 1.9826,
+      "step": 1018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028546171693630745,
+      "loss": 2.0473,
+      "step": 1019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028452626121105545,
+      "loss": 2.1385,
+      "step": 1020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028359173081344305,
+      "loss": 2.0357,
+      "step": 1021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028265812975670525,
+      "loss": 2.0814,
+      "step": 1022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002817254620500868,
+      "loss": 2.045,
+      "step": 1023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002807937316988235,
+      "loss": 2.0983,
+      "step": 1024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002798629427041262,
+      "loss": 2.0121,
+      "step": 1025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002789330990631635,
+      "loss": 2.0904,
+      "step": 1026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027800420476904335,
+      "loss": 2.0913,
+      "step": 1027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000277076263810798,
+      "loss": 2.1313,
+      "step": 1028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002761492801733645,
+      "loss": 2.0072,
+      "step": 1029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027522325783756993,
+      "loss": 2.1575,
+      "step": 1030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002742982007801121,
+      "loss": 2.2156,
+      "step": 1031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027337411297354453,
+      "loss": 2.0812,
+      "step": 1032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027245099838625803,
+      "loss": 2.0107,
+      "step": 1033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002715288609824635,
+      "loss": 2.1334,
+      "step": 1034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002706077047221763,
+      "loss": 1.8868,
+      "step": 1035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002696875335611977,
+      "loss": 2.0042,
+      "step": 1036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002687683514510989,
+      "loss": 1.9757,
+      "step": 1037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026785016233920396,
+      "loss": 2.0251,
+      "step": 1038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002669329701685719,
+      "loss": 2.0353,
+      "step": 1039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026601677887798126,
+      "loss": 2.1606,
+      "step": 1040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000265101592401912,
+      "loss": 2.1006,
+      "step": 1041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026418741467052885,
+      "loss": 2.0603,
+      "step": 1042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026327424960966505,
+      "loss": 2.153,
+      "step": 1043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002623621011408049,
+      "loss": 2.1767,
+      "step": 1044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000261450973181067,
+      "loss": 1.99,
+      "step": 1045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026054086964318756,
+      "loss": 2.0899,
+      "step": 1046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025963179443550297,
+      "loss": 2.1129,
+      "step": 1047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002587237514619347,
+      "loss": 2.1879,
+      "step": 1048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025781674462197027,
+      "loss": 2.099,
+      "step": 1049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002569107778106483,
+      "loss": 2.1399,
+      "step": 1050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002560058549185412,
+      "loss": 2.0716,
+      "step": 1051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025510197983173765,
+      "loss": 2.016,
+      "step": 1052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025419915643182765,
+      "loss": 2.0325,
+      "step": 1053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002532973885958839,
+      "loss": 2.0582,
+      "step": 1054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002523966801964468,
+      "loss": 2.0752,
+      "step": 1055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025149703510150704,
+      "loss": 2.1091,
+      "step": 1056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002505984571744884,
+      "loss": 2.1413,
+      "step": 1057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024970095027423284,
+      "loss": 1.8833,
+      "step": 1058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002488045182549819,
+      "loss": 1.971,
+      "step": 1059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002479091649663621,
+      "loss": 2.0568,
+      "step": 1060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024701489425336666,
+      "loss": 1.9776,
+      "step": 1061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002461217099563403,
+      "loss": 2.2035,
+      "step": 1062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024522961591096244,
+      "loss": 2.0639,
+      "step": 1063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002443386159482297,
+      "loss": 1.9484,
+      "step": 1064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002434487138944413,
+      "loss": 2.1296,
+      "step": 1065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024255991357118086,
+      "loss": 2.0624,
+      "step": 1066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002416722187953006,
+      "loss": 2.1542,
+      "step": 1067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024078563337890635,
+      "loss": 2.1578,
+      "step": 1068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023990016112933866,
+      "loss": 2.1219,
+      "step": 1069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023901580584915783,
+      "loss": 2.0283,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023813257133612826,
+      "loss": 2.0599,
+      "step": 1071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023725046138320033,
+      "loss": 2.0152,
+      "step": 1072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023636947977849593,
+      "loss": 2.0702,
+      "step": 1073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002354896303052912,
+      "loss": 2.0128,
+      "step": 1074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023461091674199997,
+      "loss": 2.1287,
+      "step": 1075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002337333428621588,
+      "loss": 2.0941,
+      "step": 1076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023285691243440926,
+      "loss": 2.1369,
+      "step": 1077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023198162922248323,
+      "loss": 1.9025,
+      "step": 1078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002311074969851852,
+      "loss": 2.0787,
+      "step": 1079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023023451947637758,
+      "loss": 2.0229,
+      "step": 1080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022936270044496388,
+      "loss": 2.0799,
+      "step": 1081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022849204363487214,
+      "loss": 2.016,
+      "step": 1082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022762255278504007,
+      "loss": 1.9756,
+      "step": 1083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022675423162939757,
+      "loss": 1.9795,
+      "step": 1084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022588708389685192,
+      "loss": 2.1467,
+      "step": 1085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002250211133112714,
+      "loss": 1.9688,
+      "step": 1086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022415632359146855,
+      "loss": 2.1594,
+      "step": 1087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022329271845118541,
+      "loss": 2.0094,
+      "step": 1088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022243030159907668,
+      "loss": 2.078,
+      "step": 1089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022156907673869392,
+      "loss": 2.0417,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022070904756847022,
+      "loss": 2.0016,
+      "step": 1091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021985021778170373,
+      "loss": 1.8752,
+      "step": 1092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002189925910665421,
+      "loss": 2.1363,
+      "step": 1093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021813617110596646,
+      "loss": 2.0353,
+      "step": 1094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021728096157777515,
+      "loss": 2.0782,
+      "step": 1095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021642696615456952,
+      "loss": 1.9352,
+      "step": 1096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000215574188503736,
+      "loss": 2.0556,
+      "step": 1097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021472263228743206,
+      "loss": 2.051,
+      "step": 1098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021387230116257,
+      "loss": 1.9908,
+      "step": 1099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021302319878080034,
+      "loss": 1.9891,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021217532878849789,
+      "loss": 1.9989,
+      "step": 1101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021132869482674404,
+      "loss": 2.0781,
+      "step": 1102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002104833005313131,
+      "loss": 2.0379,
+      "step": 1103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020963914953265545,
+      "loss": 2.0529,
+      "step": 1104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020879624545588188,
+      "loss": 2.0158,
+      "step": 1105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020795459192074923,
+      "loss": 2.1177,
+      "step": 1106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002071141925416431,
+      "loss": 1.9829,
+      "step": 1107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020627505092756415,
+      "loss": 2.002,
+      "step": 1108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020543717068211093,
+      "loss": 2.0185,
+      "step": 1109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002046005554034656,
+      "loss": 2.0862,
+      "step": 1110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020376520868437836,
+      "loss": 2.1291,
+      "step": 1111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020293113411215085,
+      "loss": 2.078,
+      "step": 1112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020209833526862265,
+      "loss": 2.0553,
+      "step": 1113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000201266815730154,
+      "loss": 2.0603,
+      "step": 1114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020043657906761142,
+      "loss": 1.9504,
+      "step": 1115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001996076288463532,
+      "loss": 1.9397,
+      "step": 1116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019877996862621206,
+      "loss": 2.1393,
+      "step": 1117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019795360196148127,
+      "loss": 2.1321,
+      "step": 1118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001971285324008994,
+      "loss": 2.1099,
+      "step": 1119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001963047634876342,
+      "loss": 2.1581,
+      "step": 1120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001954822987592685,
+      "loss": 2.0484,
+      "step": 1121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019466114174778422,
+      "loss": 1.9932,
+      "step": 1122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001938412959795472,
+      "loss": 2.0662,
+      "step": 1123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001930227649752928,
+      "loss": 2.0545,
+      "step": 1124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019220555225010966,
+      "loss": 2.0381,
+      "step": 1125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001913896613134259,
+      "loss": 2.0394,
+      "step": 1126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019057509566899268,
+      "loss": 2.1844,
+      "step": 1127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018976185881487028,
+      "loss": 2.0882,
+      "step": 1128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018894995424341283,
+      "loss": 2.1727,
+      "step": 1129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018813938544125236,
+      "loss": 2.0092,
+      "step": 1130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018733015588928549,
+      "loss": 1.9774,
+      "step": 1131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018652226906265695,
+      "loss": 1.9787,
+      "step": 1132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001857157284307452,
+      "loss": 2.109,
+      "step": 1133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018491053745714846,
+      "loss": 2.0125,
+      "step": 1134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018410669959966803,
+      "loss": 2.0888,
+      "step": 1135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018330421831029515,
+      "loss": 1.9985,
+      "step": 1136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018250309703519497,
+      "loss": 2.1737,
+      "step": 1137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018170333921469205,
+      "loss": 2.0432,
+      "step": 1138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018090494828325632,
+      "loss": 1.8715,
+      "step": 1139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001801079276694874,
+      "loss": 2.0333,
+      "step": 1140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017931228079610056,
+      "loss": 1.9958,
+      "step": 1141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017851801107991128,
+      "loss": 2.1071,
+      "step": 1142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017772512193182095,
+      "loss": 2.1363,
+      "step": 1143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001769336167568028,
+      "loss": 2.0867,
+      "step": 1144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017614349895388615,
+      "loss": 2.0741,
+      "step": 1145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001753547719161428,
+      "loss": 2.0356,
+      "step": 1146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017456743903067218,
+      "loss": 2.0378,
+      "step": 1147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017378150367858608,
+      "loss": 2.0892,
+      "step": 1148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017299696923499543,
+      "loss": 2.144,
+      "step": 1149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017221383906899451,
+      "loss": 2.0567,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001714321165436476,
+      "loss": 2.0739,
+      "step": 1151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017065180501597387,
+      "loss": 1.9071,
+      "step": 1152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016987290783693283,
+      "loss": 1.9898,
+      "step": 1153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016909542835141073,
+      "loss": 2.1216,
+      "step": 1154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016831936989820506,
+      "loss": 2.147,
+      "step": 1155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016754473581001168,
+      "loss": 2.0879,
+      "step": 1156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016677152941340872,
+      "loss": 1.9584,
+      "step": 1157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016599975402884404,
+      "loss": 2.1087,
+      "step": 1158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016522941297061994,
+      "loss": 2.0596,
+      "step": 1159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016446050954687885,
+      "loss": 2.1234,
+      "step": 1160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016369304705959,
+      "loss": 2.0916,
+      "step": 1161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001629270288045342,
+      "loss": 2.1231,
+      "step": 1162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016216245807129003,
+      "loss": 1.9881,
+      "step": 1163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016139933814322084,
+      "loss": 2.0208,
+      "step": 1164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016063767229745858,
+      "loss": 2.0831,
+      "step": 1165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015987746380489114,
+      "loss": 1.9929,
+      "step": 1166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015911871593014837,
+      "loss": 2.1134,
+      "step": 1167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015836143193158697,
+      "loss": 2.0022,
+      "step": 1168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001576056150612778,
+      "loss": 2.0163,
+      "step": 1169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001568512685649911,
+      "loss": 2.0716,
+      "step": 1170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001560983956821825,
+      "loss": 2.1022,
+      "step": 1171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015534699964597982,
+      "loss": 2.0575,
+      "step": 1172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015459708368316822,
+      "loss": 2.1312,
+      "step": 1173
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015384865101417743,
+      "loss": 2.0264,
+      "step": 1174
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015310170485306672,
+      "loss": 2.117,
+      "step": 1175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015235624840751223,
+      "loss": 2.2048,
+      "step": 1176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015161228487879254,
+      "loss": 2.0683,
+      "step": 1177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015086981746177474,
+      "loss": 2.1411,
+      "step": 1178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015012884934490167,
+      "loss": 2.1484,
+      "step": 1179
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014938938371017706,
+      "loss": 2.0971,
+      "step": 1180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014865142373315227,
+      "loss": 2.1303,
+      "step": 1181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001479149725829137,
+      "loss": 2.0389,
+      "step": 1182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001471800334220672,
+      "loss": 2.0711,
+      "step": 1183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014644660940672628,
+      "loss": 2.0522,
+      "step": 1184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001457147036864974,
+      "loss": 2.0749,
+      "step": 1185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014498431940446693,
+      "loss": 2.1095,
+      "step": 1186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000144255459697188,
+      "loss": 2.0136,
+      "step": 1187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014352812769466577,
+      "loss": 2.0678,
+      "step": 1188
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014280232652034609,
+      "loss": 2.0472,
+      "step": 1189
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001420780592910998,
+      "loss": 2.0866,
+      "step": 1190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001413553291172106,
+      "loss": 1.975,
+      "step": 1191
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001406341391023619,
+      "loss": 1.996,
+      "step": 1192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013991449234362246,
+      "loss": 1.978,
+      "step": 1193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013919639193143418,
+      "loss": 2.1107,
+      "step": 1194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013847984094959842,
+      "loss": 2.0115,
+      "step": 1195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013776484247526204,
+      "loss": 1.9402,
+      "step": 1196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013705139957890562,
+      "loss": 2.0726,
+      "step": 1197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001363395153243288,
+      "loss": 2.0066,
+      "step": 1198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013562919276863844,
+      "loss": 2.0812,
+      "step": 1199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013492043496223416,
+      "loss": 1.9382,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013421324494879662,
+      "loss": 2.0392,
+      "step": 1201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013350762576527347,
+      "loss": 2.0958,
+      "step": 1202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013280358044186647,
+      "loss": 2.0135,
+      "step": 1203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001321011120020189,
+      "loss": 1.9997,
+      "step": 1204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013140022346240217,
+      "loss": 2.1455,
+      "step": 1205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001307009178329025,
+      "loss": 2.0862,
+      "step": 1206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013000319811660966,
+      "loss": 2.024,
+      "step": 1207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012930706730980186,
+      "loss": 1.9954,
+      "step": 1208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012861252840193406,
+      "loss": 2.0401,
+      "step": 1209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012791958437562552,
+      "loss": 2.0393,
+      "step": 1210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012722823820664574,
+      "loss": 2.0206,
+      "step": 1211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000126538492863903,
+      "loss": 2.0376,
+      "step": 1212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012585035130943095,
+      "loss": 2.1911,
+      "step": 1213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012516381649837544,
+      "loss": 2.0106,
+      "step": 1214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012447889137898294,
+      "loss": 2.0499,
+      "step": 1215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012379557889258686,
+      "loss": 2.0506,
+      "step": 1216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001231138819735957,
+      "loss": 2.1426,
+      "step": 1217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012243380354947964,
+      "loss": 2.1219,
+      "step": 1218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012175534654075887,
+      "loss": 1.954,
+      "step": 1219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012107851386099044,
+      "loss": 2.1988,
+      "step": 1220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012040330841675574,
+      "loss": 1.9507,
+      "step": 1221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011972973310764846,
+      "loss": 1.9815,
+      "step": 1222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011905779082626139,
+      "loss": 1.9717,
+      "step": 1223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011838748445817505,
+      "loss": 1.9872,
+      "step": 1224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011771881688194452,
+      "loss": 2.0511,
+      "step": 1225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011705179096908691,
+      "loss": 2.0517,
+      "step": 1226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011638640958406998,
+      "loss": 2.0622,
+      "step": 1227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011572267558429878,
+      "loss": 2.1418,
+      "step": 1228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011506059182010392,
+      "loss": 2.0083,
+      "step": 1229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001144001611347295,
+      "loss": 2.0656,
+      "step": 1230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011374138636432052,
+      "loss": 2.0848,
+      "step": 1231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011308427033791097,
+      "loss": 1.8887,
+      "step": 1232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011242881587741127,
+      "loss": 2.0152,
+      "step": 1233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011177502579759646,
+      "loss": 2.01,
+      "step": 1234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011112290290609445,
+      "loss": 2.0655,
+      "step": 1235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011047245000337302,
+      "loss": 1.9779,
+      "step": 1236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010982366988272924,
+      "loss": 2.021,
+      "step": 1237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010917656533027576,
+      "loss": 2.0764,
+      "step": 1238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001085311391249299,
+      "loss": 2.0037,
+      "step": 1239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010788739403840192,
+      "loss": 2.0077,
+      "step": 1240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010724533283518206,
+      "loss": 2.0069,
+      "step": 1241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010660495827252986,
+      "loss": 2.0557,
+      "step": 1242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010596627310046165,
+      "loss": 2.0399,
+      "step": 1243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010532928006173852,
+      "loss": 2.0151,
+      "step": 1244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010469398189185542,
+      "loss": 2.1227,
+      "step": 1245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010406038131902834,
+      "loss": 1.9816,
+      "step": 1246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010342848106418367,
+      "loss": 1.9101,
+      "step": 1247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010279828384094542,
+      "loss": 2.1743,
+      "step": 1248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010216979235562452,
+      "loss": 2.163,
+      "step": 1249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010154300930720683,
+      "loss": 1.9514,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010091793738734111,
+      "loss": 1.9639,
+      "step": 1251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010029457928032843,
+      "loss": 2.1185,
+      "step": 1252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.967293766310958e-05,
+      "loss": 2.0083,
+      "step": 1253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.905301520525412e-05,
+      "loss": 2.0035,
+      "step": 1254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.843481456894948e-05,
+      "loss": 2.0419,
+      "step": 1255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78183384089883e-05,
+      "loss": 1.9924,
+      "step": 1256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.720358937275764e-05,
+      "loss": 1.9788,
+      "step": 1257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.659057010022809e-05,
+      "loss": 2.0713,
+      "step": 1258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.59792832239415e-05,
+      "loss": 2.1188,
+      "step": 1259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.536973136900034e-05,
+      "loss": 2.1621,
+      "step": 1260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.476191715305649e-05,
+      "loss": 2.0539,
+      "step": 1261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.415584318629911e-05,
+      "loss": 2.049,
+      "step": 1262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.35515120714447e-05,
+      "loss": 2.0543,
+      "step": 1263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.29489264037247e-05,
+      "loss": 2.0494,
+      "step": 1264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.234808877087553e-05,
+      "loss": 2.019,
+      "step": 1265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.174900175312628e-05,
+      "loss": 2.1056,
+      "step": 1266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.115166792318858e-05,
+      "loss": 2.0569,
+      "step": 1267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.05560898462453e-05,
+      "loss": 2.1397,
+      "step": 1268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.996227007993895e-05,
+      "loss": 2.1158,
+      "step": 1269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.937021117436178e-05,
+      "loss": 1.91,
+      "step": 1270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.877991567204352e-05,
+      "loss": 1.9289,
+      "step": 1271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.819138610794176e-05,
+      "loss": 1.9655,
+      "step": 1272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.76046250094304e-05,
+      "loss": 2.0962,
+      "step": 1273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.701963489628839e-05,
+      "loss": 2.0683,
+      "step": 1274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.643641828069004e-05,
+      "loss": 1.9939,
+      "step": 1275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.585497766719313e-05,
+      "loss": 1.9661,
+      "step": 1276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.527531555272849e-05,
+      "loss": 2.1734,
+      "step": 1277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.469743442658984e-05,
+      "loss": 2.0604,
+      "step": 1278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.412133677042238e-05,
+      "loss": 2.0125,
+      "step": 1279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.35470250582126e-05,
+      "loss": 2.1259,
+      "step": 1280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.297450175627713e-05,
+      "loss": 2.0728,
+      "step": 1281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.24037693232525e-05,
+      "loss": 2.0009,
+      "step": 1282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.183483021008498e-05,
+      "loss": 1.9845,
+      "step": 1283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.126768686001912e-05,
+      "loss": 2.0634,
+      "step": 1284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.070234170858803e-05,
+      "loss": 2.117,
+      "step": 1285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.013879718360285e-05,
+      "loss": 2.1212,
+      "step": 1286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.957705570514162e-05,
+      "loss": 1.9513,
+      "step": 1287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.901711968553987e-05,
+      "loss": 2.1499,
+      "step": 1288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.845899152937946e-05,
+      "loss": 2.1485,
+      "step": 1289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.79026736334788e-05,
+      "loss": 1.9962,
+      "step": 1290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.734816838688247e-05,
+      "loss": 2.0017,
+      "step": 1291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.679547817085026e-05,
+      "loss": 2.0559,
+      "step": 1292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.62446053588482e-05,
+      "loss": 1.9789,
+      "step": 1293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.569555231653719e-05,
+      "loss": 2.0361,
+      "step": 1294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.51483214017637e-05,
+      "loss": 2.0423,
+      "step": 1295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.460291496454886e-05,
+      "loss": 1.9793,
+      "step": 1296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.405933534707914e-05,
+      "loss": 2.0369,
+      "step": 1297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.351758488369598e-05,
+      "loss": 1.9934,
+      "step": 1298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297766590088528e-05,
+      "loss": 1.9649,
+      "step": 1299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.243958071726841e-05,
+      "loss": 1.9614,
+      "step": 1300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.190333164359137e-05,
+      "loss": 1.9227,
+      "step": 1301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.136892098271491e-05,
+      "loss": 2.0098,
+      "step": 1302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.083635102960584e-05,
+      "loss": 2.1107,
+      "step": 1303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.030562407132535e-05,
+      "loss": 1.9936,
+      "step": 1304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.977674238702037e-05,
+      "loss": 2.1293,
+      "step": 1305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.924970824791382e-05,
+      "loss": 1.8941,
+      "step": 1306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8724523917294e-05,
+      "loss": 2.0207,
+      "step": 1307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.820119165050592e-05,
+      "loss": 2.0245,
+      "step": 1308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.767971369494097e-05,
+      "loss": 2.1325,
+      "step": 1309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.716009229002706e-05,
+      "loss": 2.0091,
+      "step": 1310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.664232966721994e-05,
+      "loss": 2.0626,
+      "step": 1311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.612642804999253e-05,
+      "loss": 2.0543,
+      "step": 1312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.561238965382621e-05,
+      "loss": 2.0264,
+      "step": 1313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.510021668620058e-05,
+      "loss": 1.9749,
+      "step": 1314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.458991134658487e-05,
+      "loss": 2.0976,
+      "step": 1315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.408147582642771e-05,
+      "loss": 1.9672,
+      "step": 1316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.357491230914786e-05,
+      "loss": 2.1487,
+      "step": 1317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.307022297012539e-05,
+      "loss": 2.1202,
+      "step": 1318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.256740997669142e-05,
+      "loss": 1.99,
+      "step": 1319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.20664754881196e-05,
+      "loss": 2.1795,
+      "step": 1320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.156742165561685e-05,
+      "loss": 1.9783,
+      "step": 1321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.107025062231309e-05,
+      "loss": 2.0304,
+      "step": 1322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.057496452325345e-05,
+      "loss": 1.9992,
+      "step": 1323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0081565485388065e-05,
+      "loss": 2.1068,
+      "step": 1324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9590055627563256e-05,
+      "loss": 2.0077,
+      "step": 1325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.910043706051277e-05,
+      "loss": 1.9689,
+      "step": 1326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.861271188684819e-05,
+      "loss": 2.0364,
+      "step": 1327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8126882201050336e-05,
+      "loss": 2.106,
+      "step": 1328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.76429500894598e-05,
+      "loss": 2.0746,
+      "step": 1329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7160917630268315e-05,
+      "loss": 2.0777,
+      "step": 1330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.668078689351008e-05,
+      "loss": 2.0527,
+      "step": 1331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6202559941052e-05,
+      "loss": 2.0564,
+      "step": 1332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.572623882658595e-05,
+      "loss": 1.9755,
+      "step": 1333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.525182559561926e-05,
+      "loss": 2.1418,
+      "step": 1334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.477932228546573e-05,
+      "loss": 2.002,
+      "step": 1335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4308730925237635e-05,
+      "loss": 2.0443,
+      "step": 1336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.384005353583632e-05,
+      "loss": 2.0262,
+      "step": 1337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.337329212994396e-05,
+      "loss": 2.0728,
+      "step": 1338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.290844871201483e-05,
+      "loss": 2.0627,
+      "step": 1339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.244552527826624e-05,
+      "loss": 2.0657,
+      "step": 1340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.19845238166709e-05,
+      "loss": 2.0977,
+      "step": 1341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.152544630694722e-05,
+      "loss": 2.0809,
+      "step": 1342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.106829472055202e-05,
+      "loss": 2.061,
+      "step": 1343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.061307102067092e-05,
+      "loss": 2.0563,
+      "step": 1344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0159777162210764e-05,
+      "loss": 2.0787,
+      "step": 1345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9708415091791024e-05,
+      "loss": 2.0334,
+      "step": 1346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.925898674773488e-05,
+      "loss": 2.1453,
+      "step": 1347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.88114940600618e-05,
+      "loss": 2.0326,
+      "step": 1348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.836593895047853e-05,
+      "loss": 2.0689,
+      "step": 1349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7922323332370954e-05,
+      "loss": 1.9353,
+      "step": 1350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7480649110796695e-05,
+      "loss": 2.0438,
+      "step": 1351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.704091818247558e-05,
+      "loss": 2.061,
+      "step": 1352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.660313243578229e-05,
+      "loss": 1.9906,
+      "step": 1353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6167293750738584e-05,
+      "loss": 2.0689,
+      "step": 1354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.573340399900416e-05,
+      "loss": 2.002,
+      "step": 1355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5301465043869785e-05,
+      "loss": 1.9093,
+      "step": 1356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.487147874024855e-05,
+      "loss": 2.0036,
+      "step": 1357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4443446934667845e-05,
+      "loss": 2.1165,
+      "step": 1358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.401737146526219e-05,
+      "loss": 2.0492,
+      "step": 1359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.35932541617644e-05,
+      "loss": 2.0545,
+      "step": 1360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.317109684549847e-05,
+      "loss": 2.0776,
+      "step": 1361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.275090132937115e-05,
+      "loss": 2.1115,
+      "step": 1362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2332669417864735e-05,
+      "loss": 2.1846,
+      "step": 1363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1916402907029026e-05,
+      "loss": 2.0127,
+      "step": 1364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.150210358447343e-05,
+      "loss": 1.8958,
+      "step": 1365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.10897732293597e-05,
+      "loss": 2.0652,
+      "step": 1366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0679413612393855e-05,
+      "loss": 1.9985,
+      "step": 1367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.027102649581899e-05,
+      "loss": 1.9486,
+      "step": 1368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9864613633407545e-05,
+      "loss": 2.0108,
+      "step": 1369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9460176770453545e-05,
+      "loss": 1.9678,
+      "step": 1370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9057717643765535e-05,
+      "loss": 2.0813,
+      "step": 1371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8657237981658766e-05,
+      "loss": 2.0249,
+      "step": 1372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.825873950394776e-05,
+      "loss": 2.0555,
+      "step": 1373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7862223921939385e-05,
+      "loss": 2.064,
+      "step": 1374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7467692938425056e-05,
+      "loss": 2.1184,
+      "step": 1375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7075148247673496e-05,
+      "loss": 2.0194,
+      "step": 1376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6684591535423586e-05,
+      "loss": 2.0099,
+      "step": 1377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6296024478876865e-05,
+      "loss": 2.1091,
+      "step": 1378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.590944874669089e-05,
+      "loss": 2.0349,
+      "step": 1379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.552486599897126e-05,
+      "loss": 2.0592,
+      "step": 1380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5142277887265373e-05,
+      "loss": 2.1006,
+      "step": 1381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.47616860545546e-05,
+      "loss": 1.9213,
+      "step": 1382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.438309213524754e-05,
+      "loss": 2.0228,
+      "step": 1383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.400649775517312e-05,
+      "loss": 2.1505,
+      "step": 1384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3631904531573276e-05,
+      "loss": 2.0152,
+      "step": 1385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.325931407309635e-05,
+      "loss": 2.0249,
+      "step": 1386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.288872797979009e-05,
+      "loss": 2.0274,
+      "step": 1387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.252014784309443e-05,
+      "loss": 1.9768,
+      "step": 1388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.215357524583534e-05,
+      "loss": 1.9595,
+      "step": 1389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.178901176221727e-05,
+      "loss": 2.07,
+      "step": 1390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.142645895781715e-05,
+      "loss": 2.1422,
+      "step": 1391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1065918389576784e-05,
+      "loss": 1.9993,
+      "step": 1392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.070739160579711e-05,
+      "loss": 1.9802,
+      "step": 1393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0350880146130967e-05,
+      "loss": 2.0366,
+      "step": 1394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9996385541576355e-05,
+      "loss": 2.0166,
+      "step": 1395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.964390931447053e-05,
+      "loss": 1.936,
+      "step": 1396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9293452978482793e-05,
+      "loss": 1.9417,
+      "step": 1397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8945018038608105e-05,
+      "loss": 1.9738,
+      "step": 1398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8598605991161263e-05,
+      "loss": 1.9199,
+      "step": 1399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.825421832376962e-05,
+      "loss": 2.0702,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.791185651536693e-05,
+      "loss": 2.0496,
+      "step": 1401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7571522036187513e-05,
+      "loss": 2.0071,
+      "step": 1402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7233216347759272e-05,
+      "loss": 2.0777,
+      "step": 1403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.689694090289779e-05,
+      "loss": 2.0003,
+      "step": 1404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6562697145700086e-05,
+      "loss": 2.0075,
+      "step": 1405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6230486511538143e-05,
+      "loss": 2.1925,
+      "step": 1406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5900310427053044e-05,
+      "loss": 1.9654,
+      "step": 1407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5572170310148635e-05,
+      "loss": 2.0415,
+      "step": 1408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5246067569985707e-05,
+      "loss": 2.093,
+      "step": 1409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4922003606975497e-05,
+      "loss": 1.9399,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4599979812774233e-05,
+      "loss": 2.0331,
+      "step": 1411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.427999757027677e-05,
+      "loss": 2.0575,
+      "step": 1412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3962058253610586e-05,
+      "loss": 2.1091,
+      "step": 1413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3646163228130292e-05,
+      "loss": 2.1252,
+      "step": 1414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3332313850411236e-05,
+      "loss": 1.9467,
+      "step": 1415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3020511468244297e-05,
+      "loss": 2.0912,
+      "step": 1416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2710757420629557e-05,
+      "loss": 2.1514,
+      "step": 1417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2403053037770738e-05,
+      "loss": 2.0652,
+      "step": 1418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.209739964106966e-05,
+      "loss": 2.0069,
+      "step": 1419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1793798543120303e-05,
+      "loss": 2.014,
+      "step": 1420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1492251047703137e-05,
+      "loss": 2.0119,
+      "step": 1421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.119275844977997e-05,
+      "loss": 1.9937,
+      "step": 1422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.089532203548794e-05,
+      "loss": 1.9342,
+      "step": 1423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.05999430821342e-05,
+      "loss": 2.0152,
+      "step": 1424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0306622858190238e-05,
+      "loss": 2.0578,
+      "step": 1425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0015362623286515e-05,
+      "loss": 2.0387,
+      "step": 1426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.972616362820745e-05,
+      "loss": 2.0085,
+      "step": 1427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.943902711488532e-05,
+      "loss": 2.0803,
+      "step": 1428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.915395431639544e-05,
+      "loss": 2.0791,
+      "step": 1429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8870946456950922e-05,
+      "loss": 1.9776,
+      "step": 1430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.859000475189687e-05,
+      "loss": 1.9127,
+      "step": 1431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.831113040770588e-05,
+      "loss": 2.0195,
+      "step": 1432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8034324621972133e-05,
+      "loss": 2.049,
+      "step": 1433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7759588583406928e-05,
+      "loss": 1.9749,
+      "step": 1434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7486923471833283e-05,
+      "loss": 2.0562,
+      "step": 1435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7216330458180496e-05,
+      "loss": 2.0913,
+      "step": 1436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6947810704479872e-05,
+      "loss": 2.056,
+      "step": 1437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.668136536385906e-05,
+      "loss": 1.9869,
+      "step": 1438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6416995580537664e-05,
+      "loss": 2.041,
+      "step": 1439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.615470248982165e-05,
+      "loss": 2.1099,
+      "step": 1440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5894487218099162e-05,
+      "loss": 2.1563,
+      "step": 1441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.563635088283538e-05,
+      "loss": 2.1086,
+      "step": 1442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5380294592567512e-05,
+      "loss": 1.9234,
+      "step": 1443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5126319446900515e-05,
+      "loss": 2.006,
+      "step": 1444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4874426536501894e-05,
+      "loss": 2.1003,
+      "step": 1445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4624616943097247e-05,
+      "loss": 2.0895,
+      "step": 1446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4376891739465824e-05,
+      "loss": 2.0987,
+      "step": 1447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4131251989435434e-05,
+      "loss": 1.9545,
+      "step": 1448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3887698747878263e-05,
+      "loss": 2.0026,
+      "step": 1449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3646233060706226e-05,
+      "loss": 2.0492,
+      "step": 1450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3406855964866405e-05,
+      "loss": 2.0705,
+      "step": 1451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.316956848833667e-05,
+      "loss": 2.1291,
+      "step": 1452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2934371650121457e-05,
+      "loss": 2.167,
+      "step": 1453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2701266460246885e-05,
+      "loss": 2.049,
+      "step": 1454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.247025391975698e-05,
+      "loss": 2.1115,
+      "step": 1455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2241335020708844e-05,
+      "loss": 1.9536,
+      "step": 1456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2014510746168995e-05,
+      "loss": 1.9942,
+      "step": 1457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1789782070208588e-05,
+      "loss": 1.972,
+      "step": 1458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1567149957899593e-05,
+      "loss": 2.0336,
+      "step": 1459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1346615365310453e-05,
+      "loss": 2.0645,
+      "step": 1460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1128179239502046e-05,
+      "loss": 2.1064,
+      "step": 1461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0911842518523673e-05,
+      "loss": 2.1099,
+      "step": 1462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0697606131408966e-05,
+      "loss": 1.9866,
+      "step": 1463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.048547099817182e-05,
+      "loss": 2.0888,
+      "step": 1464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0275438029802797e-05,
+      "loss": 2.0214,
+      "step": 1465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0067508128264568e-05,
+      "loss": 2.0284,
+      "step": 1466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.861682186488696e-06,
+      "loss": 2.0187,
+      "step": 1467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.65796108837147e-06,
+      "loss": 2.1388,
+      "step": 1468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.456345708770075e-06,
+      "loss": 2.0939,
+      "step": 1469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.25683691349899e-06,
+      "loss": 2.0337,
+      "step": 1470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.059435559326258e-06,
+      "loss": 2.0728,
+      "step": 1471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.864142493969718e-06,
+      "loss": 1.986,
+      "step": 1472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.6709585560934e-06,
+      "loss": 1.9078,
+      "step": 1473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.479884575303798e-06,
+      "loss": 2.045,
+      "step": 1474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.290921372146654e-06,
+      "loss": 1.9891,
+      "step": 1475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.104069758103017e-06,
+      "loss": 2.0907,
+      "step": 1476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.919330535586133e-06,
+      "loss": 2.0122,
+      "step": 1477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.73670449793773e-06,
+      "loss": 2.0846,
+      "step": 1478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.556192429424569e-06,
+      "loss": 2.0389,
+      "step": 1479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.377795105235563e-06,
+      "loss": 1.9995,
+      "step": 1480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.201513291477669e-06,
+      "loss": 2.0477,
+      "step": 1481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.0273477451733295e-06,
+      "loss": 2.0437,
+      "step": 1482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.855299214256816e-06,
+      "loss": 2.0877,
+      "step": 1483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.68536843757106e-06,
+      "loss": 1.9569,
+      "step": 1484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.517556144864712e-06,
+      "loss": 2.0714,
+      "step": 1485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.351863056788587e-06,
+      "loss": 1.9801,
+      "step": 1486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.188289884893061e-06,
+      "loss": 1.9997,
+      "step": 1487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.026837331624513e-06,
+      "loss": 2.0261,
+      "step": 1488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.867506090322772e-06,
+      "loss": 2.0547,
+      "step": 1489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7102968452179e-06,
+      "loss": 2.1389,
+      "step": 1490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.555210271427191e-06,
+      "loss": 1.9253,
+      "step": 1491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.402247034952456e-06,
+      "loss": 2.0889,
+      "step": 1492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.251407792677021e-06,
+      "loss": 2.0328,
+      "step": 1493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.102693192362895e-06,
+      "loss": 2.0667,
+      "step": 1494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.956103872648332e-06,
+      "loss": 2.1119,
+      "step": 1495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.811640463044498e-06,
+      "loss": 1.9774,
+      "step": 1496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6693035839331375e-06,
+      "loss": 2.1061,
+      "step": 1497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.529093846564025e-06,
+      "loss": 2.0502,
+      "step": 1498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.391011853051963e-06,
+      "loss": 2.1556,
+      "step": 1499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.255058196374506e-06,
+      "loss": 2.1004,
+      "step": 1500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1212334603693e-06,
+      "loss": 2.048,
+      "step": 1501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.98953821973147e-06,
+      "loss": 2.111,
+      "step": 1502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.859973040011511e-06,
+      "loss": 2.0643,
+      "step": 1503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.732538477612346e-06,
+      "loss": 2.0215,
+      "step": 1504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.60723507978733e-06,
+      "loss": 2.0689,
+      "step": 1505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.484063384637748e-06,
+      "loss": 2.1398,
+      "step": 1506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3630239211105417e-06,
+      "loss": 2.0233,
+      "step": 1507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.244117208996089e-06,
+      "loss": 1.9022,
+      "step": 1508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1273437589257048e-06,
+      "loss": 2.0525,
+      "step": 1509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.012704072369921e-06,
+      "loss": 2.0592,
+      "step": 1510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.900198641635876e-06,
+      "loss": 2.159,
+      "step": 1511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.78982794986532e-06,
+      "loss": 2.0346,
+      "step": 1512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6815924710329455e-06,
+      "loss": 2.1132,
+      "step": 1513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.575492669943613e-06,
+      "loss": 1.8871,
+      "step": 1514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.471529002231021e-06,
+      "loss": 2.0232,
+      "step": 1515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3697019143554268e-06,
+      "loss": 2.1101,
+      "step": 1516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2700118436016494e-06,
+      "loss": 1.9881,
+      "step": 1517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1724592180775713e-06,
+      "loss": 2.0167,
+      "step": 1518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0770444567118076e-06,
+      "loss": 2.0315,
+      "step": 1519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.98376796925237e-06,
+      "loss": 2.0213,
+      "step": 1520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8926301562645631e-06,
+      "loss": 1.9744,
+      "step": 1521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8036314091294248e-06,
+      "loss": 1.9864,
+      "step": 1522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7167721100420087e-06,
+      "loss": 2.1043,
+      "step": 1523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6320526320098283e-06,
+      "loss": 2.1207,
+      "step": 1524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5494733388510818e-06,
+      "loss": 2.1295,
+      "step": 1525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4690345851932074e-06,
+      "loss": 2.0264,
+      "step": 1526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3907367164713303e-06,
+      "loss": 1.9259,
+      "step": 1527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3145800689268738e-06,
+      "loss": 1.9978,
+      "step": 1528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.240564969605895e-06,
+      "loss": 2.1096,
+      "step": 1529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1686917363579187e-06,
+      "loss": 2.1318,
+      "step": 1530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0989606778344375e-06,
+      "loss": 2.0157,
+      "step": 1531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.031372093487526e-06,
+      "loss": 2.0984,
+      "step": 1532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.6592627356884e-07,
+      "loss": 2.081,
+      "step": 1533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.02623499128008e-07,
+      "loss": 1.9643,
+      "step": 1534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.414640420116304e-07,
+      "loss": 2.0651,
+      "step": 1535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.824481648621151e-07,
+      "loss": 2.0144,
+      "step": 1536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.255761211165113e-07,
+      "loss": 1.9446,
+      "step": 1537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.708481550052881e-07,
+      "loss": 2.1112,
+      "step": 1538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.182645015516131e-07,
+      "loss": 2.0769,
+      "step": 1539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.678253865700756e-07,
+      "loss": 2.123,
+      "step": 1540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.195310266656317e-07,
+      "loss": 2.042,
+      "step": 1541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.733816292330495e-07,
+      "loss": 2.0879,
+      "step": 1542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.293773924556321e-07,
+      "loss": 1.8932,
+      "step": 1543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.875185053047181e-07,
+      "loss": 1.9342,
+      "step": 1544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4780514753851577e-07,
+      "loss": 1.8892,
+      "step": 1545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1023748970149257e-07,
+      "loss": 1.9717,
+      "step": 1546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7481569312381993e-07,
+      "loss": 2.0279,
+      "step": 1547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4153990992031863e-07,
+      "loss": 2.0209,
+      "step": 1548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1041028299012554e-07,
+      "loss": 2.0255,
+      "step": 1549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.814269460158613e-07,
+      "loss": 2.1133,
+      "step": 1550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5459002346324135e-07,
+      "loss": 2.112,
+      "step": 1551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.298996305804101e-07,
+      "loss": 1.9504,
+      "step": 1552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0735587339749664e-07,
+      "loss": 2.033,
+      "step": 1553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.695884872600423e-08,
+      "loss": 2.0499,
+      "step": 1554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.870864415875477e-08,
+      "loss": 1.9979,
+      "step": 1555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.260533806911161e-08,
+      "loss": 2.0888,
+      "step": 1556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.864899961097956e-08,
+      "loss": 2.0155,
+      "step": 1557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.683968871824982e-08,
+      "loss": 2.0571,
+      "step": 1558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7177456104688903e-08,
+      "loss": 2.1045,
+      "step": 1559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.6623432636056e-09,
+      "loss": 1.9497,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.294382467906477e-09,
+      "loss": 1.9055,
+      "step": 1561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0735967695962856e-09,
+      "loss": 2.1567,
+      "step": 1562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.0059,
+      "step": 1563
+    },
+    {
+      "epoch": 1.0,
+      "step": 1563,
+      "total_flos": 5.05049150555947e+17,
+      "train_loss": 2.211227413788867,
+      "train_runtime": 13197.7803,
+      "train_samples_per_second": 15.154,
+      "train_steps_per_second": 0.118
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1563,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 24000,
+  "total_flos": 5.05049150555947e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}