diff --git "a/lora-checkpoints-1/trainer_state.json" "b/lora-checkpoints-1/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/lora-checkpoints-1/trainer_state.json"
@@ -0,0 +1,10675 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 1775,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 1.6156,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 1.5653,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 1.3969,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 1.2164,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 1.2245,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 1.4574,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5925925925925925e-05,
+      "loss": 1.2275,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 1.2384,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.3397,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 1.0733,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 1.047,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 1.0879,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 1.4383,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.185185185185185e-05,
+      "loss": 1.0471,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 1.1159,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.925925925925926e-05,
+      "loss": 1.569,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.296296296296296e-05,
+      "loss": 1.3859,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.0769,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.037037037037038e-05,
+      "loss": 0.8342,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.407407407407407e-05,
+      "loss": 1.2498,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 0.9921,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.148148148148148e-05,
+      "loss": 1.0894,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.518518518518518e-05,
+      "loss": 1.2262,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 1.0083,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.25925925925926e-05,
+      "loss": 1.0188,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.62962962962963e-05,
+      "loss": 1.2602,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 1.0527,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001037037037037037,
+      "loss": 0.9503,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010740740740740742,
+      "loss": 1.0101,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011111111111111112,
+      "loss": 1.0957,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011481481481481482,
+      "loss": 0.976,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011851851851851852,
+      "loss": 0.9268,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012222222222222224,
+      "loss": 0.8646,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012592592592592592,
+      "loss": 1.0684,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012962962962962963,
+      "loss": 0.915,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 1.2329,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013703703703703705,
+      "loss": 1.1997,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014074074074074076,
+      "loss": 0.7975,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014444444444444444,
+      "loss": 0.8977,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014814814814814815,
+      "loss": 0.8541,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015185185185185185,
+      "loss": 0.843,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015555555555555556,
+      "loss": 1.0504,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015925925925925927,
+      "loss": 1.1616,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016296296296296295,
+      "loss": 1.0504,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 1.0339,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017037037037037037,
+      "loss": 1.3559,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017407407407407408,
+      "loss": 0.9241,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017777777777777779,
+      "loss": 0.8159,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001814814814814815,
+      "loss": 0.9279,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001851851851851852,
+      "loss": 0.8833,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 0.7633,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001925925925925926,
+      "loss": 0.8833,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001962962962962963,
+      "loss": 1.1069,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8578,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999983338738205,
+      "loss": 0.795,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999933355008333,
+      "loss": 0.9512,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999850048976945,
+      "loss": 0.9342,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999973342092164,
+      "loss": 1.0502,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999583471231046,
+      "loss": 0.9382,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999940020040484,
+      "loss": 1.0215,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999918360905372,
+      "loss": 0.8625,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999893369789943,
+      "loss": 1.0226,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998650467774733,
+      "loss": 1.111,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998333919623424,
+      "loss": 0.8417,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999798405450032,
+      "loss": 1.1361,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997600873571263,
+      "loss": 0.7799,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997184378113105,
+      "loss": 0.9244,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996734569513715,
+      "loss": 0.8649,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996251449271973,
+      "loss": 0.7906,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999573501899775,
+      "loss": 0.9595,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995185280411928,
+      "loss": 1.0123,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994602235346372,
+      "loss": 0.8628,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993985885743938,
+      "loss": 0.9477,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993336233658455,
+      "loss": 1.0705,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999265328125473,
+      "loss": 0.8122,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991937030808528,
+      "loss": 0.9656,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991187484706585,
+      "loss": 0.9207,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999040464544657,
+      "loss": 0.9395,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989588515637106,
+      "loss": 0.9005,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998873909799774,
+      "loss": 0.7768,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987856395358946,
+      "loss": 0.9209,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986940410662114,
+      "loss": 0.8841,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985991146959537,
+      "loss": 0.8465,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199850086074144,
+      "loss": 1.5621,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998399279530077,
+      "loss": 1.0367,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998294371400359,
+      "loss": 1.0315,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981861367018668,
+      "loss": 0.9351,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980745757952655,
+      "loss": 0.9467,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979596890523041,
+      "loss": 0.9819,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978414768558143,
+      "loss": 0.7389,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977199395997087,
+      "loss": 0.841,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975950776889804,
+      "loss": 0.8481,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997466891539701,
+      "loss": 1.0625,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973353815790187,
+      "loss": 0.9027,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972005482451582,
+      "loss": 0.8663,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970623919874178,
+      "loss": 1.0159,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996920913266169,
+      "loss": 0.9159,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967761125528553,
+      "loss": 1.0981,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966279903299884,
+      "loss": 0.8146,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964765470911492,
+      "loss": 0.7222,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963217833409854,
+      "loss": 0.9921,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996163699595208,
+      "loss": 1.3971,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960022963805922,
+      "loss": 1.1027,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958375742349746,
+      "loss": 0.7822,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956695337072504,
+      "loss": 1.0715,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954981753573733,
+      "loss": 1.0565,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995323499756353,
+      "loss": 1.3881,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951455074862517,
+      "loss": 0.9813,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949641991401856,
+      "loss": 1.0453,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994779575322319,
+      "loss": 0.8274,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994591636647866,
+      "loss": 0.7891,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019944003837430845,
+      "loss": 0.9952,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942058172452783,
+      "loss": 0.8288,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940079378027919,
+      "loss": 0.8955,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938067460750097,
+      "loss": 0.8128,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019936022427323528,
+      "loss": 0.8954,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933944284562785,
+      "loss": 0.9127,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931833039392758,
+      "loss": 1.5297,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929688698848655,
+      "loss": 1.4747,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992751127007596,
+      "loss": 0.9704,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925300760330407,
+      "loss": 1.0599,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923057176977977,
+      "loss": 0.8956,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992078052749486,
+      "loss": 0.8015,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918470819467422,
+      "loss": 0.8177,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916128060592196,
+      "loss": 0.9083,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913752258675843,
+      "loss": 1.3935,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911343421635133,
+      "loss": 1.1009,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908901557496927,
+      "loss": 0.9699,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906426674398122,
+      "loss": 0.8572,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903918780585656,
+      "loss": 0.8358,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019901377884416473,
+      "loss": 1.0103,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989880399435747,
+      "loss": 0.8636,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198961971189855,
+      "loss": 0.8481,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893557266987332,
+      "loss": 0.9202,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989088444715962,
+      "loss": 0.9665,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988817866840887,
+      "loss": 1.1398,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988543993975142,
+      "loss": 0.8528,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882668270313413,
+      "loss": 0.813,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879863669330743,
+      "loss": 0.7961,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987702614614905,
+      "loss": 1.1777,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874155710223677,
+      "loss": 1.0353,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871252371119643,
+      "loss": 0.8068,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019868316138511603,
+      "loss": 0.8066,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019865347022183827,
+      "loss": 1.0352,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986234503203016,
+      "loss": 0.8572,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019859310178053994,
+      "loss": 0.9557,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856242470368222,
+      "loss": 1.3891,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019853141919195223,
+      "loss": 0.9101,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019850008534866818,
+      "loss": 0.9016,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984684232782423,
+      "loss": 0.7849,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019843643308618068,
+      "loss": 1.1141,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001984041148790826,
+      "loss": 0.8641,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837146876464058,
+      "loss": 0.9155,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833849485163962,
+      "loss": 0.945,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983051932499572,
+      "loss": 1.0809,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827156407056264,
+      "loss": 0.9264,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823760742551685,
+      "loss": 0.6897,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019820332342797192,
+      "loss": 0.7899,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981687121921708,
+      "loss": 0.8011,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813377383344686,
+      "loss": 0.8901,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980985084682235,
+      "loss": 0.8281,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806291621401388,
+      "loss": 0.8769,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802699718942033,
+      "loss": 0.7309,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799075151413412,
+      "loss": 0.9114,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795417930893494,
+      "loss": 0.9296,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019791728069569066,
+      "loss": 1.5162,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788005579735675,
+      "loss": 0.836,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000197842504737976,
+      "loss": 0.7553,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019780462764267798,
+      "loss": 1.0793,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977664246376787,
+      "loss": 0.8704,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977278958502803,
+      "loss": 0.8467,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019768904140887033,
+      "loss": 0.9529,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764986144292165,
+      "loss": 0.9642,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976103560829918,
+      "loss": 0.9745,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757052546072256,
+      "loss": 0.9918,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975303697088397,
+      "loss": 0.9228,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019748988896115221,
+      "loss": 0.8553,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019744908335255222,
+      "loss": 1.459,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019740795301901435,
+      "loss": 0.8777,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736649809759516,
+      "loss": 1.1507,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019732471872643297,
+      "loss": 1.5318,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972826150447472,
+      "loss": 0.7829,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972401871928379,
+      "loss": 0.7352,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971974353120854,
+      "loss": 0.7926,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019715435954494975,
+      "loss": 0.737,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019711096003497026,
+      "loss": 0.7938,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970672369267651,
+      "loss": 0.8829,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702319036603067,
+      "loss": 0.7432,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969788204995412,
+      "loss": 1.0677,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969341274751483,
+      "loss": 1.0395,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019688911144178045,
+      "loss": 0.7546,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968437725494424,
+      "loss": 0.8733,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019679811094921478,
+      "loss": 1.068,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019675212679325354,
+      "loss": 0.886,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019670582023478952,
+      "loss": 0.8922,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665919142812784,
+      "loss": 0.8347,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019661224052864752,
+      "loss": 1.0944,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965649676928007,
+      "loss": 1.0164,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019651737307811246,
+      "loss": 1.0235,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019646945684318007,
+      "loss": 1.032,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964212191476725,
+      "loss": 1.0406,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019637266015232993,
+      "loss": 0.9182,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963237800189632,
+      "loss": 0.9003,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001962745789104532,
+      "loss": 1.0019,
+      "step": 204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622505699075048,
+      "loss": 0.871,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961752144248746,
+      "loss": 0.7815,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019612505137891353,
+      "loss": 1.337,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960745680200232,
+      "loss": 1.0935,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960237645164269,
+      "loss": 1.0427,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019597264103741473,
+      "loss": 0.7992,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019592119775334302,
+      "loss": 1.4345,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958694348356338,
+      "loss": 0.9539,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019581735245677412,
+      "loss": 1.5357,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001957649507903157,
+      "loss": 0.8563,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571223001087402,
+      "loss": 1.0575,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019565919029412808,
+      "loss": 1.0911,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019560583181681956,
+      "loss": 0.8674,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955521547567524,
+      "loss": 0.7726,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019549815929279212,
+      "loss": 0.7347,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544384560486524,
+      "loss": 0.775,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953892138739586,
+      "loss": 1.0373,
+      "step": 221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000195334264282119,
+      "loss": 1.1528,
+      "step": 222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527899701245232,
+      "loss": 0.8832,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019522341224912303,
+      "loss": 0.9369,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019516751017735362,
+      "loss": 0.7424,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019511129098342383,
+      "loss": 1.2663,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019505475485467031,
+      "loss": 1.0327,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019499790197948565,
+      "loss": 0.7705,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019494073254731794,
+      "loss": 0.8773,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019488324674867024,
+      "loss": 0.9251,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019482544477509968,
+      "loss": 0.7985,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019476732681921702,
+      "loss": 0.8283,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019470889307468601,
+      "loss": 1.0076,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019465014373622256,
+      "loss": 0.9597,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019459107899959434,
+      "loss": 1.3785,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019453169906161994,
+      "loss": 0.9615,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019447200412016832,
+      "loss": 0.8347,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019441199437415807,
+      "loss": 1.0783,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019435167002355682,
+      "loss": 0.7429,
+      "step": 239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019429103126938048,
+      "loss": 0.7747,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019423007831369273,
+      "loss": 1.0095,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019416881135960422,
+      "loss": 1.3889,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019410723061127186,
+      "loss": 0.9639,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019404533627389827,
+      "loss": 0.9272,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019398312855373098,
+      "loss": 0.8153,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392060765806185,
+      "loss": 0.8101,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019385777379522622,
+      "loss": 0.7923,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937946271746024,
+      "loss": 0.8118,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373116800661092,
+      "loss": 0.657,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019366739650271368,
+      "loss": 0.7612,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019360331287541344,
+      "loss": 0.8982,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019353891733825304,
+      "loss": 0.9628,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019347421010581465,
+      "loss": 0.9822,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019340919139371903,
+      "loss": 0.9402,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019334386141862505,
+      "loss": 0.7639,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019327822039822863,
+      "loss": 0.983,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932122685512622,
+      "loss": 0.7007,
+      "step": 257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019314600609749395,
+      "loss": 0.7921,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019307943325772714,
+      "loss": 1.046,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001930125502537992,
+      "loss": 0.6867,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019294535730858127,
+      "loss": 0.8388,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928778546459771,
+      "loss": 0.8017,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019281004249092272,
+      "loss": 0.9773,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019274192106938522,
+      "loss": 1.2289,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019267349060836245,
+      "loss": 0.8394,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019260475133588194,
+      "loss": 0.8306,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019253570348100029,
+      "loss": 0.7998,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019246634727380238,
+      "loss": 0.88,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923966829454006,
+      "loss": 0.9796,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019232671072793412,
+      "loss": 1.4552,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019225643085456792,
+      "loss": 0.8625,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019218584355949235,
+      "loss": 0.7941,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921149490779221,
+      "loss": 0.8139,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019204374764609546,
+      "loss": 0.8766,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019197223950127352,
+      "loss": 0.9313,
+      "step": 275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019190042488173952,
+      "loss": 0.8172,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019182830402679786,
+      "loss": 1.0091,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917558771767735,
+      "loss": 0.8192,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019168314457301087,
+      "loss": 0.9843,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019161010645787343,
+      "loss": 0.871,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019153676307474263,
+      "loss": 0.8803,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019146311466801709,
+      "loss": 1.3221,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019138916148311186,
+      "loss": 0.7727,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019131490376645769,
+      "loss": 0.8603,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019124034176549998,
+      "loss": 0.9902,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019116547572869816,
+      "loss": 1.0216,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019109030590552478,
+      "loss": 0.8401,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910148325464646,
+      "loss": 1.0606,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019093905590301388,
+      "loss": 0.78,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019086297622767964,
+      "loss": 0.9343,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019078659377397842,
+      "loss": 0.8022,
+      "step": 291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070990879643595,
+      "loss": 0.775,
+      "step": 292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019063292155058586,
+      "loss": 0.9838,
+      "step": 293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055563229296906,
+      "loss": 0.8544,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019047804128113295,
+      "loss": 1.0533,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019040014877363024,
+      "loss": 0.7176,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019032195503001852,
+      "loss": 0.8775,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019024346031085908,
+      "loss": 1.0344,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019016466487771605,
+      "loss": 0.9044,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019008556899315575,
+      "loss": 1.0232,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019000617292074562,
+      "loss": 1.0462,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018992647692505344,
+      "loss": 0.9079,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018984648127164637,
+      "loss": 0.8502,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018976618622709005,
+      "loss": 0.9477,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896855920589479,
+      "loss": 0.8901,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018960469903578,
+      "loss": 0.9076,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018952350742714236,
+      "loss": 0.7829,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018944201750358585,
+      "loss": 0.8432,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893602295366555,
+      "loss": 0.9024,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001892781437988894,
+      "loss": 0.7057,
+      "step": 310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018919576056381803,
+      "loss": 0.7502,
+      "step": 311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018911308010596302,
+      "loss": 0.9563,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018903010270083663,
+      "loss": 0.8973,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018894682862494042,
+      "loss": 0.9902,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888632581557647,
+      "loss": 1.0053,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018877939157178726,
+      "loss": 0.8633,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018869522915247283,
+      "loss": 0.7928,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886107711782718,
+      "loss": 1.134,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018852601793061945,
+      "loss": 0.8021,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018844096969193494,
+      "loss": 1.5151,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018835562674562054,
+      "loss": 0.7987,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018826998937606043,
+      "loss": 0.9297,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018818405786861996,
+      "loss": 0.8499,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880978325096446,
+      "loss": 0.7189,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188011313586459,
+      "loss": 0.7418,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018792450138736607,
+      "loss": 0.7652,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018783739620164592,
+      "loss": 0.7955,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018774999831955506,
+      "loss": 0.8644,
+      "step": 328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876623080323252,
+      "loss": 0.7352,
+      "step": 329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757432563216265,
+      "loss": 1.0154,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018748605141224683,
+      "loss": 0.7747,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018739748566672982,
+      "loss": 0.8373,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018730862869073493,
+      "loss": 0.8233,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018721948078035614,
+      "loss": 0.9781,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018713004223265672,
+      "loss": 0.9515,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001870403133456685,
+      "loss": 1.0041,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018695029441839075,
+      "loss": 0.855,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018685998575078928,
+      "loss": 0.9206,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018676938764379539,
+      "loss": 1.379,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018667850039930477,
+      "loss": 1.0057,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865873243201767,
+      "loss": 0.9008,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018649585971023284,
+      "loss": 0.7521,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001864041068742564,
+      "loss": 1.1338,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000186312066117991,
+      "loss": 0.7557,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018621973774813957,
+      "loss": 1.0157,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018612712207236365,
+      "loss": 0.7837,
+      "step": 346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000186034219399282,
+      "loss": 0.8779,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018594103003846972,
+      "loss": 0.8541,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018584755430045734,
+      "loss": 0.8277,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001857537924967296,
+      "loss": 0.6714,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856597449397245,
+      "loss": 0.9007,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018556541194283223,
+      "loss": 1.2025,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018547079382039407,
+      "loss": 1.2175,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853758908877016,
+      "loss": 0.9114,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018528070346099525,
+      "loss": 0.7793,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851852318574636,
+      "loss": 0.8605,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018508947639524212,
+      "loss": 0.8143,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018499343739341212,
+      "loss": 0.8905,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018489711517199987,
+      "loss": 0.9302,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018480051005197528,
+      "loss": 0.7711,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018470362235525095,
+      "loss": 0.8955,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018460645240468124,
+      "loss": 0.9303,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018450900052406085,
+      "loss": 0.8873,
+      "step": 363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018441126703812408,
+      "loss": 0.9313,
+      "step": 364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018431325227254353,
+      "loss": 0.8167,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842149565539292,
+      "loss": 0.8987,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018411638020982717,
+      "loss": 1.2625,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018401752356871876,
+      "loss": 0.7683,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018391838696001919,
+      "loss": 0.7065,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018381897071407665,
+      "loss": 0.8583,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837192751621712,
+      "loss": 1.0727,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836193006365136,
+      "loss": 1.0953,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018351904747024414,
+      "loss": 1.0089,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018341851599743168,
+      "loss": 0.8007,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018331770655307246,
+      "loss": 0.7265,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018321661947308902,
+      "loss": 1.1326,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000183115255094329,
+      "loss": 0.9342,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018301361375456406,
+      "loss": 1.2021,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018291169579248885,
+      "loss": 0.7478,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018280950154771973,
+      "loss": 0.8206,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018270703136079365,
+      "loss": 0.7933,
+      "step": 381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018260428557316722,
+      "loss": 0.9599,
+      "step": 382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018250126452721527,
+      "loss": 1.0447,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018239796856622998,
+      "loss": 0.8113,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018229439803441947,
+      "loss": 0.8049,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000182190553276907,
+      "loss": 0.8866,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001820864346397294,
+      "loss": 0.7476,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018198204246983632,
+      "loss": 0.8021,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818773771150888,
+      "loss": 0.8439,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018177243892425818,
+      "loss": 0.7912,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018166722824702506,
+      "loss": 0.7333,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018156174543397792,
+      "loss": 0.9418,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001814559908366121,
+      "loss": 0.9187,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018134996480732865,
+      "loss": 0.9454,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018124366769943303,
+      "loss": 0.86,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181137099867134,
+      "loss": 1.0415,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018103026166554253,
+      "loss": 0.8858,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018092315345067043,
+      "loss": 0.9148,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018081577557942934,
+      "loss": 0.8514,
+      "step": 399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018070812840962938,
+      "loss": 0.9579,
+      "step": 400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001806002122999781,
+      "loss": 0.7442,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018049202761007921,
+      "loss": 0.8042,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018038357470043145,
+      "loss": 0.8746,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018027485393242717,
+      "loss": 0.7722,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801658656683515,
+      "loss": 0.776,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018005661027138082,
+      "loss": 0.792,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017994708810558169,
+      "loss": 0.8237,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017983729953590958,
+      "loss": 0.884,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797272449282077,
+      "loss": 0.8552,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017961692464920582,
+      "loss": 0.8579,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017950633906651896,
+      "loss": 0.7286,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001793954885486461,
+      "loss": 0.8987,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017928437346496922,
+      "loss": 1.0588,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017917299418575183,
+      "loss": 0.7697,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017906135108213773,
+      "loss": 1.4389,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017894944452614994,
+      "loss": 1.4422,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788372748906894,
+      "loss": 0.8038,
+      "step": 417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017872484254953354,
+      "loss": 1.1945,
+      "step": 418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786121478773354,
+      "loss": 0.9607,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017849919124962197,
+      "loss": 1.2117,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017838597304279335,
+      "loss": 0.9041,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017827249363412104,
+      "loss": 0.9118,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781587534017472,
+      "loss": 0.9351,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001780447527246829,
+      "loss": 0.9212,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779304919828072,
+      "loss": 1.0091,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017781597155686572,
+      "loss": 0.7895,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017770119182846942,
+      "loss": 0.9084,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017758615318009334,
+      "loss": 0.7537,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017747085599507524,
+      "loss": 0.8332,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017735530065761447,
+      "loss": 0.8075,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017723948755277063,
+      "loss": 0.6878,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001771234170664621,
+      "loss": 0.7899,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017700708958546507,
+      "loss": 0.9414,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768905054974121,
+      "loss": 0.916,
+      "step": 434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017677366519079074,
+      "loss": 0.7936,
+      "step": 435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017665656905494246,
+      "loss": 1.0211,
+      "step": 436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017653921748006106,
+      "loss": 0.9869,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764216108571916,
+      "loss": 0.8318,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763037495782291,
+      "loss": 0.8525,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000176185634035917,
+      "loss": 0.727,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017606726462384616,
+      "loss": 0.6808,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759486417364533,
+      "loss": 0.8207,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017582976576901982,
+      "loss": 1.3319,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757106371176704,
+      "loss": 0.8764,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017559125617937186,
+      "loss": 0.7765,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017547162335193156,
+      "loss": 0.9151,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017535173903399627,
+      "loss": 0.8388,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001752316036250508,
+      "loss": 1.3553,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017511121752541664,
+      "loss": 1.3692,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017499058113625067,
+      "loss": 0.8119,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748696948595438,
+      "loss": 1.4146,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017474855909811954,
+      "loss": 0.7804,
+      "step": 452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017462717425563288,
+      "loss": 0.8669,
+      "step": 453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017450554073656873,
+      "loss": 0.8945,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017438365894624069,
+      "loss": 0.792,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017426152929078963,
+      "loss": 1.0091,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017413915217718236,
+      "loss": 1.4239,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740165280132103,
+      "loss": 0.8425,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738936572074882,
+      "loss": 0.8218,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017377054016945247,
+      "loss": 0.816,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017364717730936025,
+      "loss": 0.8223,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017352356903828763,
+      "loss": 0.9625,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017339971576812863,
+      "loss": 0.6561,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017327561791159358,
+      "loss": 0.9804,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017315127588220786,
+      "loss": 1.0037,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001730266900943105,
+      "loss": 0.8521,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017290186096305274,
+      "loss": 0.7289,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001727767889043968,
+      "loss": 0.8317,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017265147433511433,
+      "loss": 1.1957,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725259176727851,
+      "loss": 0.6649,
+      "step": 470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017240011933579557,
+      "loss": 0.9748,
+      "step": 471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017227407974333757,
+      "loss": 0.8207,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017214779931540682,
+      "loss": 0.9306,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017202127847280158,
+      "loss": 0.7632,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001718945176371212,
+      "loss": 0.8285,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017176751723076483,
+      "loss": 0.8848,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017164027767692982,
+      "loss": 0.7527,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017151279939961052,
+      "loss": 0.7575,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017138508282359667,
+      "loss": 0.7868,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017125712837447215,
+      "loss": 1.3437,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017112893647861348,
+      "loss": 1.0907,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017100050756318843,
+      "loss": 0.7053,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001708718420561545,
+      "loss": 0.9962,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017074294038625767,
+      "loss": 0.7606,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017061380298303086,
+      "loss": 0.7304,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017048443027679243,
+      "loss": 0.905,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001703548226986449,
+      "loss": 0.7237,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017022498068047342,
+      "loss": 1.3507,
+      "step": 488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001700949046549444,
+      "loss": 0.8248,
+      "step": 489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000169964595055504,
+      "loss": 0.6065,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698340523163766,
+      "loss": 0.9477,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016970327687256364,
+      "loss": 0.838,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016957226915984184,
+      "loss": 0.9129,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016944102961476197,
+      "loss": 0.8447,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016930955867464737,
+      "loss": 1.2855,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001691778567775923,
+      "loss": 0.7336,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690459243624608,
+      "loss": 0.9687,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016891376186888493,
+      "loss": 1.2448,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016878136973726347,
+      "loss": 1.0912,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016864874840876045,
+      "loss": 1.0546,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685158983253035,
+      "loss": 1.1884,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016838281992958277,
+      "loss": 0.8389,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016824951366504895,
+      "loss": 0.8537,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001681159799759122,
+      "loss": 1.26,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016798221930714043,
+      "loss": 0.7816,
+      "step": 505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016784823210445806,
+      "loss": 0.7435,
+      "step": 506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016771401881434409,
+      "loss": 0.9165,
+      "step": 507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001675795798840312,
+      "loss": 1.3665,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001674449157615038,
+      "loss": 0.8221,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016731002689549673,
+      "loss": 0.8232,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016717491373549373,
+      "loss": 0.9572,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016703957673172592,
+      "loss": 0.8271,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016690401633517039,
+      "loss": 0.8954,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016676823299754856,
+      "loss": 0.8934,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016663222717132477,
+      "loss": 0.7789,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016649599930970482,
+      "loss": 0.8003,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016635954986663428,
+      "loss": 0.8889,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662228792967971,
+      "loss": 0.7812,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016608598805561414,
+      "loss": 0.775,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016594887659924154,
+      "loss": 0.767,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016581154538456935,
+      "loss": 0.8532,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016567399486921972,
+      "loss": 0.9951,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553622551154578,
+      "loss": 0.9263,
+      "step": 523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016539823777062971,
+      "loss": 0.9887,
+      "step": 524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016526003210628155,
+      "loss": 0.8265,
+      "step": 525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016512160897903745,
+      "loss": 0.7895,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016498296885015815,
+      "loss": 0.7976,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016484411218162758,
+      "loss": 0.8204,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016470503943615122,
+      "loss": 1.0237,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001645657510771545,
+      "loss": 0.8149,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016442624756878143,
+      "loss": 0.8061,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016428652937589293,
+      "loss": 0.7127,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641465969640652,
+      "loss": 0.9787,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640064507995884,
+      "loss": 0.8259,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016386609134946492,
+      "loss": 0.7714,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016372551908140786,
+      "loss": 0.8236,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016358473446383945,
+      "loss": 0.7797,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634437379658896,
+      "loss": 0.7276,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016330253005739423,
+      "loss": 0.7547,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016316111120889376,
+      "loss": 1.2456,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001630194818916314,
+      "loss": 1.0233,
+      "step": 541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016287764257755182,
+      "loss": 0.8906,
+      "step": 542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016273559373929946,
+      "loss": 0.8644,
+      "step": 543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001625933358502168,
+      "loss": 1.12,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016245086938434307,
+      "loss": 0.7379,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016230819481641246,
+      "loss": 0.8455,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621653126218527,
+      "loss": 0.8303,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016202222327678327,
+      "loss": 0.9664,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016187892725801398,
+      "loss": 0.8248,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617354250430434,
+      "loss": 0.784,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016159171711005698,
+      "loss": 0.8401,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016144780393792594,
+      "loss": 1.4073,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016130368600620528,
+      "loss": 0.8576,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016115936379513222,
+      "loss": 0.709,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001610148377856249,
+      "loss": 1.0814,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608701084592804,
+      "loss": 0.8172,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016072517629837335,
+      "loss": 0.7244,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016058004178585433,
+      "loss": 0.7825,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604347054053481,
+      "loss": 0.6828,
+      "step": 559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016028916764115224,
+      "loss": 0.8382,
+      "step": 560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016014342897823522,
+      "loss": 1.0926,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015999748990223512,
+      "loss": 0.7837,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015985135089945774,
+      "loss": 1.0089,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015970501245687508,
+      "loss": 0.9058,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015955847506212383,
+      "loss": 0.8571,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015941173920350348,
+      "loss": 0.85,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000159264805369975,
+      "loss": 0.9401,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015911767405115903,
+      "loss": 0.9412,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001589703457373342,
+      "loss": 1.2477,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015882282091943565,
+      "loss": 0.7708,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015867510008905333,
+      "loss": 0.8532,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585271837384303,
+      "loss": 1.1211,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015837907236046115,
+      "loss": 0.9289,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015823076644869038,
+      "loss": 0.9935,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015808226649731072,
+      "loss": 0.8153,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015793357300116151,
+      "loss": 0.8336,
+      "step": 576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015778468645572696,
+      "loss": 1.003,
+      "step": 577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015763560735713463,
+      "loss": 0.748,
+      "step": 578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015748633620215368,
+      "loss": 0.8705,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015733687348819334,
+      "loss": 0.8002,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000157187219713301,
+      "loss": 0.8333,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015703737537616085,
+      "loss": 1.3624,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015688734097609204,
+      "loss": 0.8917,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015673711701304704,
+      "loss": 1.1263,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015658670398761002,
+      "loss": 0.9955,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015643610240099508,
+      "loss": 0.8186,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015628531275504477,
+      "loss": 0.7674,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015613433555222828,
+      "loss": 1.0408,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015598317129563967,
+      "loss": 0.8282,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001558318204889964,
+      "loss": 1.1224,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015568028363663761,
+      "loss": 0.8067,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001555285612435223,
+      "loss": 0.874,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015537665381522775,
+      "loss": 0.8293,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015522456185794787,
+      "loss": 0.7221,
+      "step": 594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015507228587849147,
+      "loss": 0.8911,
+      "step": 595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549198263842805,
+      "loss": 1.2569,
+      "step": 596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015476718388334848,
+      "loss": 0.8417,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015461435888433872,
+      "loss": 0.9617,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015446135189650274,
+      "loss": 0.6472,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001543081634296984,
+      "loss": 0.8647,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015415479399438837,
+      "loss": 0.8457,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015400124410163823,
+      "loss": 0.7784,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015384751426311504,
+      "loss": 1.032,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001536936049910854,
+      "loss": 0.8277,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015353951679841386,
+      "loss": 0.9134,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015338525019856114,
+      "loss": 0.7634,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015323080570558246,
+      "loss": 0.8142,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001530761838341259,
+      "loss": 0.8594,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015292138509943052,
+      "loss": 0.8096,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015276641001732476,
+      "loss": 0.8135,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015261125910422476,
+      "loss": 0.7904,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015245593287713247,
+      "loss": 1.1609,
+      "step": 612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015230043185363409,
+      "loss": 1.2596,
+      "step": 613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015214475655189825,
+      "loss": 0.759,
+      "step": 614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015198890749067436,
+      "loss": 0.8159,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015183288518929084,
+      "loss": 0.7402,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015167669016765336,
+      "loss": 0.8436,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515203229462431,
+      "loss": 0.8034,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001513637840461152,
+      "loss": 1.3711,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015120707398889668,
+      "loss": 0.7619,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015105019329678506,
+      "loss": 0.9544,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508931424925464,
+      "loss": 0.74,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015073592209951354,
+      "loss": 0.7063,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505785326415846,
+      "loss": 0.8065,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001504209746432209,
+      "loss": 1.0547,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015026324862944552,
+      "loss": 0.9022,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501053551258413,
+      "loss": 0.6711,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001499472946585492,
+      "loss": 1.3809,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014978906775426666,
+      "loss": 0.8689,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001496306749402456,
+      "loss": 0.8555,
+      "step": 630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001494721167442909,
+      "loss": 1.0124,
+      "step": 631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014931339369475842,
+      "loss": 0.9591,
+      "step": 632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001491545063205535,
+      "loss": 0.7966,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014899545515112888,
+      "loss": 0.7895,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014883624071648327,
+      "loss": 0.8954,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014867686354715927,
+      "loss": 0.9035,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001485173241742419,
+      "loss": 0.8505,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014835762312935655,
+      "loss": 0.7879,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014819776094466742,
+      "loss": 1.3412,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001480377381528757,
+      "loss": 0.9046,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014787755528721766,
+      "loss": 0.7757,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014771721288146306,
+      "loss": 0.804,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014755671146991325,
+      "loss": 1.1423,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014739605158739941,
+      "loss": 0.7792,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001472352337692809,
+      "loss": 0.7975,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014707425855144316,
+      "loss": 1.0372,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014691312647029635,
+      "loss": 0.8304,
+      "step": 647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014675183806277313,
+      "loss": 0.6675,
+      "step": 648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014659039386632724,
+      "loss": 0.8444,
+      "step": 649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014642879441893145,
+      "loss": 1.0982,
+      "step": 650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001462670402590759,
+      "loss": 0.9841,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001461051319257663,
+      "loss": 0.9167,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014594306995852207,
+      "loss": 0.8576,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014578085489737457,
+      "loss": 1.044,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014561848728286532,
+      "loss": 1.1567,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454559676560442,
+      "loss": 0.9446,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014529329655846754,
+      "loss": 1.3375,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014513047453219665,
+      "loss": 1.2952,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014496750211979547,
+      "loss": 0.9073,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001448043798643293,
+      "loss": 0.8585,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014464110830936258,
+      "loss": 0.9106,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014447768799895734,
+      "loss": 0.6596,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014431411947767137,
+      "loss": 0.6683,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001441504032905562,
+      "loss": 1.1954,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014398653998315552,
+      "loss": 0.7304,
+      "step": 665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014382253010150316,
+      "loss": 0.8847,
+      "step": 666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001436583741921215,
+      "loss": 0.8294,
+      "step": 667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014349407280201944,
+      "loss": 0.8156,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014332962647869066,
+      "loss": 0.8993,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014316503577011178,
+      "loss": 0.8424,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014300030122474063,
+      "loss": 1.0077,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014283542339151427,
+      "loss": 0.8677,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014267040281984726,
+      "loss": 0.6751,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014250524005962977,
+      "loss": 0.7614,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001423399356612258,
+      "loss": 0.7395,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014217449017547132,
+      "loss": 1.0816,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014200890415367248,
+      "loss": 0.7824,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001418431781476036,
+      "loss": 0.8916,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001416773127095057,
+      "loss": 1.1028,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014151130839208412,
+      "loss": 0.7553,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001413451657485072,
+      "loss": 0.6484,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014117888533240423,
+      "loss": 0.7217,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001410124676978634,
+      "loss": 0.8028,
+      "step": 683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014084591339943032,
+      "loss": 0.7917,
+      "step": 684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014067922299210595,
+      "loss": 0.7778,
+      "step": 685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001405123970313448,
+      "loss": 0.8221,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014034543607305305,
+      "loss": 1.024,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014017834067358672,
+      "loss": 0.829,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014001111138974996,
+      "loss": 0.8764,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013984374877879283,
+      "loss": 0.8883,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013967625339840982,
+      "loss": 0.9232,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013950862580673783,
+      "loss": 0.7454,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013934086656235424,
+      "loss": 0.7852,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013917297622427526,
+      "loss": 0.7646,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390049553519538,
+      "loss": 0.7732,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013883680450527786,
+      "loss": 0.6403,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013866852424456846,
+      "loss": 0.808,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013850011513057796,
+      "loss": 0.9614,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013833157772448793,
+      "loss": 1.2961,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001381629125879076,
+      "loss": 0.7336,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013799412028287175,
+      "loss": 0.8196,
+      "step": 701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013782520137183897,
+      "loss": 0.7323,
+      "step": 702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013765615641768964,
+      "loss": 0.7116,
+      "step": 703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013748698598372428,
+      "loss": 0.8247,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001373176906336614,
+      "loss": 0.9065,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013714827093163585,
+      "loss": 0.7677,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001369787274421969,
+      "loss": 0.7408,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013680906073030612,
+      "loss": 0.747,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013663927136133589,
+      "loss": 0.7476,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001364693599010672,
+      "loss": 0.927,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013629932691568798,
+      "loss": 0.6891,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000136129172971791,
+      "loss": 0.8696,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001359588986363721,
+      "loss": 0.8205,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013578850447682842,
+      "loss": 0.7585,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013561799106095621,
+      "loss": 0.7709,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013544735895694928,
+      "loss": 0.7173,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013527660873339682,
+      "loss": 0.8324,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013510574095928167,
+      "loss": 0.8532,
+      "step": 718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013493475620397842,
+      "loss": 0.9455,
+      "step": 719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013476365503725134,
+      "loss": 0.7647,
+      "step": 720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013459243802925276,
+      "loss": 0.8985,
+      "step": 721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013442110575052096,
+      "loss": 0.8071,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342496587719783,
+      "loss": 0.8594,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013407809766492935,
+      "loss": 1.3122,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013390642300105908,
+      "loss": 0.8905,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013373463535243073,
+      "loss": 0.7611,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001335627352914842,
+      "loss": 0.7507,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013339072339103376,
+      "loss": 0.7598,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013321860022426653,
+      "loss": 0.9462,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013304636636474033,
+      "loss": 0.744,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013287402238638184,
+      "loss": 0.9065,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013270156886348474,
+      "loss": 1.1016,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013252900637070755,
+      "loss": 0.5736,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013235633548307218,
+      "loss": 0.7388,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013218355677596154,
+      "loss": 0.933,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001320106708251179,
+      "loss": 0.7731,
+      "step": 736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013183767820664088,
+      "loss": 1.036,
+      "step": 737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013166457949698553,
+      "loss": 0.789,
+      "step": 738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013149137527296046,
+      "loss": 0.8962,
+      "step": 739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001313180661117258,
+      "loss": 0.7209,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013114465259079152,
+      "loss": 0.7199,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001309711352880151,
+      "loss": 0.7151,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013079751478160007,
+      "loss": 0.8211,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013062379165009372,
+      "loss": 1.0261,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013044996647238542,
+      "loss": 0.8577,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013027603982770447,
+      "loss": 0.8477,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013010201229561838,
+      "loss": 0.7291,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012992788445603083,
+      "loss": 0.7069,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012975365688917964,
+      "loss": 0.9096,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001295793301756351,
+      "loss": 0.7728,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012940490489629777,
+      "loss": 0.8063,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012923038163239678,
+      "loss": 0.7136,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012905576096548758,
+      "loss": 0.7207,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001288810434774504,
+      "loss": 0.7587,
+      "step": 754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012870622975048786,
+      "loss": 0.8364,
+      "step": 755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012853132036712357,
+      "loss": 0.571,
+      "step": 756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012835631591019965,
+      "loss": 1.2064,
+      "step": 757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012818121696287515,
+      "loss": 0.8322,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012800602410862393,
+      "loss": 0.8146,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012783073793123278,
+      "loss": 0.7418,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001276553590147995,
+      "loss": 0.7244,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001274798879437309,
+      "loss": 0.9075,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012730432530274087,
+      "loss": 0.9277,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001271286716768484,
+      "loss": 0.7937,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012695292765137577,
+      "loss": 0.6439,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012677709381194637,
+      "loss": 0.7473,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012660117074448293,
+      "loss": 1.3543,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012642515903520552,
+      "loss": 1.0504,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012624905927062955,
+      "loss": 0.7844,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012607287203756387,
+      "loss": 0.9319,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012589659792310885,
+      "loss": 0.8132,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001257202375146543,
+      "loss": 0.7019,
+      "step": 772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012554379139987764,
+      "loss": 0.9469,
+      "step": 773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001253672601667418,
+      "loss": 0.6914,
+      "step": 774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012519064440349342,
+      "loss": 0.9497,
+      "step": 775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012501394469866076,
+      "loss": 0.9676,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012483716164105187,
+      "loss": 0.9162,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012466029581975253,
+      "loss": 0.8334,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244833478241243,
+      "loss": 0.5986,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001243063182438024,
+      "loss": 0.8698,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012412920766869426,
+      "loss": 0.737,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012395201668897692,
+      "loss": 1.1016,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012377474589509547,
+      "loss": 0.7145,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012359739587776091,
+      "loss": 0.8291,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012341996722794825,
+      "loss": 0.9367,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012324246053689453,
+      "loss": 0.9221,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012306487639609687,
+      "loss": 0.6626,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012288721539731044,
+      "loss": 1.0209,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012270947813254645,
+      "loss": 0.8634,
+      "step": 789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001225316651940704,
+      "loss": 0.941,
+      "step": 790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012235377717439983,
+      "loss": 1.2792,
+      "step": 791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012217581466630255,
+      "loss": 0.7631,
+      "step": 792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012199777826279448,
+      "loss": 0.938,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012181966855713798,
+      "loss": 0.828,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012164148614283937,
+      "loss": 0.9448,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012146323161364754,
+      "loss": 0.9555,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012128490556355153,
+      "loss": 0.9826,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001211065085867787,
+      "loss": 0.8737,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012092804127779285,
+      "loss": 0.724,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012074950423129206,
+      "loss": 0.9347,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012057089804220687,
+      "loss": 0.7682,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001203922233056981,
+      "loss": 0.7982,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012021348061715516,
+      "loss": 0.9748,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001200346705721937,
+      "loss": 0.7878,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000119855793766654,
+      "loss": 0.6317,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011967685079659863,
+      "loss": 0.829,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001194978422583108,
+      "loss": 0.8489,
+      "step": 807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011931876874829209,
+      "loss": 0.8484,
+      "step": 808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011913963086326061,
+      "loss": 0.7864,
+      "step": 809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011896042920014907,
+      "loss": 0.8398,
+      "step": 810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011878116435610255,
+      "loss": 0.788,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011860183692847682,
+      "loss": 0.749,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011842244751483609,
+      "loss": 1.0909,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011824299671295116,
+      "loss": 0.9441,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011806348512079741,
+      "loss": 0.8497,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011788391333655273,
+      "loss": 0.7769,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011770428195859567,
+      "loss": 0.6826,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011752459158550322,
+      "loss": 0.8367,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011734484281604916,
+      "loss": 0.7551,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011716503624920166,
+      "loss": 0.8194,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011698517248412166,
+      "loss": 0.7992,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011680525212016054,
+      "loss": 0.9062,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011662527575685842,
+      "loss": 0.8486,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011644524399394194,
+      "loss": 1.0532,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011626515743132233,
+      "loss": 0.7406,
+      "step": 825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011608501666909355,
+      "loss": 0.8136,
+      "step": 826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011590482230752996,
+      "loss": 0.9184,
+      "step": 827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011572457494708475,
+      "loss": 0.6595,
+      "step": 828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011554427518838756,
+      "loss": 0.8759,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011536392363224268,
+      "loss": 0.7572,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011518352087962703,
+      "loss": 0.7624,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011500306753168812,
+      "loss": 0.7308,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011482256418974204,
+      "loss": 0.8867,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011464201145527143,
+      "loss": 1.2239,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011446140992992362,
+      "loss": 0.7615,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011428076021550849,
+      "loss": 0.9746,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001141000629139964,
+      "loss": 0.7348,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001139193186275164,
+      "loss": 0.9122,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011373852795835405,
+      "loss": 0.8045,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001135576915089495,
+      "loss": 0.7014,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011337680988189543,
+      "loss": 1.0354,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011319588367993508,
+      "loss": 0.8719,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001130149135059602,
+      "loss": 0.7174,
+      "step": 843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001128338999630091,
+      "loss": 0.9047,
+      "step": 844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011265284365426454,
+      "loss": 0.9501,
+      "step": 845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011247174518305188,
+      "loss": 0.8435,
+      "step": 846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001122906051528369,
+      "loss": 0.8896,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011210942416722392,
+      "loss": 0.7462,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011192820282995364,
+      "loss": 0.6714,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011174694174490136,
+      "loss": 0.6543,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011156564151607474,
+      "loss": 0.7926,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001113843027476119,
+      "loss": 0.8468,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011120292604377938,
+      "loss": 0.7742,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011102151200897006,
+      "loss": 0.7602,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011084006124770141,
+      "loss": 0.6848,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011065857436461304,
+      "loss": 1.0019,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011047705196446514,
+      "loss": 0.8663,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011029549465213609,
+      "loss": 0.862,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011011390303262071,
+      "loss": 0.7396,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010993227771102805,
+      "loss": 0.8101,
+      "step": 860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010975061929257958,
+      "loss": 0.7328,
+      "step": 861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010956892838260695,
+      "loss": 0.9228,
+      "step": 862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010938720558655016,
+      "loss": 0.7539,
+      "step": 863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010920545150995541,
+      "loss": 0.9107,
+      "step": 864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001090236667584731,
+      "loss": 0.7255,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010884185193785599,
+      "loss": 0.8278,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010866000765395687,
+      "loss": 0.767,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010847813451272686,
+      "loss": 0.7239,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010829623312021307,
+      "loss": 0.7715,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010811430408255692,
+      "loss": 1.265,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010793234800599182,
+      "loss": 0.7945,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010775036549684139,
+      "loss": 0.7773,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010756835716151722,
+      "loss": 0.6833,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010738632360651704,
+      "loss": 0.8564,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010720426543842261,
+      "loss": 0.7287,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010702218326389766,
+      "loss": 1.04,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010684007768968598,
+      "loss": 0.8489,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010665794932260928,
+      "loss": 0.8004,
+      "step": 878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010647579876956526,
+      "loss": 0.8323,
+      "step": 879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010629362663752547,
+      "loss": 0.7133,
+      "step": 880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010611143353353348,
+      "loss": 0.795,
+      "step": 881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010592922006470269,
+      "loss": 0.744,
+      "step": 882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010574698683821434,
+      "loss": 0.8919,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010556473446131554,
+      "loss": 0.7654,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010538246354131718,
+      "loss": 0.8182,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010520017468559202,
+      "loss": 0.8344,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010501786850157247,
+      "loss": 0.8392,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010483554559674879,
+      "loss": 0.812,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010465320657866688,
+      "loss": 0.6947,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010447085205492636,
+      "loss": 0.7484,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010428848263317854,
+      "loss": 0.9025,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010410609892112435,
+      "loss": 0.7653,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010392370152651234,
+      "loss": 0.7782,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010374129105713664,
+      "loss": 0.7214,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010355886812083501,
+      "loss": 0.7799,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010337643332548668,
+      "loss": 0.7359,
+      "step": 896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010319398727901044,
+      "loss": 0.7248,
+      "step": 897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010301153058936252,
+      "loss": 0.7559,
+      "step": 898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010282906386453474,
+      "loss": 0.749,
+      "step": 899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001026465877125522,
+      "loss": 0.9649,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010246410274147151,
+      "loss": 0.6759,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010228160955937867,
+      "loss": 0.7119,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010209910877438696,
+      "loss": 1.3652,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001019166009946351,
+      "loss": 0.8787,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010173408682828503,
+      "loss": 1.0258,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010155156688352006,
+      "loss": 0.8565,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010136904176854266,
+      "loss": 0.8563,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010118651209157261,
+      "loss": 0.861,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010100397846084487,
+      "loss": 0.7176,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001008214414846075,
+      "loss": 0.8259,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010063890177111978,
+      "loss": 0.7398,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010045635992865019,
+      "loss": 0.7487,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010027381656547413,
+      "loss": 0.6734,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010009127228987214,
+      "loss": 0.7231,
+      "step": 914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990872771012788e-05,
+      "loss": 0.7961,
+      "step": 915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97261834345259e-05,
+      "loss": 0.6519,
+      "step": 916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954364007134984e-05,
+      "loss": 0.7494,
+      "step": 917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936109822888023e-05,
+      "loss": 0.8155,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917855851539252e-05,
+      "loss": 0.9403,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899602153915517e-05,
+      "loss": 1.0906,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88134879084274e-05,
+      "loss": 0.8357,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863095823145735e-05,
+      "loss": 0.7773,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844843311647994e-05,
+      "loss": 0.833,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.826591317171497e-05,
+      "loss": 0.634,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808339900536495e-05,
+      "loss": 1.3448,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790089122561309e-05,
+      "loss": 0.6122,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771839044062138e-05,
+      "loss": 0.8677,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753589725852852e-05,
+      "loss": 0.6828,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735341228744782e-05,
+      "loss": 0.8646,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717093613546528e-05,
+      "loss": 0.9494,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698846941063749e-05,
+      "loss": 0.8121,
+      "step": 931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.68060127209896e-05,
+      "loss": 0.6661,
+      "step": 932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662356667451336e-05,
+      "loss": 0.8467,
+      "step": 933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644113187916501e-05,
+      "loss": 0.8629,
+      "step": 934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625870894286338e-05,
+      "loss": 0.835,
+      "step": 935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607629847348769e-05,
+      "loss": 0.799,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589390107887568e-05,
+      "loss": 1.2037,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57115173668215e-05,
+      "loss": 0.7077,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552914794507364e-05,
+      "loss": 0.6415,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534679342133314e-05,
+      "loss": 0.743,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516445440325122e-05,
+      "loss": 0.8041,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.498213149842752e-05,
+      "loss": 0.7966,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479982531440797e-05,
+      "loss": 0.6566,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46175364586828e-05,
+      "loss": 0.7146,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443526553868448e-05,
+      "loss": 1.1755,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425301316178571e-05,
+      "loss": 0.7073,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407077993529735e-05,
+      "loss": 1.1089,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388856646646657e-05,
+      "loss": 0.7497,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370637336247456e-05,
+      "loss": 0.8511,
+      "step": 949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352420123043479e-05,
+      "loss": 0.7657,
+      "step": 950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334205067739075e-05,
+      "loss": 0.8039,
+      "step": 951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315992231031403e-05,
+      "loss": 0.8257,
+      "step": 952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297781673610235e-05,
+      "loss": 0.7441,
+      "step": 953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279573456157741e-05,
+      "loss": 0.8553,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261367639348297e-05,
+      "loss": 1.2339,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24316428384828e-05,
+      "loss": 0.7821,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224963450315865e-05,
+      "loss": 0.6806,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20676519940082e-05,
+      "loss": 1.1895,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18856959174431e-05,
+      "loss": 1.1241,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170376687978694e-05,
+      "loss": 0.8757,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152186548727316e-05,
+      "loss": 1.0129,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133999234604312e-05,
+      "loss": 0.7279,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115814806214402e-05,
+      "loss": 0.6364,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09763332415269e-05,
+      "loss": 0.6729,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079454849004463e-05,
+      "loss": 0.9003,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061279441344985e-05,
+      "loss": 0.6094,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043107161739308e-05,
+      "loss": 1.2343,
+      "step": 967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024938070742047e-05,
+      "loss": 0.781,
+      "step": 968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006772228897198e-05,
+      "loss": 0.7707,
+      "step": 969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988609696737936e-05,
+      "loss": 0.7868,
+      "step": 970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970450534786397e-05,
+      "loss": 0.9101,
+      "step": 971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95229480355349e-05,
+      "loss": 0.8388,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934142563538699e-05,
+      "loss": 0.7094,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915993875229862e-05,
+      "loss": 0.7161,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897848799102995e-05,
+      "loss": 0.8132,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879707395622066e-05,
+      "loss": 0.8077,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861569725238811e-05,
+      "loss": 0.6754,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843435848392527e-05,
+      "loss": 0.7358,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825305825509864e-05,
+      "loss": 1.2141,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807179717004638e-05,
+      "loss": 0.647,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789057583277612e-05,
+      "loss": 0.7078,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770939484716311e-05,
+      "loss": 0.6399,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752825481694813e-05,
+      "loss": 0.8801,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734715634573547e-05,
+      "loss": 0.7671,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716610003699092e-05,
+      "loss": 0.7765,
+      "step": 985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698508649403979e-05,
+      "loss": 0.9701,
+      "step": 986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680411632006493e-05,
+      "loss": 0.6288,
+      "step": 987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66231901181046e-05,
+      "loss": 0.6859,
+      "step": 988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644230849105053e-05,
+      "loss": 0.7826,
+      "step": 989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626147204164599e-05,
+      "loss": 0.9293,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608068137248367e-05,
+      "loss": 0.9309,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589993708600364e-05,
+      "loss": 0.6977,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571923978449156e-05,
+      "loss": 0.6077,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553859007007638e-05,
+      "loss": 0.677,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.535798854472858e-05,
+      "loss": 1.2033,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517743581025799e-05,
+      "loss": 0.6451,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49969324683119e-05,
+      "loss": 0.7049,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4816479120373e-05,
+      "loss": 0.9854,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463607636775734e-05,
+      "loss": 0.6911,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445572481161248e-05,
+      "loss": 0.8871,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.427542505291527e-05,
+      "loss": 0.7191,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409517769247005e-05,
+      "loss": 0.8305,
+      "step": 1002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.391498333090647e-05,
+      "loss": 1.0712,
+      "step": 1003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373484256867766e-05,
+      "loss": 0.7403,
+      "step": 1004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355475600605808e-05,
+      "loss": 0.6185,
+      "step": 1005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337472424314157e-05,
+      "loss": 0.8923,
+      "step": 1006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319474787983945e-05,
+      "loss": 1.1525,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301482751587835e-05,
+      "loss": 0.8489,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283496375079838e-05,
+      "loss": 0.7807,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265515718395089e-05,
+      "loss": 0.7583,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247540841449683e-05,
+      "loss": 0.7964,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22957180414044e-05,
+      "loss": 0.6783,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211608666344732e-05,
+      "loss": 0.7656,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193651487920262e-05,
+      "loss": 0.8755,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175700328704887e-05,
+      "loss": 0.652,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157755248516395e-05,
+      "loss": 0.7741,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139816307152319e-05,
+      "loss": 0.6736,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121883564389747e-05,
+      "loss": 1.0194,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103957079985096e-05,
+      "loss": 0.8974,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08603691367394e-05,
+      "loss": 1.1692,
+      "step": 1020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068123125170794e-05,
+      "loss": 1.1772,
+      "step": 1021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050215774168923e-05,
+      "loss": 1.0947,
+      "step": 1022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032314920340138e-05,
+      "loss": 0.7914,
+      "step": 1023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014420623334602e-05,
+      "loss": 0.7016,
+      "step": 1024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996532942780631e-05,
+      "loss": 0.7949,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978651938284485e-05,
+      "loss": 0.7486,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960777669430188e-05,
+      "loss": 0.6871,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942910195779313e-05,
+      "loss": 0.6712,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925049576870794e-05,
+      "loss": 0.8071,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907195872220719e-05,
+      "loss": 1.3063,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889349141322136e-05,
+      "loss": 0.803,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871509443644852e-05,
+      "loss": 0.8878,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85367683863525e-05,
+      "loss": 0.8877,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.835851385716064e-05,
+      "loss": 0.9689,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818033144286207e-05,
+      "loss": 0.7097,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800222173720553e-05,
+      "loss": 0.9477,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782418533369749e-05,
+      "loss": 0.7715,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764622282560018e-05,
+      "loss": 0.8333,
+      "step": 1038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.746833480592961e-05,
+      "loss": 0.8228,
+      "step": 1039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729052186745358e-05,
+      "loss": 0.7362,
+      "step": 1040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71127846026896e-05,
+      "loss": 0.6737,
+      "step": 1041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.693512360390314e-05,
+      "loss": 1.0346,
+      "step": 1042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675753946310549e-05,
+      "loss": 0.6864,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658003277205176e-05,
+      "loss": 1.2056,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640260412223912e-05,
+      "loss": 0.7851,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622525410490453e-05,
+      "loss": 0.784,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604798331102307e-05,
+      "loss": 0.8944,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587079233130573e-05,
+      "loss": 1.2137,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569368175619759e-05,
+      "loss": 0.7153,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551665217587576e-05,
+      "loss": 0.7135,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.53397041802475e-05,
+      "loss": 0.9567,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516283835894812e-05,
+      "loss": 0.8759,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498605530133926e-05,
+      "loss": 0.6813,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480935559650661e-05,
+      "loss": 0.852,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463273983325823e-05,
+      "loss": 0.696,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445620860012238e-05,
+      "loss": 0.9231,
+      "step": 1056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42797624853457e-05,
+      "loss": 0.8307,
+      "step": 1057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410340207689116e-05,
+      "loss": 0.7858,
+      "step": 1058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392712796243614e-05,
+      "loss": 0.8062,
+      "step": 1059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37509407293705e-05,
+      "loss": 0.7053,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357484096479452e-05,
+      "loss": 0.9195,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339882925551711e-05,
+      "loss": 0.5775,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322290618805367e-05,
+      "loss": 0.7362,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304707234862425e-05,
+      "loss": 0.8341,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28713283231516e-05,
+      "loss": 0.7233,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269567469725913e-05,
+      "loss": 0.9734,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25201120562691e-05,
+      "loss": 0.8685,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23446409852005e-05,
+      "loss": 0.8805,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216926206876723e-05,
+      "loss": 0.7806,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199397589137609e-05,
+      "loss": 0.7547,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181878303712486e-05,
+      "loss": 1.245,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164368408980038e-05,
+      "loss": 0.9564,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146867963287647e-05,
+      "loss": 0.8624,
+      "step": 1073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.129377024951216e-05,
+      "loss": 0.8537,
+      "step": 1074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111895652254968e-05,
+      "loss": 0.7774,
+      "step": 1075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094423903451246e-05,
+      "loss": 0.8802,
+      "step": 1076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076961836760327e-05,
+      "loss": 0.7884,
+      "step": 1077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059509510370224e-05,
+      "loss": 0.8482,
+      "step": 1078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042066982436493e-05,
+      "loss": 0.793,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024634311082039e-05,
+      "loss": 0.7387,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00721155439692e-05,
+      "loss": 0.6628,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.989798770438163e-05,
+      "loss": 0.864,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972396017229555e-05,
+      "loss": 0.931,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95500335276146e-05,
+      "loss": 0.9182,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93762083499063e-05,
+      "loss": 1.0012,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920248521839994e-05,
+      "loss": 0.6599,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902886471198491e-05,
+      "loss": 0.7585,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885534740920849e-05,
+      "loss": 0.8336,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868193388827417e-05,
+      "loss": 0.7118,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850862472703957e-05,
+      "loss": 0.951,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.833542050301446e-05,
+      "loss": 0.7977,
+      "step": 1091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816232179335913e-05,
+      "loss": 1.2463,
+      "step": 1092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798932917488214e-05,
+      "loss": 0.8156,
+      "step": 1093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781644322403849e-05,
+      "loss": 0.6278,
+      "step": 1094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764366451692787e-05,
+      "loss": 0.6757,
+      "step": 1095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74709936292925e-05,
+      "loss": 0.7331,
+      "step": 1096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729843113651533e-05,
+      "loss": 0.8496,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712597761361818e-05,
+      "loss": 0.9914,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695363363525968e-05,
+      "loss": 0.7282,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678139977573348e-05,
+      "loss": 1.2082,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660927660896627e-05,
+      "loss": 0.8302,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643726470851584e-05,
+      "loss": 0.7812,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626536464756928e-05,
+      "loss": 1.3404,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609357699894094e-05,
+      "loss": 0.8111,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592190233507067e-05,
+      "loss": 0.7004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575034122802174e-05,
+      "loss": 0.8852,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557889424947907e-05,
+      "loss": 0.6664,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540756197074726e-05,
+      "loss": 0.8352,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523634496274866e-05,
+      "loss": 0.8103,
+      "step": 1109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506524379602161e-05,
+      "loss": 0.8449,
+      "step": 1110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489425904071831e-05,
+      "loss": 0.8378,
+      "step": 1111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472339126660318e-05,
+      "loss": 0.708,
+      "step": 1112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.455264104305072e-05,
+      "loss": 0.6392,
+      "step": 1113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438200893904382e-05,
+      "loss": 0.8068,
+      "step": 1114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421149552317162e-05,
+      "loss": 0.8159,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404110136362793e-05,
+      "loss": 0.8443,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387082702820904e-05,
+      "loss": 0.6619,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370067308431204e-05,
+      "loss": 0.5855,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35306400989328e-05,
+      "loss": 0.6532,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336072863866414e-05,
+      "loss": 0.69,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.319093926969392e-05,
+      "loss": 0.6479,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302127255780314e-05,
+      "loss": 0.9962,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285172906836416e-05,
+      "loss": 0.7883,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268230936633861e-05,
+      "loss": 0.9452,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251301401627575e-05,
+      "loss": 0.7097,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234384358231035e-05,
+      "loss": 0.7686,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217479862816104e-05,
+      "loss": 0.7341,
+      "step": 1127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200587971712827e-05,
+      "loss": 0.7494,
+      "step": 1128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183708741209241e-05,
+      "loss": 0.791,
+      "step": 1129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166842227551209e-05,
+      "loss": 0.8723,
+      "step": 1130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149988486942207e-05,
+      "loss": 0.6656,
+      "step": 1131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133147575543155e-05,
+      "loss": 0.7851,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.116319549472213e-05,
+      "loss": 0.6059,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099504464804621e-05,
+      "loss": 0.6957,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08270237757248e-05,
+      "loss": 0.7432,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065913343764581e-05,
+      "loss": 0.8713,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0491374193262226e-05,
+      "loss": 0.7453,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032374660159023e-05,
+      "loss": 0.6517,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0156251221207206e-05,
+      "loss": 0.707,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998888861025007e-05,
+      "loss": 0.8519,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982165932641328e-05,
+      "loss": 0.8042,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965456392694697e-05,
+      "loss": 0.7355,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948760296865523e-05,
+      "loss": 0.7415,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9320777007894056e-05,
+      "loss": 0.7645,
+      "step": 1144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.915408660056969e-05,
+      "loss": 0.7685,
+      "step": 1145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8987532302136617e-05,
+      "loss": 0.7363,
+      "step": 1146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882111466759581e-05,
+      "loss": 0.76,
+      "step": 1147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865483425149281e-05,
+      "loss": 0.8771,
+      "step": 1148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84886916079159e-05,
+      "loss": 0.8025,
+      "step": 1149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832268729049435e-05,
+      "loss": 0.7262,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815682185239639e-05,
+      "loss": 0.9002,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799109584632751e-05,
+      "loss": 0.6659,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782550982452866e-05,
+      "loss": 0.8896,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76600643387742e-05,
+      "loss": 0.7628,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.749475994037024e-05,
+      "loss": 0.7253,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732959718015276e-05,
+      "loss": 0.7756,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716457660848574e-05,
+      "loss": 0.7273,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699969877525939e-05,
+      "loss": 0.6803,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6834964229888254e-05,
+      "loss": 0.9369,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6670373521309414e-05,
+      "loss": 1.4763,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650592719798059e-05,
+      "loss": 0.8925,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6341625807878515e-05,
+      "loss": 0.741,
+      "step": 1162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617746989849686e-05,
+      "loss": 0.6336,
+      "step": 1163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601346001684453e-05,
+      "loss": 0.8073,
+      "step": 1164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58495967094438e-05,
+      "loss": 0.7732,
+      "step": 1165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.568588052232864e-05,
+      "loss": 0.7729,
+      "step": 1166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552231200104265e-05,
+      "loss": 0.8639,
+      "step": 1167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535889169063746e-05,
+      "loss": 1.1933,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519562013567074e-05,
+      "loss": 0.7513,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5032497880204516e-05,
+      "loss": 0.6787,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486952546780335e-05,
+      "loss": 0.6883,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4706703441532435e-05,
+      "loss": 0.8568,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454403234395584e-05,
+      "loss": 0.7288,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.438151271713472e-05,
+      "loss": 0.796,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421914510262542e-05,
+      "loss": 0.6233,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4056930041477926e-05,
+      "loss": 0.6704,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389486807423374e-05,
+      "loss": 0.7369,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37329597409241e-05,
+      "loss": 0.8298,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3571205581068586e-05,
+      "loss": 0.7091,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3409606133672805e-05,
+      "loss": 0.7377,
+      "step": 1180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.32481619372269e-05,
+      "loss": 0.6684,
+      "step": 1181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.308687352970371e-05,
+      "loss": 0.7677,
+      "step": 1182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2925741448556834e-05,
+      "loss": 0.654,
+      "step": 1183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276476623071913e-05,
+      "loss": 0.8953,
+      "step": 1184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.260394841260059e-05,
+      "loss": 0.6554,
+      "step": 1185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2443288530086785e-05,
+      "loss": 0.6928,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228278711853698e-05,
+      "loss": 0.949,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212244471278234e-05,
+      "loss": 0.6176,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1962261847124314e-05,
+      "loss": 0.7151,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1802239055332593e-05,
+      "loss": 0.8021,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164237687064348e-05,
+      "loss": 0.922,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148267582575812e-05,
+      "loss": 1.1719,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132313645284072e-05,
+      "loss": 0.7822,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1163759283516755e-05,
+      "loss": 0.8073,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1004544848871113e-05,
+      "loss": 0.9652,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0845493679446524e-05,
+      "loss": 0.8506,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0686606305241536e-05,
+      "loss": 0.5911,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052788325570913e-05,
+      "loss": 0.7214,
+      "step": 1198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.036932505975442e-05,
+      "loss": 0.7009,
+      "step": 1199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021093224573339e-05,
+      "loss": 0.7813,
+      "step": 1200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005270534145081e-05,
+      "loss": 0.7713,
+      "step": 1201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9894644874158746e-05,
+      "loss": 0.5524,
+      "step": 1202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973675137055451e-05,
+      "loss": 0.9035,
+      "step": 1203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9579025356779116e-05,
+      "loss": 0.8986,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942146735841541e-05,
+      "loss": 0.8919,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926407790048646e-05,
+      "loss": 0.8009,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910685750745364e-05,
+      "loss": 0.8372,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894980670321496e-05,
+      "loss": 0.88,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8792926011103344e-05,
+      "loss": 0.7772,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863621595388481e-05,
+      "loss": 0.6813,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847967705375689e-05,
+      "loss": 0.6814,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8323309832346666e-05,
+      "loss": 1.1277,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.816711481070918e-05,
+      "loss": 0.7837,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8011092509325665e-05,
+      "loss": 0.6197,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785524344810174e-05,
+      "loss": 0.7878,
+      "step": 1215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7699568146365923e-05,
+      "loss": 0.8079,
+      "step": 1216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7544067122867544e-05,
+      "loss": 0.653,
+      "step": 1217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7388740895775253e-05,
+      "loss": 0.6598,
+      "step": 1218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723358998267524e-05,
+      "loss": 0.9466,
+      "step": 1219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707861490056952e-05,
+      "loss": 0.7171,
+      "step": 1220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692381616587415e-05,
+      "loss": 0.8317,
+      "step": 1221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676919429441758e-05,
+      "loss": 0.7442,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661474980143889e-05,
+      "loss": 0.8542,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646048320158617e-05,
+      "loss": 0.7098,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630639500891463e-05,
+      "loss": 0.7172,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615248573688499e-05,
+      "loss": 0.7384,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599875589836181e-05,
+      "loss": 0.6196,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584520600561165e-05,
+      "loss": 0.6813,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5691836570301605e-05,
+      "loss": 0.8495,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553864810349727e-05,
+      "loss": 0.8749,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538564111566129e-05,
+      "loss": 1.2387,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5232816116651564e-05,
+      "loss": 0.6926,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5080173615719513e-05,
+      "loss": 0.6915,
+      "step": 1233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492771412150855e-05,
+      "loss": 0.6305,
+      "step": 1234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477543814205214e-05,
+      "loss": 0.729,
+      "step": 1235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462334618477227e-05,
+      "loss": 0.8001,
+      "step": 1236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44714387564777e-05,
+      "loss": 0.8624,
+      "step": 1237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4319716363362395e-05,
+      "loss": 0.7888,
+      "step": 1238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416817951100359e-05,
+      "loss": 1.1895,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401682870436039e-05,
+      "loss": 0.7076,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386566444777176e-05,
+      "loss": 0.9692,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371468724495524e-05,
+      "loss": 0.7307,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356389759900495e-05,
+      "loss": 0.7183,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341329601239005e-05,
+      "loss": 0.801,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326288298695301e-05,
+      "loss": 0.6987,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311265902390798e-05,
+      "loss": 0.7158,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296262462383916e-05,
+      "loss": 0.7609,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281278028669902e-05,
+      "loss": 0.7812,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2663126511806695e-05,
+      "loss": 0.7929,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251366379784629e-05,
+      "loss": 0.7697,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236439264286538e-05,
+      "loss": 0.7162,
+      "step": 1251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221531354427306e-05,
+      "loss": 0.6982,
+      "step": 1252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2066426998838516e-05,
+      "loss": 0.8633,
+      "step": 1253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1917733502689304e-05,
+      "loss": 0.7189,
+      "step": 1254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176923355130962e-05,
+      "loss": 1.191,
+      "step": 1255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1620927639538864e-05,
+      "loss": 0.6681,
+      "step": 1256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147281626156973e-05,
+      "loss": 0.6732,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132489991094668e-05,
+      "loss": 0.6049,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117717908056435e-05,
+      "loss": 1.1005,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102965426266578e-05,
+      "loss": 0.8842,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0882325948841e-05,
+      "loss": 0.6485,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0735194630025017e-05,
+      "loss": 0.7956,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058826079649654e-05,
+      "loss": 0.6693,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0441524937876216e-05,
+      "loss": 0.6803,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029498754312495e-05,
+      "loss": 0.6943,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0148649100542314e-05,
+      "loss": 0.8999,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000251009776491e-05,
+      "loss": 0.6775,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9856571021764764e-05,
+      "loss": 0.7986,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971083235884777e-05,
+      "loss": 0.8054,
+      "step": 1269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95652945946519e-05,
+      "loss": 1.2107,
+      "step": 1270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94199582141457e-05,
+      "loss": 0.5779,
+      "step": 1271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9274823701626686e-05,
+      "loss": 0.5977,
+      "step": 1272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912989154071961e-05,
+      "loss": 0.712,
+      "step": 1273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898516221437512e-05,
+      "loss": 0.8499,
+      "step": 1274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884063620486779e-05,
+      "loss": 0.8122,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869631399379476e-05,
+      "loss": 0.7614,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855219606207404e-05,
+      "loss": 0.7019,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8408282889943006e-05,
+      "loss": 0.8213,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826457495695662e-05,
+      "loss": 0.6838,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812107274198601e-05,
+      "loss": 0.9081,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797777672321674e-05,
+      "loss": 0.7745,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.783468737814732e-05,
+      "loss": 0.6556,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769180518358756e-05,
+      "loss": 0.7071,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.754913061565699e-05,
+      "loss": 0.6592,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7406664149783266e-05,
+      "loss": 0.8203,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726440626070058e-05,
+      "loss": 0.8143,
+      "step": 1286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.712235742244819e-05,
+      "loss": 0.7371,
+      "step": 1287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698051810836862e-05,
+      "loss": 0.7613,
+      "step": 1288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683888879110629e-05,
+      "loss": 0.8131,
+      "step": 1289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6697469942605756e-05,
+      "loss": 0.8962,
+      "step": 1290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65562620341104e-05,
+      "loss": 0.752,
+      "step": 1291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.641526553616058e-05,
+      "loss": 0.7432,
+      "step": 1292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627448091859219e-05,
+      "loss": 0.7085,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613390865053512e-05,
+      "loss": 0.7702,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5993549200411594e-05,
+      "loss": 0.9452,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5853403035934796e-05,
+      "loss": 0.6424,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.571347062410708e-05,
+      "loss": 0.6489,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5573752431218555e-05,
+      "loss": 0.6594,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5434248922845505e-05,
+      "loss": 0.933,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5294960563848776e-05,
+      "loss": 0.7299,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515588781837241e-05,
+      "loss": 0.7209,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.501703114984186e-05,
+      "loss": 1.1838,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487839102096258e-05,
+      "loss": 0.882,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473996789371846e-05,
+      "loss": 0.641,
+      "step": 1304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.46017622293703e-05,
+      "loss": 0.7765,
+      "step": 1305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446377448845426e-05,
+      "loss": 0.9985,
+      "step": 1306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4326005130780305e-05,
+      "loss": 1.0759,
+      "step": 1307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4188454615430674e-05,
+      "loss": 0.6849,
+      "step": 1308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.405112340075845e-05,
+      "loss": 0.6083,
+      "step": 1309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3914011944385884e-05,
+      "loss": 1.2482,
+      "step": 1310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.377712070320294e-05,
+      "loss": 0.7582,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.364045013336577e-05,
+      "loss": 0.7463,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.350400069029518e-05,
+      "loss": 0.7687,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.336777282867523e-05,
+      "loss": 1.123,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323176700245146e-05,
+      "loss": 0.8186,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3095983664829645e-05,
+      "loss": 0.7585,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296042326827408e-05,
+      "loss": 0.6138,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282508626450628e-05,
+      "loss": 1.3862,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2689973104503266e-05,
+      "loss": 0.9298,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255508423849619e-05,
+      "loss": 0.8358,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24204201159688e-05,
+      "loss": 0.7221,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228598118565588e-05,
+      "loss": 0.5925,
+      "step": 1322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2151767895541964e-05,
+      "loss": 0.7204,
+      "step": 1323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201778069285958e-05,
+      "loss": 0.6297,
+      "step": 1324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1884020024087845e-05,
+      "loss": 0.8329,
+      "step": 1325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.175048633495108e-05,
+      "loss": 0.5306,
+      "step": 1326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1617180070417276e-05,
+      "loss": 0.6373,
+      "step": 1327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148410167469652e-05,
+      "loss": 1.2421,
+      "step": 1328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1351251591239614e-05,
+      "loss": 0.7341,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1218630262736584e-05,
+      "loss": 0.7945,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.108623813111509e-05,
+      "loss": 1.1648,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0954075637539216e-05,
+      "loss": 0.7461,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0822143222407706e-05,
+      "loss": 0.6602,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069044132535267e-05,
+      "loss": 0.8502,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055897038523802e-05,
+      "loss": 0.7536,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0427730840158162e-05,
+      "loss": 0.8572,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029672312743638e-05,
+      "loss": 0.7058,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.016594768362341e-05,
+      "loss": 0.5814,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0035404944496026e-05,
+      "loss": 0.9938,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9905095345055567e-05,
+      "loss": 0.6144,
+      "step": 1340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9775019319526565e-05,
+      "loss": 0.6979,
+      "step": 1341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9645177301355108e-05,
+      "loss": 0.5378,
+      "step": 1342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9515569723207594e-05,
+      "loss": 0.8483,
+      "step": 1343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9386197016969175e-05,
+      "loss": 0.7246,
+      "step": 1344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9257059613742343e-05,
+      "loss": 0.6906,
+      "step": 1345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9128157943845536e-05,
+      "loss": 0.7399,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8999492436811625e-05,
+      "loss": 0.7527,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.887106352138653e-05,
+      "loss": 0.8131,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8742871625527868e-05,
+      "loss": 0.7536,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8614917176403365e-05,
+      "loss": 0.6283,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848720060038952e-05,
+      "loss": 0.6545,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8359722323070203e-05,
+      "loss": 0.7307,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823248276923517e-05,
+      "loss": 0.6074,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.810548236287879e-05,
+      "loss": 0.8015,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7978721527198437e-05,
+      "loss": 0.9709,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7852200684593198e-05,
+      "loss": 0.6414,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.772592025666245e-05,
+      "loss": 0.6862,
+      "step": 1357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7599880664204425e-05,
+      "loss": 0.7695,
+      "step": 1358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7474082327214912e-05,
+      "loss": 0.6348,
+      "step": 1359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.734852566488567e-05,
+      "loss": 0.611,
+      "step": 1360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7223211095603207e-05,
+      "loss": 0.9686,
+      "step": 1361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7098139036947246e-05,
+      "loss": 0.6882,
+      "step": 1362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6973309905689514e-05,
+      "loss": 0.6687,
+      "step": 1363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6848724117792147e-05,
+      "loss": 0.7788,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672438208840644e-05,
+      "loss": 0.6981,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6600284231871396e-05,
+      "loss": 0.5941,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6476430961712394e-05,
+      "loss": 0.7236,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.63528226906398e-05,
+      "loss": 1.2892,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6229459830547563e-05,
+      "loss": 0.6632,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.610634279251185e-05,
+      "loss": 0.8121,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5983471986789687e-05,
+      "loss": 0.8787,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5860847822817658e-05,
+      "loss": 0.7232,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5738470709210406e-05,
+      "loss": 0.7711,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.561634105375933e-05,
+      "loss": 0.7352,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5494459263431257e-05,
+      "loss": 0.7158,
+      "step": 1375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5372825744367123e-05,
+      "loss": 0.9452,
+      "step": 1376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5251440901880475e-05,
+      "loss": 0.8272,
+      "step": 1377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5130305140456246e-05,
+      "loss": 0.6915,
+      "step": 1378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5009418863749357e-05,
+      "loss": 0.6645,
+      "step": 1379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488878247458336e-05,
+      "loss": 1.2046,
+      "step": 1380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.476839637494922e-05,
+      "loss": 0.5982,
+      "step": 1381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4648260966003746e-05,
+      "loss": 0.7337,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4528376648068464e-05,
+      "loss": 0.8128,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4408743820628165e-05,
+      "loss": 1.1632,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428936288232958e-05,
+      "loss": 0.6012,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4170234230980204e-05,
+      "loss": 0.7624,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4051358263546743e-05,
+      "loss": 0.6402,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3932735376153848e-05,
+      "loss": 0.7716,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3814365964082995e-05,
+      "loss": 0.8992,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3696250421770917e-05,
+      "loss": 0.8011,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35783891428084e-05,
+      "loss": 1.3291,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3460782519938972e-05,
+      "loss": 0.624,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3343430945057552e-05,
+      "loss": 0.6847,
+      "step": 1393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3226334809209248e-05,
+      "loss": 0.6807,
+      "step": 1394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3109494502587924e-05,
+      "loss": 0.8035,
+      "step": 1395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2992910414534962e-05,
+      "loss": 0.6652,
+      "step": 1396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287658293353795e-05,
+      "loss": 0.764,
+      "step": 1397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276051244722941e-05,
+      "loss": 0.8451,
+      "step": 1398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.264469934238552e-05,
+      "loss": 0.7084,
+      "step": 1399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2529144004924763e-05,
+      "loss": 0.6521,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.241384681990668e-05,
+      "loss": 0.6879,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229880817153057e-05,
+      "loss": 0.7663,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2184028443134275e-05,
+      "loss": 0.7211,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206950801719281e-05,
+      "loss": 0.6841,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1955247275317114e-05,
+      "loss": 0.6492,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1841246598252828e-05,
+      "loss": 0.8999,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1727506365878946e-05,
+      "loss": 0.851,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16140269572067e-05,
+      "loss": 0.6837,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1500808750378053e-05,
+      "loss": 0.6737,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1387852122664653e-05,
+      "loss": 0.7889,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1275157450466465e-05,
+      "loss": 0.6176,
+      "step": 1411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1162725109310644e-05,
+      "loss": 0.9195,
+      "step": 1412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1050555473850077e-05,
+      "loss": 0.7529,
+      "step": 1413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09386489178623e-05,
+      "loss": 0.6822,
+      "step": 1414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082700581424821e-05,
+      "loss": 0.7278,
+      "step": 1415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0715626535030773e-05,
+      "loss": 1.0093,
+      "step": 1416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0604511451353912e-05,
+      "loss": 0.7962,
+      "step": 1417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0493660933481086e-05,
+      "loss": 0.7193,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038307535079419e-05,
+      "loss": 0.6861,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0272755071792294e-05,
+      "loss": 0.7653,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0162700464090435e-05,
+      "loss": 0.7624,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0052911894418334e-05,
+      "loss": 0.7622,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9943389728619187e-05,
+      "loss": 0.7902,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9834134331648514e-05,
+      "loss": 0.8742,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972514606757282e-05,
+      "loss": 0.6926,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9616425299568576e-05,
+      "loss": 0.8748,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9507972389920782e-05,
+      "loss": 0.7248,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939978770002191e-05,
+      "loss": 0.776,
+      "step": 1428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9291871590370636e-05,
+      "loss": 0.8257,
+      "step": 1429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9184224420570683e-05,
+      "loss": 0.7844,
+      "step": 1430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9076846549329587e-05,
+      "loss": 0.9173,
+      "step": 1431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8969738334457498e-05,
+      "loss": 0.7085,
+      "step": 1432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8862900132866e-05,
+      "loss": 0.6768,
+      "step": 1433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875633230056699e-05,
+      "loss": 0.7307,
+      "step": 1434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8650035192671378e-05,
+      "loss": 0.7171,
+      "step": 1435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.854400916338791e-05,
+      "loss": 0.7878,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8438254566022107e-05,
+      "loss": 0.7052,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8332771752974942e-05,
+      "loss": 0.7995,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8227561075741806e-05,
+      "loss": 0.6204,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8122622884911212e-05,
+      "loss": 0.8637,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8017957530163687e-05,
+      "loss": 0.7789,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7913565360270612e-05,
+      "loss": 0.5202,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7809446723093015e-05,
+      "loss": 0.7306,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7705601965580532e-05,
+      "loss": 0.7822,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.760203143377005e-05,
+      "loss": 0.7891,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7498735472784733e-05,
+      "loss": 0.6841,
+      "step": 1446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7395714426832775e-05,
+      "loss": 0.6354,
+      "step": 1447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.729296863920634e-05,
+      "loss": 0.7875,
+      "step": 1448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.719049845228029e-05,
+      "loss": 0.6412,
+      "step": 1449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088304207511174e-05,
+      "loss": 0.8256,
+      "step": 1450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6986386245435947e-05,
+      "loss": 0.654,
+      "step": 1451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6884744905671034e-05,
+      "loss": 0.8571,
+      "step": 1452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.678338052691101e-05,
+      "loss": 0.7086,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6682293446927565e-05,
+      "loss": 0.717,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.658148400256836e-05,
+      "loss": 1.041,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6480952529755888e-05,
+      "loss": 1.132,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.638069936348643e-05,
+      "loss": 0.5997,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6280724837828796e-05,
+      "loss": 0.8224,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6181029285923367e-05,
+      "loss": 0.8324,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6081613039980827e-05,
+      "loss": 0.6056,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5982476431281258e-05,
+      "loss": 0.7354,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5883619790172832e-05,
+      "loss": 0.669,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5785043446070823e-05,
+      "loss": 0.6532,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5686747727456485e-05,
+      "loss": 0.7035,
+      "step": 1464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.558873296187594e-05,
+      "loss": 0.7205,
+      "step": 1465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5490999475939172e-05,
+      "loss": 0.5845,
+      "step": 1466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539354759531878e-05,
+      "loss": 0.5977,
+      "step": 1467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5296377644749037e-05,
+      "loss": 0.6224,
+      "step": 1468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5199489948024747e-05,
+      "loss": 0.6859,
+      "step": 1469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5102884828000119e-05,
+      "loss": 0.6523,
+      "step": 1470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5006562606587882e-05,
+      "loss": 0.6151,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491052360475792e-05,
+      "loss": 0.8936,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4814768142536395e-05,
+      "loss": 0.8074,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4719296539004746e-05,
+      "loss": 0.6527,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4624109112298413e-05,
+      "loss": 0.9028,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4529206179605936e-05,
+      "loss": 0.7406,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.443458805716782e-05,
+      "loss": 0.7372,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4340255060275509e-05,
+      "loss": 0.7179,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4246207503270404e-05,
+      "loss": 1.0827,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4152445699542671e-05,
+      "loss": 0.6431,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40589699615303e-05,
+      "loss": 0.6615,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396578060071806e-05,
+      "loss": 0.839,
+      "step": 1482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3872877927636363e-05,
+      "loss": 1.2344,
+      "step": 1483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3780262251860433e-05,
+      "loss": 0.6666,
+      "step": 1484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.368793388200904e-05,
+      "loss": 1.4326,
+      "step": 1485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3595893125743609e-05,
+      "loss": 0.6695,
+      "step": 1486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3504140289767153e-05,
+      "loss": 0.8144,
+      "step": 1487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3412675679823317e-05,
+      "loss": 0.6141,
+      "step": 1488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3321499600695243e-05,
+      "loss": 1.0655,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.323061235620463e-05,
+      "loss": 1.1028,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.314001424921072e-05,
+      "loss": 0.6158,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3049705581609261e-05,
+      "loss": 0.7058,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2959686654331526e-05,
+      "loss": 0.7304,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.286995776734331e-05,
+      "loss": 1.0598,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2780519219643894e-05,
+      "loss": 1.0134,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2691371309265066e-05,
+      "loss": 0.7382,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260251433327022e-05,
+      "loss": 0.6257,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2513948587753188e-05,
+      "loss": 0.9225,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2425674367837392e-05,
+      "loss": 0.8195,
+      "step": 1499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.233769196767478e-05,
+      "loss": 0.7473,
+      "step": 1500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2250001680444967e-05,
+      "loss": 0.7794,
+      "step": 1501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2162603798354088e-05,
+      "loss": 0.6547,
+      "step": 1502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2075498612633951e-05,
+      "loss": 0.6557,
+      "step": 1503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1988686413541007e-05,
+      "loss": 0.7643,
+      "step": 1504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1902167490355387e-05,
+      "loss": 0.8573,
+      "step": 1505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1815942131380043e-05,
+      "loss": 0.7683,
+      "step": 1506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.173001062393958e-05,
+      "loss": 0.7734,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1644373254379482e-05,
+      "loss": 0.666,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1559030308065078e-05,
+      "loss": 0.7856,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1473982069380573e-05,
+      "loss": 0.8065,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1389228821728203e-05,
+      "loss": 0.8726,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1304770847527169e-05,
+      "loss": 0.7335,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1220608428212764e-05,
+      "loss": 0.7631,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1136741844235343e-05,
+      "loss": 0.7019,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1053171375059601e-05,
+      "loss": 0.7676,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.09698972991634e-05,
+      "loss": 0.6385,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0886919894036995e-05,
+      "loss": 0.65,
+      "step": 1517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0804239436181995e-05,
+      "loss": 0.7539,
+      "step": 1518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0721856201110603e-05,
+      "loss": 0.6454,
+      "step": 1519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0639770463344534e-05,
+      "loss": 0.7002,
+      "step": 1520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0557982496414176e-05,
+      "loss": 0.6548,
+      "step": 1521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0476492572857666e-05,
+      "loss": 0.6824,
+      "step": 1522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.039530096421999e-05,
+      "loss": 0.8705,
+      "step": 1523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0314407941052095e-05,
+      "loss": 1.1379,
+      "step": 1524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0233813772909962e-05,
+      "loss": 1.0184,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0153518728353661e-05,
+      "loss": 0.7102,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0073523074946567e-05,
+      "loss": 0.6832,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.99382707925437e-06,
+      "loss": 0.7868,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.914431006844261e-06,
+      "loss": 0.8014,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.835335122283973e-06,
+      "loss": 0.7465,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.756539689140965e-06,
+      "loss": 0.9276,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.67804496998147e-06,
+      "loss": 0.788,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59985122636976e-06,
+      "loss": 0.699,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.521958718867108e-06,
+      "loss": 1.2395,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.444367707030966e-06,
+      "loss": 0.6367,
+      "step": 1535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.36707844941417e-06,
+      "loss": 0.7273,
+      "step": 1536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.290091203564067e-06,
+      "loss": 0.7883,
+      "step": 1537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.21340622602158e-06,
+      "loss": 0.6715,
+      "step": 1538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.137023772320396e-06,
+      "loss": 1.2498,
+      "step": 1539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.060944096986123e-06,
+      "loss": 0.6353,
+      "step": 1540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.985167453535436e-06,
+      "loss": 0.6712,
+      "step": 1541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.909694094475252e-06,
+      "loss": 0.7074,
+      "step": 1542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.834524271301846e-06,
+      "loss": 0.7446,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.759658234500023e-06,
+      "loss": 0.933,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.685096233542311e-06,
+      "loss": 0.6554,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.610838516888142e-06,
+      "loss": 0.614,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.536885331982946e-06,
+      "loss": 0.658,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463236925257401e-06,
+      "loss": 0.642,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.389893542126592e-06,
+      "loss": 0.7048,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.316855426989145e-06,
+      "loss": 0.7164,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.24412282322653e-06,
+      "loss": 0.6432,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.171695973202143e-06,
+      "loss": 0.6439,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.099575118260493e-06,
+      "loss": 0.7891,
+      "step": 1553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.0277604987265e-06,
+      "loss": 1.3786,
+      "step": 1554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.95625235390457e-06,
+      "loss": 0.7763,
+      "step": 1555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.885050922077908e-06,
+      "loss": 0.6437,
+      "step": 1556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.814156440507658e-06,
+      "loss": 0.6883,
+      "step": 1557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.743569145432084e-06,
+      "loss": 0.5832,
+      "step": 1558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.673289272065908e-06,
+      "loss": 1.1491,
+      "step": 1559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.603317054599401e-06,
+      "loss": 0.7369,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5336527261976376e-06,
+      "loss": 0.591,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.464296518999736e-06,
+      "loss": 0.752,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.395248664118071e-06,
+      "loss": 0.6564,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.326509391637559e-06,
+      "loss": 0.711,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.258078930614787e-06,
+      "loss": 0.6925,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.189957509077305e-06,
+      "loss": 0.7056,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.122145354022891e-06,
+      "loss": 0.7495,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0546426914187334e-06,
+      "loss": 0.6962,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.987449746200792e-06,
+      "loss": 0.8689,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.920566742272883e-06,
+      "loss": 0.6532,
+      "step": 1570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8539939025060506e-06,
+      "loss": 0.6556,
+      "step": 1571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7877314487377996e-06,
+      "loss": 0.6364,
+      "step": 1572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7217796017713605e-06,
+      "loss": 0.6829,
+      "step": 1573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.656138581374938e-06,
+      "loss": 0.7143,
+      "step": 1574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590808606280951e-06,
+      "loss": 0.8128,
+      "step": 1575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.525789894185385e-06,
+      "loss": 0.6871,
+      "step": 1576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.461082661746976e-06,
+      "loss": 0.9848,
+      "step": 1577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.396687124586576e-06,
+      "loss": 0.676,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.332603497286338e-06,
+      "loss": 0.7569,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.268831993389102e-06,
+      "loss": 0.7303,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.205372825397593e-06,
+      "loss": 0.7696,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142226204773804e-06,
+      "loss": 0.796,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.079392341938184e-06,
+      "loss": 0.6206,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0168714462690345e-06,
+      "loss": 0.6779,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.954663726101739e-06,
+      "loss": 0.6859,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892769388728137e-06,
+      "loss": 0.7141,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83118864039579e-06,
+      "loss": 0.8449,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.769921686307267e-06,
+      "loss": 0.7929,
+      "step": 1588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.708968730619535e-06,
+      "loss": 0.6283,
+      "step": 1589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648329976443212e-06,
+      "loss": 0.6658,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5880056258419415e-06,
+      "loss": 0.7919,
+      "step": 1591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5279958798316934e-06,
+      "loss": 0.7154,
+      "step": 1592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.468300938380056e-06,
+      "loss": 0.7585,
+      "step": 1593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.408921000405664e-06,
+      "loss": 0.7413,
+      "step": 1594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349856263777431e-06,
+      "loss": 0.7866,
+      "step": 1595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.291106925314004e-06,
+      "loss": 0.6647,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2326731807829875e-06,
+      "loss": 0.8724,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.174555224900346e-06,
+      "loss": 0.78,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116753251329764e-06,
+      "loss": 0.7836,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.059267452682059e-06,
+      "loss": 0.7018,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0020980205143785e-06,
+      "loss": 0.6307,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.945245145329702e-06,
+      "loss": 0.6927,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888709016576154e-06,
+      "loss": 0.6812,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.832489822646402e-06,
+      "loss": 0.7431,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7765877508769815e-06,
+      "loss": 0.818,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.721002987547696e-06,
+      "loss": 0.7638,
+      "step": 1606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.665735717881015e-06,
+      "loss": 0.6414,
+      "step": 1607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.610786126041399e-06,
+      "loss": 0.6879,
+      "step": 1608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556154395134782e-06,
+      "loss": 0.7139,
+      "step": 1609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5018407072078824e-06,
+      "loss": 0.6976,
+      "step": 1610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.447845243247606e-06,
+      "loss": 1.0527,
+      "step": 1611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.394168183180458e-06,
+      "loss": 0.6635,
+      "step": 1612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.34080970587194e-06,
+      "loss": 0.668,
+      "step": 1613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2877699891260005e-06,
+      "loss": 0.7677,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2350492096843274e-06,
+      "loss": 0.6171,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.18264754322587e-06,
+      "loss": 0.9829,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.130565164366207e-06,
+      "loss": 0.7367,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.078802246656987e-06,
+      "loss": 0.65,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.027358962585293e-06,
+      "loss": 0.9596,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9762354835731315e-06,
+      "loss": 0.7187,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.925431979976824e-06,
+      "loss": 0.6839,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.874948621086483e-06,
+      "loss": 0.6716,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.824785575125412e-06,
+      "loss": 0.6431,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7749430092495208e-06,
+      "loss": 0.6687,
+      "step": 1624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.72542108954681e-06,
+      "loss": 0.7704,
+      "step": 1625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6762199810368214e-06,
+      "loss": 1.033,
+      "step": 1626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.627339847670075e-06,
+      "loss": 0.8748,
+      "step": 1627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5787808523275124e-06,
+      "loss": 0.7874,
+      "step": 1628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5305431568199433e-06,
+      "loss": 0.7576,
+      "step": 1629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482626921887533e-06,
+      "loss": 1.005,
+      "step": 1630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.435032307199304e-06,
+      "loss": 0.8381,
+      "step": 1631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.387759471352503e-06,
+      "loss": 0.7558,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.340808571872156e-06,
+      "loss": 0.652,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2941797652104923e-06,
+      "loss": 0.7443,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.247873206746477e-06,
+      "loss": 0.8704,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.201889050785245e-06,
+      "loss": 0.7245,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.156227450557614e-06,
+      "loss": 0.6882,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.110888558219549e-06,
+      "loss": 0.7225,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0658725248516874e-06,
+      "loss": 0.6717,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.021179500458815e-06,
+      "loss": 1.0969,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9768096339693684e-06,
+      "loss": 0.6257,
+      "step": 1641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.932763073234923e-06,
+      "loss": 0.693,
+      "step": 1642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8890399650297493e-06,
+      "loss": 0.6313,
+      "step": 1643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.84564045505028e-06,
+      "loss": 0.6905,
+      "step": 1644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802564687914633e-06,
+      "loss": 0.7461,
+      "step": 1645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.759812807162132e-06,
+      "loss": 0.6584,
+      "step": 1646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.717384955252833e-06,
+      "loss": 0.8549,
+      "step": 1647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.675281273567032e-06,
+      "loss": 0.6854,
+      "step": 1648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6335019024048447e-06,
+      "loss": 0.9723,
+      "step": 1649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.592046980985674e-06,
+      "loss": 0.6804,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5509166474477763e-06,
+      "loss": 0.7268,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5101110388477956e-06,
+      "loss": 0.8594,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4696302911603187e-06,
+      "loss": 0.6487,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.429474539277432e-06,
+      "loss": 0.772,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3896439170082106e-06,
+      "loss": 0.6503,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.350138557078352e-06,
+      "loss": 0.6615,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.310958591129675e-06,
+      "loss": 0.5987,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2721041497197226e-06,
+      "loss": 0.6036,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2335753623213053e-06,
+      "loss": 0.7292,
+      "step": 1659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1953723573220565e-06,
+      "loss": 0.9554,
+      "step": 1660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1574952620240228e-06,
+      "loss": 0.6425,
+      "step": 1661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1199442026432536e-06,
+      "loss": 0.8448,
+      "step": 1662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.082719304309355e-06,
+      "loss": 0.7427,
+      "step": 1663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.045820691065081e-06,
+      "loss": 0.6028,
+      "step": 1664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.009248485865922e-06,
+      "loss": 0.7538,
+      "step": 1665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.973002810579683e-06,
+      "loss": 0.8006,
+      "step": 1666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9370837859861268e-06,
+      "loss": 0.7291,
+      "step": 1667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9014915317765003e-06,
+      "loss": 0.6773,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8662261665531645e-06,
+      "loss": 0.6836,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8312878078292072e-06,
+      "loss": 0.5893,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7966765720280887e-06,
+      "loss": 0.7676,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7623925744831626e-06,
+      "loss": 0.682,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7284359294373663e-06,
+      "loss": 0.7338,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6948067500427989e-06,
+      "loss": 0.7846,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.661505148360376e-06,
+      "loss": 0.6528,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.628531235359454e-06,
+      "loss": 0.7176,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5958851209174064e-06,
+      "loss": 0.7479,
+      "step": 1677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5635669138193476e-06,
+      "loss": 0.5771,
+      "step": 1678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5315767217576994e-06,
+      "loss": 0.7066,
+      "step": 1679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4999146513318352e-06,
+      "loss": 0.6572,
+      "step": 1680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4685808080477814e-06,
+      "loss": 0.8481,
+      "step": 1681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4375752963178057e-06,
+      "loss": 0.759,
+      "step": 1682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4068982194600844e-06,
+      "loss": 0.7587,
+      "step": 1683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3765496796984024e-06,
+      "loss": 0.69,
+      "step": 1684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3465297781617426e-06,
+      "loss": 0.8593,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3168386148839974e-06,
+      "loss": 0.6419,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2874762888036019e-06,
+      "loss": 0.7105,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2584428977632457e-06,
+      "loss": 0.6939,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2297385385095173e-06,
+      "loss": 0.6837,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2013633066925821e-06,
+      "loss": 0.5718,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1733172968658723e-06,
+      "loss": 0.887,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1456006024857746e-06,
+      "loss": 0.8166,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1182133159113095e-06,
+      "loss": 0.6817,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.09115552840382e-06,
+      "loss": 0.7303,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0644273301266826e-06,
+      "loss": 0.6797,
+      "step": 1695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0380288101450086e-06,
+      "loss": 0.5043,
+      "step": 1696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0119600564253206e-06,
+      "loss": 0.7003,
+      "step": 1697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.862211558352873e-07,
+      "loss": 0.6463,
+      "step": 1698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.608121941434234e-07,
+      "loss": 0.7284,
+      "step": 1699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.357332560188004e-07,
+      "loss": 0.7904,
+      "step": 1700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.109844250307586e-07,
+      "loss": 0.6598,
+      "step": 1701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.865657836486629e-07,
+      "loss": 0.7398,
+      "step": 1702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.624774132415914e-07,
+      "loss": 0.695,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.38719394078058e-07,
+      "loss": 0.7711,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.152918053257796e-07,
+      "loss": 0.6413,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.921947250513983e-07,
+      "loss": 0.7671,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.69428230220226e-07,
+      "loss": 0.8222,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.469923966959447e-07,
+      "loss": 0.7758,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.248872992404399e-07,
+      "loss": 0.6497,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.031130115134565e-07,
+      "loss": 1.1094,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.816696060724215e-07,
+      "loss": 0.6189,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.605571543721767e-07,
+      "loss": 0.7411,
+      "step": 1712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.397757267647353e-07,
+      "loss": 0.6935,
+      "step": 1713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.193253924990483e-07,
+      "loss": 0.6793,
+      "step": 1714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.99206219720816e-07,
+      "loss": 0.6601,
+      "step": 1715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.794182754721766e-07,
+      "loss": 0.6191,
+      "step": 1716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.59961625691563e-07,
+      "loss": 0.6234,
+      "step": 1717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.40836335213446e-07,
+      "loss": 1.0931,
+      "step": 1718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.220424677681024e-07,
+      "loss": 0.734,
+      "step": 1719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.035800859814588e-07,
+      "loss": 0.6727,
+      "step": 1720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.854492513748365e-07,
+      "loss": 0.6986,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.676500243647297e-07,
+      "loss": 0.8588,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5018246426267176e-07,
+      "loss": 0.6938,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.330466292749691e-07,
+      "loss": 0.9598,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1624257650256794e-07,
+      "loss": 0.6753,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9977036194078777e-07,
+      "loss": 0.7669,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.836300404792215e-07,
+      "loss": 0.9529,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6782166590147993e-07,
+      "loss": 0.6985,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5234529088506993e-07,
+      "loss": 0.5871,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3720096700117223e-07,
+      "loss": 0.7948,
+      "step": 1730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2238874471449687e-07,
+      "loss": 0.6262,
+      "step": 1731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0790867338310603e-07,
+      "loss": 0.6591,
+      "step": 1732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.937608012582471e-07,
+      "loss": 0.6008,
+      "step": 1733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7994517548420864e-07,
+      "loss": 0.5115,
+      "step": 1734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.664618420981424e-07,
+      "loss": 1.0844,
+      "step": 1735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.533108460299083e-07,
+      "loss": 0.739,
+      "step": 1736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4049223110196307e-07,
+      "loss": 0.6477,
+      "step": 1737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2800604002913838e-07,
+      "loss": 0.9106,
+      "step": 1738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1585231441859642e-07,
+      "loss": 0.6617,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.040310947696078e-07,
+      "loss": 0.6735,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9254242047346272e-07,
+      "loss": 0.7457,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813863298133267e-07,
+      "loss": 0.8435,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7056285996409627e-07,
+      "loss": 0.6986,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6007204699232113e-07,
+      "loss": 0.6174,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.499139258560378e-07,
+      "loss": 0.776,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.400885304046473e-07,
+      "loss": 0.7365,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.305958933788598e-07,
+      "loss": 0.8178,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2143604641055018e-07,
+      "loss": 1.1905,
+      "step": 1748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1260902002262485e-07,
+      "loss": 0.7967,
+      "step": 1749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0411484362896629e-07,
+      "loss": 0.7972,
+      "step": 1750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.59535455343108e-08,
+      "loss": 0.8899,
+      "step": 1751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.812515293415979e-08,
+      "loss": 0.8666,
+      "step": 1752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.062969191471315e-08,
+      "loss": 0.8428,
+      "step": 1753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.346718745272485e-08,
+      "loss": 0.6602,
+      "step": 1754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.663766341545864e-08,
+      "loss": 0.6813,
+      "step": 1755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.014114256063241e-08,
+      "loss": 0.6162,
+      "step": 1756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.397764653627402e-08,
+      "loss": 0.5943,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.814719588072114e-08,
+      "loss": 0.7395,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.264981002249924e-08,
+      "loss": 1.085,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.748550728028821e-08,
+      "loss": 0.6146,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.265430486284471e-08,
+      "loss": 1.2175,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8156218868957694e-08,
+      "loss": 0.9247,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3991264287381852e-08,
+      "loss": 0.95,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0159454996804272e-08,
+      "loss": 0.8374,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6660803765777834e-08,
+      "loss": 0.7599,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3495322252687902e-08,
+      "loss": 0.5751,
+      "step": 1766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.066302100570793e-08,
+      "loss": 0.7389,
+      "step": 1767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.163909462799435e-09,
+      "loss": 0.6544,
+      "step": 1768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9979959516232035e-09,
+      "loss": 0.7415,
+      "step": 1769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.165287689550379e-09,
+      "loss": 1.0343,
+      "step": 1770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.665790783629163e-09,
+      "loss": 0.6682,
+      "step": 1771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4995102305626064e-09,
+      "loss": 0.743,
+      "step": 1772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.664499166864069e-10,
+      "loss": 0.7826,
+      "step": 1773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.666126179689087e-10,
+      "loss": 0.6494,
+      "step": 1774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5891,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0,
+      "step": 1775,
+      "total_flos": 89872441638912.0,
+      "train_loss": 0.8474402057620841,
+      "train_runtime": 10515.967,
+      "train_samples_per_second": 2.699,
+      "train_steps_per_second": 0.169
+    }
+  ],
+  "max_steps": 1775,
+  "num_train_epochs": 1,
+  "total_flos": 89872441638912.0,
+  "trial_name": null,
+  "trial_params": null
+}