diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7374 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9993876301285977,
+  "eval_steps": 500,
+  "global_step": 1224,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000499999176532081,
+      "loss": 2.1594,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999967061337493,
+      "loss": 2.1003,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999925888212787,
+      "loss": 1.8351,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999868246217933,
+      "loss": 1.9533,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999794135732661,
+      "loss": 1.9901,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0004999703557245193,
+      "loss": 1.8264,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999596511352234,
+      "loss": 1.9285,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999472998758978,
+      "loss": 1.811,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999333020279093,
+      "loss": 1.8615,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999176576834721,
+      "loss": 1.7243,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999003669456472,
+      "loss": 1.9144,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004999003669456472,
+      "loss": 1.6444,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998814299283415,
+      "loss": 1.6947,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998608467563072,
+      "loss": 1.6256,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998386175651409,
+      "loss": 1.6551,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004998147425012829,
+      "loss": 1.8031,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000499789221722016,
+      "loss": 1.7751,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004997620553954645,
+      "loss": 1.6896,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004997332437005932,
+      "loss": 1.675,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499702786827206,
+      "loss": 1.5378,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996706849759452,
+      "loss": 1.6407,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004996369383582895,
+      "loss": 1.5797,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000499601547196553,
+      "loss": 1.5717,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995645117238834,
+      "loss": 1.5683,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004995258321842611,
+      "loss": 1.6953,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994855088324968,
+      "loss": 1.5689,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004994435419342305,
+      "loss": 1.5684,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993999317659293,
+      "loss": 1.592,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993546786148858,
+      "loss": 1.5773,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004993077827792162,
+      "loss": 1.7244,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992592445678582,
+      "loss": 1.5262,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004992090643005691,
+      "loss": 1.5647,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991572423079236,
+      "loss": 1.5838,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004991037789313117,
+      "loss": 1.5977,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004990486745229364,
+      "loss": 1.4692,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989919294458114,
+      "loss": 1.5728,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004989335440737586,
+      "loss": 1.4687,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000498873518791406,
+      "loss": 1.5257,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004988118539941848,
+      "loss": 1.5172,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004987485500883267,
+      "loss": 1.6303,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986836074908615,
+      "loss": 1.5362,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004986170266296148,
+      "loss": 1.4999,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004985488079432037,
+      "loss": 1.3966,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984789518810357,
+      "loss": 1.5622,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004984074589033044,
+      "loss": 1.615,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004983343294809875,
+      "loss": 1.3983,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004982595640958427,
+      "loss": 1.5111,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000498183163240405,
+      "loss": 1.602,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004981051274179839,
+      "loss": 1.487,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004980254571426594,
+      "loss": 1.4709,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004979441529392784,
+      "loss": 1.5622,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004978612153434526,
+      "loss": 1.6075,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004977766449015534,
+      "loss": 1.4113,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976904421707092,
+      "loss": 1.3687,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004976026077188013,
+      "loss": 1.4445,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004975131421244607,
+      "loss": 1.511,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004974220459770638,
+      "loss": 1.6073,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004973293198767285,
+      "loss": 1.5208,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004972349644343108,
+      "loss": 1.4489,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004971389802713999,
+      "loss": 1.5419,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004970413680203149,
+      "loss": 1.4473,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004969421283241003,
+      "loss": 1.3978,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004968412618365216,
+      "loss": 1.4488,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004967387692220614,
+      "loss": 1.4734,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004966346511559149,
+      "loss": 1.4268,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000496528908323985,
+      "loss": 1.5527,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004964215414228786,
+      "loss": 1.4059,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004963125511599009,
+      "loss": 1.4936,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000496201938253052,
+      "loss": 1.5675,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004960897034310215,
+      "loss": 1.4432,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004959758474331833,
+      "loss": 1.3855,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004958603710095916,
+      "loss": 1.4734,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004957432749209755,
+      "loss": 1.3273,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004956245599387339,
+      "loss": 1.4347,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004955042268449307,
+      "loss": 1.4878,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004953822764322895,
+      "loss": 1.5084,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0004952587095041882,
+      "loss": 1.4071,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495133526874654,
+      "loss": 1.4971,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000495006729368358,
+      "loss": 1.5107,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004948783178206096,
+      "loss": 1.4152,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004947482930773512,
+      "loss": 1.6487,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004946166559951523,
+      "loss": 1.5631,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004944834074412042,
+      "loss": 1.3929,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004943485482933144,
+      "loss": 1.3685,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004942120794399002,
+      "loss": 1.493,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004940740017799833,
+      "loss": 1.4697,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000493934316223184,
+      "loss": 1.3028,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000493793023689715,
+      "loss": 1.4564,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004936501251103751,
+      "loss": 1.466,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004935056214265435,
+      "loss": 1.3609,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0004933595135901732,
+      "loss": 1.5026,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004932118025637854,
+      "loss": 1.5779,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004930624893204624,
+      "loss": 1.5107,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004929115748438414,
+      "loss": 1.4332,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004927590601281084,
+      "loss": 1.4105,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004926049461779909,
+      "loss": 1.3919,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004924492340087524,
+      "loss": 1.436,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004922919246461845,
+      "loss": 1.3344,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000492133019126601,
+      "loss": 1.5432,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004919725184968307,
+      "loss": 1.5113,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004918104238142104,
+      "loss": 1.4125,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004916467361465784,
+      "loss": 1.3211,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004914814565722671,
+      "loss": 1.4846,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0004913145861800957,
+      "loss": 1.6591,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004911461260693639,
+      "loss": 1.3527,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004909760773498432,
+      "loss": 1.6009,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004908044411417712,
+      "loss": 1.5183,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004906312185758431,
+      "loss": 1.3647,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004904564107932048,
+      "loss": 1.3824,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004902800189454451,
+      "loss": 1.4677,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004901020441945881,
+      "loss": 1.4092,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004899224877130858,
+      "loss": 1.3776,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004897413506838104,
+      "loss": 1.3135,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004895586343000458,
+      "loss": 1.5419,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000489374339765481,
+      "loss": 1.5139,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0004891884682942012,
+      "loss": 1.4482,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004890010211106795,
+      "loss": 1.4793,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048881199944977,
+      "loss": 1.4989,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000488621404556699,
+      "loss": 1.3618,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004884292376870567,
+      "loss": 1.3491,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004882355001067892,
+      "loss": 1.5396,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004880401930921897,
+      "loss": 1.5266,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004878433179298909,
+      "loss": 1.3162,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004876448759168558,
+      "loss": 1.4951,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004874448683603695,
+      "loss": 1.3594,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0004872432965780303,
+      "loss": 1.2981,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048704016189774147,
+      "loss": 1.3909,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00048683546565770215,
+      "loss": 1.2777,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048662920920639866,
+      "loss": 1.2185,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048642139390259545,
+      "loss": 1.4069,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048621202111532653,
+      "loss": 1.3784,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048600109222388604,
+      "loss": 1.3602,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048578860861781935,
+      "loss": 1.4951,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000485574571696914,
+      "loss": 1.3451,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004853589828711902,
+      "loss": 1.3153,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004851418435608919,
+      "loss": 1.3359,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004849231551964771,
+      "loss": 1.4878,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004847029192186086,
+      "loss": 1.4207,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00048448113707814456,
+      "loss": 1.3223,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004842578102361287,
+      "loss": 1.3065,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000484032940163781,
+      "loss": 1.2831,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004838065283424878,
+      "loss": 1.4204,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000483578576263792,
+      "loss": 1.5151,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004833490854293835,
+      "loss": 1.4167,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048311805735108893,
+      "loss": 1.4353,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004828854935508621,
+      "loss": 1.4536,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048265139556077344,
+      "loss": 1.4906,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004824157649230005,
+      "loss": 1.3239,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004821786031898176,
+      "loss": 1.3811,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004819399119235852,
+      "loss": 1.5675,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0004816996926967401,
+      "loss": 1.2896,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000481457947091785,
+      "loss": 1.367,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00048121467670127796,
+      "loss": 1.3822,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004809698831278217,
+      "loss": 1.5132,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004807235679840536,
+      "loss": 1.2856,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048047573289263445,
+      "loss": 1.4076,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00048022637948623843,
+      "loss": 1.3933,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047997550940754165,
+      "loss": 1.2479,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047972312430921184,
+      "loss": 1.5397,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004794692258538972,
+      "loss": 1.4965,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004792138157142157,
+      "loss": 1.4688,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047895689557274373,
+      "loss": 1.3496,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047869846712200525,
+      "loss": 1.3115,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047843853206446055,
+      "loss": 1.5092,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0004781770921124951,
+      "loss": 1.494,
+      "step": 165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047791414898840816,
+      "loss": 1.3489,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047764970442440164,
+      "loss": 1.391,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047738376016256837,
+      "loss": 1.3822,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047711631795488093,
+      "loss": 1.3905,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004768473795631799,
+      "loss": 1.4504,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047657694675916254,
+      "loss": 1.2375,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004763050213243705,
+      "loss": 1.6112,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047603160505017893,
+      "loss": 1.5109,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047575669973778413,
+      "loss": 1.3239,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047548030719819155,
+      "loss": 1.3143,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047520242925220445,
+      "loss": 1.3333,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0004749230677304114,
+      "loss": 1.3209,
+      "step": 177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004746422244731743,
+      "loss": 1.3362,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004743599013306166,
+      "loss": 1.3013,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004740761001626106,
+      "loss": 1.4733,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047379082283876563,
+      "loss": 1.2937,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004735040712384154,
+      "loss": 1.4254,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047321584725060597,
+      "loss": 1.3224,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004729261527740829,
+      "loss": 1.4175,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004726349897172791,
+      "loss": 1.3974,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047234235999830205,
+      "loss": 1.2746,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004720482655449212,
+      "loss": 1.4266,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0004717527082945554,
+      "loss": 1.3699,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00047145569019425994,
+      "loss": 1.3896,
+      "step": 189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047115721320071385,
+      "loss": 1.3213,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047085727928020696,
+      "loss": 1.2941,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047055589040862676,
+      "loss": 1.3324,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00047025304857144614,
+      "loss": 1.3306,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046994875576370914,
+      "loss": 1.4135,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004696430139900188,
+      "loss": 1.4713,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046933582526452336,
+      "loss": 1.3872,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004690271916109034,
+      "loss": 1.2793,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046871711506235814,
+      "loss": 1.5601,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004684055976615924,
+      "loss": 1.2165,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046809264146080275,
+      "loss": 1.3709,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0004677782485216644,
+      "loss": 1.2739,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00046746242091531733,
+      "loss": 1.2434,
+      "step": 202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046714516072235277,
+      "loss": 1.2875,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004668264700327994,
+      "loss": 1.2939,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046650635094610973,
+      "loss": 1.3576,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046618480557114605,
+      "loss": 1.3555,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004658618360261669,
+      "loss": 1.3495,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004655374444388127,
+      "loss": 1.4241,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004652116329460919,
+      "loss": 1.317,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046488440369436715,
+      "loss": 1.3761,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046455575883934066,
+      "loss": 1.3128,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00046422570054604047,
+      "loss": 1.2786,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004638942309888058,
+      "loss": 1.3711,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0004635613523512731,
+      "loss": 1.3548,
+      "step": 214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004632270668263614,
+      "loss": 1.3002,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004628913766162579,
+      "loss": 1.2433,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004625542839324036,
+      "loss": 1.2455,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004622157909954785,
+      "loss": 1.3323,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046187590003538725,
+      "loss": 1.3168,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046153461329124433,
+      "loss": 1.229,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046119193301135915,
+      "loss": 1.3456,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046084786145322143,
+      "loss": 1.2573,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00046050240088348635,
+      "loss": 1.4324,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004601555535779594,
+      "loss": 1.2507,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004598073218215817,
+      "loss": 1.344,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0004594577079084146,
+      "loss": 1.2898,
+      "step": 226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004591067141416249,
+      "loss": 1.3311,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045875434283346937,
+      "loss": 1.3894,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045840059630527987,
+      "loss": 1.2656,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004580454768874477,
+      "loss": 1.2229,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045768898691940843,
+      "loss": 1.4048,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045733112874962636,
+      "loss": 1.2412,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004569719047355795,
+      "loss": 1.3517,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045661131724374326,
+      "loss": 1.2536,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045624936864957554,
+      "loss": 1.6123,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00045588606133750096,
+      "loss": 1.3931,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0004555213977008946,
+      "loss": 1.3345,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000455155380142067,
+      "loss": 1.2914,
+      "step": 238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045478801107224796,
+      "loss": 1.2884,
+      "step": 239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045478801107224796,
+      "loss": 1.2817,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045441929291157056,
+      "loss": 1.3754,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045404922808905544,
+      "loss": 1.3951,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045367781904259464,
+      "loss": 1.1983,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004533050682189357,
+      "loss": 1.2189,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045293097807366534,
+      "loss": 1.3403,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004525555510711934,
+      "loss": 1.2915,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004521787896847366,
+      "loss": 1.1561,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004518006963963024,
+      "loss": 1.2843,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004514212736966722,
+      "loss": 1.2785,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00045104052408538543,
+      "loss": 1.3026,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0004506584500707228,
+      "loss": 1.3167,
+      "step": 251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004502750541696899,
+      "loss": 1.4252,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044989033890800045,
+      "loss": 1.2327,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044950430682005996,
+      "loss": 1.3087,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044911696044894856,
+      "loss": 1.249,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044872830234640493,
+      "loss": 1.2928,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004483383350728088,
+      "loss": 1.2434,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044794706119716455,
+      "loss": 1.2879,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044755448329708396,
+      "loss": 1.3037,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044716060395876955,
+      "loss": 1.2957,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0004467654257769974,
+      "loss": 1.2358,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044636895135509967,
+      "loss": 1.2179,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00044597118330494845,
+      "loss": 1.1878,
+      "step": 263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044557212424693726,
+      "loss": 1.3805,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044517177680996494,
+      "loss": 1.26,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004447701436314176,
+      "loss": 1.2524,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004443672273571516,
+      "loss": 1.3619,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004439630306414758,
+      "loss": 1.3873,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044355755614713455,
+      "loss": 1.2831,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004431508065452897,
+      "loss": 1.392,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004427427845155033,
+      "loss": 1.2447,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044233349274571976,
+      "loss": 1.3091,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00044192293393224814,
+      "loss": 1.3385,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004415111107797445,
+      "loss": 1.3911,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0004410980260011942,
+      "loss": 1.281,
+      "step": 275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004406836823178937,
+      "loss": 1.3719,
+      "step": 276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00044026808245943284,
+      "loss": 1.3391,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004398512291636768,
+      "loss": 1.2876,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004394331251767481,
+      "loss": 1.245,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043901377325300864,
+      "loss": 1.3237,
+      "step": 280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000438593176155041,
+      "loss": 1.2547,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043817133665363114,
+      "loss": 1.4516,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004377482575277494,
+      "loss": 1.3214,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004373239415645324,
+      "loss": 1.3894,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00043689839155926474,
+      "loss": 1.3069,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004364716103153609,
+      "loss": 1.1784,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0004360436006443459,
+      "loss": 1.3186,
+      "step": 287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043561436536583776,
+      "loss": 1.2797,
+      "step": 288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004351839073075283,
+      "loss": 1.3483,
+      "step": 289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043475222930516476,
+      "loss": 1.4869,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000434319334202531,
+      "loss": 1.2546,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004338852248514289,
+      "loss": 1.2437,
+      "step": 292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004334499041116593,
+      "loss": 1.2201,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004330133748510037,
+      "loss": 1.3446,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00043257563994520467,
+      "loss": 1.2456,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004321367022779476,
+      "loss": 1.3211,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004316965647408411,
+      "loss": 1.247,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004312552302333982,
+      "loss": 1.4084,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004308127016630175,
+      "loss": 1.294,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0004303689819449636,
+      "loss": 1.359,
+      "step": 300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004299240740023482,
+      "loss": 1.3,
+      "step": 301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004294779807661105,
+      "loss": 1.3214,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042903070517499837,
+      "loss": 1.2498,
+      "step": 303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004285822501755485,
+      "loss": 1.2102,
+      "step": 304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004281326187220675,
+      "loss": 1.38,
+      "step": 305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004276818137766118,
+      "loss": 1.1944,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042722983830896884,
+      "loss": 1.2296,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042677669529663686,
+      "loss": 1.3378,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042632238772480585,
+      "loss": 1.299,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004258669185863375,
+      "loss": 1.1825,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00042541029088174543,
+      "loss": 1.2156,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0004249525076191759,
+      "loss": 1.2084,
+      "step": 312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004244935718143876,
+      "loss": 1.4662,
+      "step": 313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004240334864907317,
+      "loss": 1.2716,
+      "step": 314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004235722546791323,
+      "loss": 1.3078,
+      "step": 315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042310987941806616,
+      "loss": 1.3159,
+      "step": 316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042264636375354285,
+      "loss": 1.3981,
+      "step": 317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042218171073908465,
+      "loss": 1.3738,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042171592343570634,
+      "loss": 1.2415,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00042124900491189517,
+      "loss": 1.296,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004207809582435904,
+      "loss": 1.2288,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004203117865141635,
+      "loss": 1.2131,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004198414928143974,
+      "loss": 1.2882,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004193700802424663,
+      "loss": 1.2438,
+      "step": 324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004188975519039151,
+      "loss": 1.3187,
+      "step": 325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004184239109116393,
+      "loss": 1.4187,
+      "step": 326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004179491603858643,
+      "loss": 1.2856,
+      "step": 327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041747330345412453,
+      "loss": 1.4805,
+      "step": 328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004169963432512435,
+      "loss": 1.2335,
+      "step": 329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004165182829193126,
+      "loss": 1.199,
+      "step": 330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004160391256076704,
+      "loss": 1.4151,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004155588744728826,
+      "loss": 1.2355,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00041507753267872017,
+      "loss": 1.2175,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004145951033961395,
+      "loss": 1.2043,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004141115898032607,
+      "loss": 1.344,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004136269950853473,
+      "loss": 1.2807,
+      "step": 336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000413141322434785,
+      "loss": 1.2541,
+      "step": 337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004126545750510605,
+      "loss": 1.1884,
+      "step": 338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041216675614074073,
+      "loss": 1.2217,
+      "step": 339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004116778689174514,
+      "loss": 1.1973,
+      "step": 340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004111879166018561,
+      "loss": 1.1581,
+      "step": 341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004106969024216348,
+      "loss": 1.1267,
+      "step": 342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004102048296114631,
+      "loss": 1.2995,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004097117014129903,
+      "loss": 1.2325,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040921752107481846,
+      "loss": 1.2109,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004087222918524807,
+      "loss": 1.2491,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040822601700842015,
+      "loss": 1.3012,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040772869981196803,
+      "loss": 1.4286,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00040723034353932244,
+      "loss": 1.3211,
+      "step": 349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040673095147352665,
+      "loss": 1.2208,
+      "step": 350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040623052690444726,
+      "loss": 1.2816,
+      "step": 351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004057290731287531,
+      "loss": 1.1285,
+      "step": 352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040522659344989286,
+      "loss": 1.3845,
+      "step": 353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040472309117807367,
+      "loss": 1.1976,
+      "step": 354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040421856963023927,
+      "loss": 1.2608,
+      "step": 355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004037130321300482,
+      "loss": 1.2315,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040320648200785163,
+      "loss": 1.2976,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040269892260067197,
+      "loss": 1.1766,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004021903572521802,
+      "loss": 1.1754,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040168078931267427,
+      "loss": 1.1879,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040117022213905705,
+      "loss": 1.2101,
+      "step": 361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004006586590948141,
+      "loss": 1.2959,
+      "step": 362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0004001461035499914,
+      "loss": 1.2383,
+      "step": 363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003996325588811733,
+      "loss": 1.2306,
+      "step": 364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003991180284714603,
+      "loss": 1.2667,
+      "step": 365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003986025157104467,
+      "loss": 1.1938,
+      "step": 366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003980860239941983,
+      "loss": 1.2576,
+      "step": 367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003975685567252299,
+      "loss": 1.2717,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003970501173124831,
+      "loss": 1.1934,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003965307091713037,
+      "loss": 1.2964,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039601033572341914,
+      "loss": 1.4453,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039548900039691634,
+      "loss": 1.2413,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00039496670662621843,
+      "loss": 1.2551,
+      "step": 373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003944434578520628,
+      "loss": 1.2719,
+      "step": 374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039391925752147805,
+      "loss": 1.2876,
+      "step": 375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039339410908776154,
+      "loss": 1.1293,
+      "step": 376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039286801601045625,
+      "loss": 1.0433,
+      "step": 377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003923409817553284,
+      "loss": 1.2889,
+      "step": 378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003918130097943444,
+      "loss": 1.2326,
+      "step": 379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000391284103605648,
+      "loss": 1.0871,
+      "step": 380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003907542666735374,
+      "loss": 1.1409,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00039022350248844244,
+      "loss": 1.2562,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003896918145469013,
+      "loss": 1.265,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003891592063515376,
+      "loss": 1.2468,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003886256814110376,
+      "loss": 1.3957,
+      "step": 385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038809124324012647,
+      "loss": 1.1821,
+      "step": 386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003875558953595459,
+      "loss": 1.1325,
+      "step": 387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003870196412960302,
+      "loss": 1.137,
+      "step": 388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038648248458228365,
+      "loss": 1.2205,
+      "step": 389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003859444287569567,
+      "loss": 1.3893,
+      "step": 390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038540547736462305,
+      "loss": 1.1998,
+      "step": 391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038486563395575624,
+      "loss": 1.2227,
+      "step": 392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000384324902086706,
+      "loss": 1.231,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003837832853196751,
+      "loss": 1.34,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003832407872226959,
+      "loss": 1.4598,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003826974113696065,
+      "loss": 1.2354,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038215316134002775,
+      "loss": 1.3935,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00038160804071933895,
+      "loss": 1.1612,
+      "step": 398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00038106205309865497,
+      "loss": 1.3368,
+      "step": 399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00038051520207480204,
+      "loss": 1.3126,
+      "step": 400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037996749125029456,
+      "loss": 1.2208,
+      "step": 401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037941892423331067,
+      "loss": 1.2446,
+      "step": 402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037886950463766905,
+      "loss": 1.2942,
+      "step": 403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037831923608280517,
+      "loss": 1.2843,
+      "step": 404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037776812219374693,
+      "loss": 1.2525,
+      "step": 405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037721616660109125,
+      "loss": 1.217,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00037666337294097986,
+      "loss": 1.3488,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003761097448550755,
+      "loss": 1.3377,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000375555285990538,
+      "loss": 1.253,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000375,
+      "loss": 1.2185,
+      "step": 410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037444389054154314,
+      "loss": 1.2407,
+      "step": 411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003738869612786737,
+      "loss": 1.4353,
+      "step": 412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037332921588029887,
+      "loss": 1.2325,
+      "step": 413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003727706580207021,
+      "loss": 1.2247,
+      "step": 414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003722112913795192,
+      "loss": 1.2354,
+      "step": 415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037165111964171404,
+      "loss": 1.1955,
+      "step": 416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003710901464975544,
+      "loss": 1.1976,
+      "step": 417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00037052837564258725,
+      "loss": 1.2496,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003699658107776148,
+      "loss": 1.2556,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036940245560867005,
+      "loss": 1.3031,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003688383138469923,
+      "loss": 1.2516,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00036827338920900255,
+      "loss": 1.2514,
+      "step": 422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003677076854162794,
+      "loss": 1.2642,
+      "step": 423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036714120619553436,
+      "loss": 1.3294,
+      "step": 424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000366573955278587,
+      "loss": 1.2495,
+      "step": 425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036600593640234084,
+      "loss": 1.2672,
+      "step": 426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003654371533087585,
+      "loss": 1.374,
+      "step": 427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003648676097448369,
+      "loss": 1.1804,
+      "step": 428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036429730946258297,
+      "loss": 1.1609,
+      "step": 429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036372625621898866,
+      "loss": 1.147,
+      "step": 430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000363154453776006,
+      "loss": 1.2187,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003625819059005228,
+      "loss": 1.1923,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00036200861636433753,
+      "loss": 1.3538,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003614345889441346,
+      "loss": 1.1795,
+      "step": 434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003608598274214592,
+      "loss": 1.2951,
+      "step": 435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003602843355826928,
+      "loss": 1.3016,
+      "step": 436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035970811721902803,
+      "loss": 1.1018,
+      "step": 437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003591311761264433,
+      "loss": 1.1931,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035855351610567876,
+      "loss": 1.3397,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003579751409622103,
+      "loss": 1.3428,
+      "step": 440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035739605450622473,
+      "loss": 1.251,
+      "step": 441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035681626055259527,
+      "loss": 1.2292,
+      "step": 442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003562357629208555,
+      "loss": 1.2881,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035565456543517487,
+      "loss": 1.2088,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035507267192433334,
+      "loss": 1.1621,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035449008622169586,
+      "loss": 1.302,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00035390681216518766,
+      "loss": 1.1498,
+      "step": 447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003533228535972685,
+      "loss": 1.1941,
+      "step": 448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003527382143649075,
+      "loss": 1.1975,
+      "step": 449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003521528983195579,
+      "loss": 1.1581,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035156690931713167,
+      "loss": 1.1009,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003509802512179737,
+      "loss": 1.2261,
+      "step": 452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00035039292788683734,
+      "loss": 1.2266,
+      "step": 453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003498049431928577,
+      "loss": 1.1586,
+      "step": 454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034921630100952716,
+      "loss": 1.2721,
+      "step": 455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003486270052146694,
+      "loss": 1.2289,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003480370596904137,
+      "loss": 1.199,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034744646832316986,
+      "loss": 1.2142,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00034685523500360227,
+      "loss": 1.1852,
+      "step": 459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034626336362660417,
+      "loss": 1.2193,
+      "step": 460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003456708580912725,
+      "loss": 1.3826,
+      "step": 461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003450777223008815,
+      "loss": 1.3648,
+      "step": 462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034448396016285745,
+      "loss": 1.2778,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034388957558875315,
+      "loss": 1.2513,
+      "step": 464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003432945724942215,
+      "loss": 1.233,
+      "step": 465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003426989547989902,
+      "loss": 1.191,
+      "step": 466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034210272642683584,
+      "loss": 1.194,
+      "step": 467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00034150589130555777,
+      "loss": 1.2335,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003409084533669526,
+      "loss": 1.2887,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003403104165467883,
+      "loss": 1.2632,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0003397117847847777,
+      "loss": 1.1329,
+      "step": 471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033911256202455355,
+      "loss": 1.1585,
+      "step": 472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003385127522136413,
+      "loss": 1.377,
+      "step": 473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003385127522136413,
+      "loss": 1.2704,
+      "step": 474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003379123593034342,
+      "loss": 1.2157,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003373113872491666,
+      "loss": 1.2399,
+      "step": 476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003367098400098881,
+      "loss": 1.3061,
+      "step": 477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033610772154843774,
+      "loss": 1.2058,
+      "step": 478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003355050358314172,
+      "loss": 1.3083,
+      "step": 479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003349017868291653,
+      "loss": 1.3221,
+      "step": 480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003342979785157318,
+      "loss": 1.2649,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00033369361486885085,
+      "loss": 1.3248,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0003330886998699149,
+      "loss": 1.1479,
+      "step": 483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003324832375039489,
+      "loss": 1.319,
+      "step": 484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033187723175958343,
+      "loss": 1.2363,
+      "step": 485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003312706866290288,
+      "loss": 1.2657,
+      "step": 486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033066360610804876,
+      "loss": 1.2452,
+      "step": 487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00033005599419593385,
+      "loss": 1.1893,
+      "step": 488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003294478548954754,
+      "loss": 1.1913,
+      "step": 489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003288391922129392,
+      "loss": 1.0939,
+      "step": 490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032823001015803863,
+      "loss": 1.1156,
+      "step": 491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032762031274390877,
+      "loss": 1.1812,
+      "step": 492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003270101039870797,
+      "loss": 1.3445,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032639938790745014,
+      "loss": 1.2255,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0003257881685282609,
+      "loss": 1.2436,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00032517644987606825,
+      "loss": 1.2174,
+      "step": 496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032456423598071783,
+      "loss": 1.4091,
+      "step": 497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032395153087531767,
+      "loss": 1.1183,
+      "step": 498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032333833859621156,
+      "loss": 1.292,
+      "step": 499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032272466318295303,
+      "loss": 1.1791,
+      "step": 500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032211050867827807,
+      "loss": 1.2849,
+      "step": 501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003214958791280791,
+      "loss": 1.1812,
+      "step": 502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003208807785813777,
+      "loss": 1.0762,
+      "step": 503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00032026521109029853,
+      "loss": 1.1849,
+      "step": 504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003196491807100422,
+      "loss": 1.114,
+      "step": 505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003190326914988587,
+      "loss": 1.1195,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0003184157475180208,
+      "loss": 1.3137,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000317798352831797,
+      "loss": 1.2317,
+      "step": 508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003171805115074251,
+      "loss": 1.2292,
+      "step": 509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031656222761508525,
+      "loss": 1.1485,
+      "step": 510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031594350522787295,
+      "loss": 1.1674,
+      "step": 511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003153243484217725,
+      "loss": 1.2141,
+      "step": 512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031470476127563017,
+      "loss": 1.1566,
+      "step": 513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031408474787112696,
+      "loss": 1.2299,
+      "step": 514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031346431229275194,
+      "loss": 1.2658,
+      "step": 515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003128434586277757,
+      "loss": 1.2442,
+      "step": 516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003122221909662227,
+      "loss": 1.2373,
+      "step": 517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003116005134008447,
+      "loss": 1.2988,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0003109784300270943,
+      "loss": 1.0625,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00031035594494309687,
+      "loss": 1.2094,
+      "step": 520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030973306224962437,
+      "loss": 1.1542,
+      "step": 521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003091097860500683,
+      "loss": 1.1996,
+      "step": 522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003084861204504122,
+      "loss": 1.1386,
+      "step": 523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003078620695592048,
+      "loss": 1.2001,
+      "step": 524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030723763748753355,
+      "loss": 1.2714,
+      "step": 525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030661282834899644,
+      "loss": 1.1684,
+      "step": 526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003059876462596758,
+      "loss": 1.1974,
+      "step": 527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003053620953381107,
+      "loss": 1.1926,
+      "step": 528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003047361797052702,
+      "loss": 1.2026,
+      "step": 529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030410990348452574,
+      "loss": 1.3174,
+      "step": 530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00030348327080162435,
+      "loss": 1.336,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0003028562857846614,
+      "loss": 1.1656,
+      "step": 532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003022289525640531,
+      "loss": 1.1433,
+      "step": 533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003016012752725099,
+      "loss": 1.1748,
+      "step": 534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00030097325804500865,
+      "loss": 1.2172,
+      "step": 535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0003003449050187656,
+      "loss": 1.2286,
+      "step": 536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029971622033320914,
+      "loss": 1.289,
+      "step": 537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029908720812995285,
+      "loss": 1.2428,
+      "step": 538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002984578725527675,
+      "loss": 1.192,
+      "step": 539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002978282177475545,
+      "loss": 1.2345,
+      "step": 540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029719824786231795,
+      "loss": 1.2082,
+      "step": 541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000296567967047138,
+      "loss": 1.1865,
+      "step": 542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002959373794541426,
+      "loss": 1.242,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953064892374812,
+      "loss": 1.1575,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029467530055329655,
+      "loss": 1.2655,
+      "step": 545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002940438175596978,
+      "loss": 1.3675,
+      "step": 546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029341204441673266,
+      "loss": 1.0566,
+      "step": 547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002927799852863606,
+      "loss": 1.3543,
+      "step": 548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002921476443324248,
+      "loss": 1.173,
+      "step": 549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029151502572062534,
+      "loss": 1.2981,
+      "step": 550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029088213361849123,
+      "loss": 1.2544,
+      "step": 551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029024897219535325,
+      "loss": 1.1469,
+      "step": 552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002896155456223163,
+      "loss": 1.3687,
+      "step": 553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002889818580722319,
+      "loss": 1.1021,
+      "step": 554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028834791371967137,
+      "loss": 1.1219,
+      "step": 555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00028771371674089714,
+      "loss": 1.149,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002870792713138362,
+      "loss": 1.2034,
+      "step": 557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002864445816180521,
+      "loss": 1.2111,
+      "step": 558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002858096518347179,
+      "loss": 1.2846,
+      "step": 559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028517448614658785,
+      "loss": 1.1094,
+      "step": 560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002845390887379706,
+      "loss": 1.3161,
+      "step": 561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002839034637947011,
+      "loss": 1.0742,
+      "step": 562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028326761550411347,
+      "loss": 1.252,
+      "step": 563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000282631548055013,
+      "loss": 1.1724,
+      "step": 564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002819952656376487,
+      "loss": 1.1005,
+      "step": 565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028135877244368594,
+      "loss": 1.2622,
+      "step": 566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00028072207266617854,
+      "loss": 1.3489,
+      "step": 567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002800851704995412,
+      "loss": 1.2933,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027944807013952197,
+      "loss": 1.2395,
+      "step": 569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002788107757831744,
+      "loss": 1.1678,
+      "step": 570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002781732916288303,
+      "loss": 1.2043,
+      "step": 571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027753562187607156,
+      "loss": 1.0324,
+      "step": 572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002768977707257029,
+      "loss": 1.2013,
+      "step": 573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027625974237972394,
+      "loss": 1.1292,
+      "step": 574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027562154104130175,
+      "loss": 1.1604,
+      "step": 575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002749831709147429,
+      "loss": 1.2818,
+      "step": 576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027434463620546596,
+      "loss": 1.1859,
+      "step": 577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002737059411199738,
+      "loss": 1.21,
+      "step": 578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002730670898658255,
+      "loss": 1.0954,
+      "step": 579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002724280866516094,
+      "loss": 1.1393,
+      "step": 580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002717889356869146,
+      "loss": 1.2647,
+      "step": 581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027114964118230353,
+      "loss": 1.1508,
+      "step": 582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00027051020734928445,
+      "loss": 1.2546,
+      "step": 583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026987063840028323,
+      "loss": 1.3218,
+      "step": 584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026923093854861596,
+      "loss": 1.3143,
+      "step": 585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002685911120084612,
+      "loss": 1.4008,
+      "step": 586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026795116299483193,
+      "loss": 1.1986,
+      "step": 587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026731109572354796,
+      "loss": 1.2502,
+      "step": 588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026667091441120815,
+      "loss": 1.1647,
+      "step": 589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002660306232751629,
+      "loss": 1.0875,
+      "step": 590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026539022653348577,
+      "loss": 1.0774,
+      "step": 591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000264749728404946,
+      "loss": 1.0957,
+      "step": 592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002641091331089811,
+      "loss": 1.2185,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00026346844486566833,
+      "loss": 1.0557,
+      "step": 594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002628276678956974,
+      "loss": 1.3422,
+      "step": 595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026218680642034264,
+      "loss": 1.17,
+      "step": 596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002615458646614349,
+      "loss": 1.1873,
+      "step": 597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026090484684133404,
+      "loss": 1.2195,
+      "step": 598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00026026375718290085,
+      "loss": 1.1257,
+      "step": 599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002596225999094696,
+      "loss": 1.0427,
+      "step": 600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002589813792448196,
+      "loss": 1.33,
+      "step": 601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025834009941314836,
+      "loss": 1.1926,
+      "step": 602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002576987646390426,
+      "loss": 1.2288,
+      "step": 603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002570573791474513,
+      "loss": 1.1561,
+      "step": 604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002564159471636575,
+      "loss": 1.164,
+      "step": 605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002557744729132503,
+      "loss": 1.2139,
+      "step": 606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025513296062209764,
+      "loss": 1.268,
+      "step": 607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002544914145163177,
+      "loss": 1.2438,
+      "step": 608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025384983882225173,
+      "loss": 1.077,
+      "step": 609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025320823776643566,
+      "loss": 1.3652,
+      "step": 610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025256661557557247,
+      "loss": 1.1944,
+      "step": 611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002519249764765047,
+      "loss": 1.1849,
+      "step": 612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025128332469618594,
+      "loss": 1.1933,
+      "step": 613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002506416644616533,
+      "loss": 1.2885,
+      "step": 614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025,
+      "loss": 1.1539,
+      "step": 615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002493583355383467,
+      "loss": 1.0778,
+      "step": 616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002487166753038141,
+      "loss": 1.1326,
+      "step": 617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002480750235234954,
+      "loss": 1.2101,
+      "step": 618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024743338442442754,
+      "loss": 1.2191,
+      "step": 619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002467917622335644,
+      "loss": 1.323,
+      "step": 620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024615016117774833,
+      "loss": 1.1518,
+      "step": 621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024550858548368234,
+      "loss": 1.2836,
+      "step": 622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002448670393779024,
+      "loss": 1.176,
+      "step": 623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002442255270867498,
+      "loss": 1.2703,
+      "step": 624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024358405283634265,
+      "loss": 1.2077,
+      "step": 625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024294262085254877,
+      "loss": 1.0884,
+      "step": 626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024230123536095747,
+      "loss": 1.0855,
+      "step": 627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024165990058685173,
+      "loss": 1.1083,
+      "step": 628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002410186207551804,
+      "loss": 1.102,
+      "step": 629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024037740009053054,
+      "loss": 1.1881,
+      "step": 630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023973624281709924,
+      "loss": 1.1866,
+      "step": 631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023909515315866605,
+      "loss": 1.1569,
+      "step": 632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023845413533856517,
+      "loss": 1.2431,
+      "step": 633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023781319357965745,
+      "loss": 1.206,
+      "step": 634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023717233210430257,
+      "loss": 1.1618,
+      "step": 635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023653155513433174,
+      "loss": 1.1611,
+      "step": 636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002358908668910189,
+      "loss": 1.1733,
+      "step": 637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023525027159505398,
+      "loss": 1.1068,
+      "step": 638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002346097734665143,
+      "loss": 1.1288,
+      "step": 639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023396937672483709,
+      "loss": 1.2002,
+      "step": 640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00023332908558879178,
+      "loss": 1.1308,
+      "step": 641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002326889042764521,
+      "loss": 1.1557,
+      "step": 642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002320488370051681,
+      "loss": 1.0789,
+      "step": 643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002314088879915388,
+      "loss": 1.1547,
+      "step": 644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00023076906145138407,
+      "loss": 1.1554,
+      "step": 645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002301293615997168,
+      "loss": 1.3557,
+      "step": 646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002294897926507156,
+      "loss": 1.305,
+      "step": 647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022885035881769654,
+      "loss": 1.1036,
+      "step": 648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022821106431308543,
+      "loss": 1.1285,
+      "step": 649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022757191334839063,
+      "loss": 1.0992,
+      "step": 650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022693291013417452,
+      "loss": 1.3331,
+      "step": 651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022629405888002627,
+      "loss": 1.249,
+      "step": 652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022565536379453406,
+      "loss": 1.1711,
+      "step": 653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022501682908525716,
+      "loss": 1.2036,
+      "step": 654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00022437845895869826,
+      "loss": 1.221,
+      "step": 655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002237402576202761,
+      "loss": 1.1568,
+      "step": 656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002231022292742972,
+      "loss": 1.1505,
+      "step": 657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022246437812392848,
+      "loss": 1.2478,
+      "step": 658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022182670837116975,
+      "loss": 1.2124,
+      "step": 659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022118922421682563,
+      "loss": 1.1446,
+      "step": 660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00022055192986047807,
+      "loss": 1.2629,
+      "step": 661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021991482950045884,
+      "loss": 1.2253,
+      "step": 662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002192779273338215,
+      "loss": 1.1678,
+      "step": 663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021864122755631415,
+      "loss": 1.1142,
+      "step": 664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002180047343623514,
+      "loss": 1.1841,
+      "step": 665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021736845194498717,
+      "loss": 1.0987,
+      "step": 666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021673238449588668,
+      "loss": 1.2257,
+      "step": 667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021609653620529895,
+      "loss": 1.247,
+      "step": 668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002154609112620295,
+      "loss": 1.1234,
+      "step": 669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021482551385341224,
+      "loss": 1.1887,
+      "step": 670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021419034816528218,
+      "loss": 1.1832,
+      "step": 671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021355541838194794,
+      "loss": 1.0937,
+      "step": 672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021292072868616383,
+      "loss": 1.1232,
+      "step": 673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002122862832591029,
+      "loss": 1.1895,
+      "step": 674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002116520862803286,
+      "loss": 1.136,
+      "step": 675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021101814192776803,
+      "loss": 1.0739,
+      "step": 676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021038445437768376,
+      "loss": 1.2399,
+      "step": 677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020975102780464676,
+      "loss": 1.2148,
+      "step": 678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00020911786638150873,
+      "loss": 1.1559,
+      "step": 679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020848497427937464,
+      "loss": 1.0613,
+      "step": 680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020785235566757519,
+      "loss": 1.0899,
+      "step": 681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002072200147136395,
+      "loss": 1.2101,
+      "step": 682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020658795558326743,
+      "loss": 1.1734,
+      "step": 683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002059561824403023,
+      "loss": 1.2021,
+      "step": 684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020532469944670347,
+      "loss": 1.2547,
+      "step": 685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020469351076251885,
+      "loss": 1.1917,
+      "step": 686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002040626205458574,
+      "loss": 1.1897,
+      "step": 687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002034320329528621,
+      "loss": 1.0289,
+      "step": 688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020280175213768206,
+      "loss": 1.0023,
+      "step": 689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020217178225244553,
+      "loss": 1.2928,
+      "step": 690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00020154212744723248,
+      "loss": 1.2823,
+      "step": 691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002009127918700472,
+      "loss": 1.0725,
+      "step": 692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002002837796667909,
+      "loss": 1.2401,
+      "step": 693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019965509498123446,
+      "loss": 1.1985,
+      "step": 694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019902674195499144,
+      "loss": 1.2224,
+      "step": 695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019839872472749012,
+      "loss": 1.2232,
+      "step": 696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019777104743594687,
+      "loss": 1.1954,
+      "step": 697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019714371421533866,
+      "loss": 1.1161,
+      "step": 698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019651672919837571,
+      "loss": 1.08,
+      "step": 699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001958900965154743,
+      "loss": 1.3326,
+      "step": 700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001952638202947299,
+      "loss": 1.2142,
+      "step": 701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019463790466188937,
+      "loss": 1.1674,
+      "step": 702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019401235374032428,
+      "loss": 1.0752,
+      "step": 703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019338717165100363,
+      "loss": 1.2541,
+      "step": 704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019276236251246655,
+      "loss": 1.1298,
+      "step": 705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001921379304407952,
+      "loss": 1.2117,
+      "step": 706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019151387954958793,
+      "loss": 1.2544,
+      "step": 707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019089021394993178,
+      "loss": 1.2411,
+      "step": 708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001902669377503756,
+      "loss": 1.1707,
+      "step": 709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018964405505690323,
+      "loss": 1.1408,
+      "step": 710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001890215699729057,
+      "loss": 1.187,
+      "step": 711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018839948659915524,
+      "loss": 1.0687,
+      "step": 712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018777780903377734,
+      "loss": 1.1897,
+      "step": 713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018715654137222432,
+      "loss": 1.0239,
+      "step": 714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018653568770724804,
+      "loss": 1.2316,
+      "step": 715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018591525212887302,
+      "loss": 1.2184,
+      "step": 716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001852952387243698,
+      "loss": 1.1243,
+      "step": 717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018467565157822744,
+      "loss": 1.0837,
+      "step": 718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018405649477212698,
+      "loss": 1.1351,
+      "step": 719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018343777238491476,
+      "loss": 1.1836,
+      "step": 720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001828194884925749,
+      "loss": 1.2087,
+      "step": 721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001822016471682031,
+      "loss": 1.1975,
+      "step": 722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001815842524819793,
+      "loss": 1.0393,
+      "step": 723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018096730850114136,
+      "loss": 1.2075,
+      "step": 724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001803508192899579,
+      "loss": 1.1717,
+      "step": 725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017973478890970145,
+      "loss": 1.2399,
+      "step": 726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001791192214186223,
+      "loss": 1.1503,
+      "step": 727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017850412087192096,
+      "loss": 1.035,
+      "step": 728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017788949132172194,
+      "loss": 1.2183,
+      "step": 729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017727533681704706,
+      "loss": 1.129,
+      "step": 730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017666166140378853,
+      "loss": 1.1666,
+      "step": 731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017604846912468243,
+      "loss": 1.1265,
+      "step": 732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017543576401928218,
+      "loss": 1.1714,
+      "step": 733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017482355012393176,
+      "loss": 1.2929,
+      "step": 734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017421183147173913,
+      "loss": 1.2391,
+      "step": 735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001736006120925499,
+      "loss": 1.1351,
+      "step": 736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017298989601292037,
+      "loss": 1.2163,
+      "step": 737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017237968725609127,
+      "loss": 1.1982,
+      "step": 738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017176998984196146,
+      "loss": 1.1178,
+      "step": 739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017116080778706088,
+      "loss": 1.0615,
+      "step": 740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001705521451045246,
+      "loss": 1.3243,
+      "step": 741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016994400580406624,
+      "loss": 1.1612,
+      "step": 742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016933639389195136,
+      "loss": 1.1716,
+      "step": 743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001687293133709712,
+      "loss": 1.0288,
+      "step": 744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001681227682404166,
+      "loss": 1.0895,
+      "step": 745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001675167624960512,
+      "loss": 1.087,
+      "step": 746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016691130013008512,
+      "loss": 1.2198,
+      "step": 747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016630638513114927,
+      "loss": 1.2312,
+      "step": 748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016570202148426816,
+      "loss": 1.0933,
+      "step": 749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016509821317083466,
+      "loss": 1.2823,
+      "step": 750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016449496416858284,
+      "loss": 1.1326,
+      "step": 751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016389227845156225,
+      "loss": 1.2714,
+      "step": 752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001632901599901118,
+      "loss": 1.0857,
+      "step": 753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016268861275083342,
+      "loss": 1.0933,
+      "step": 754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016208764069656578,
+      "loss": 1.0641,
+      "step": 755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001614872477863587,
+      "loss": 1.1772,
+      "step": 756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016088743797544654,
+      "loss": 1.2117,
+      "step": 757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016028821521522224,
+      "loss": 1.2231,
+      "step": 758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015968958345321176,
+      "loss": 1.2319,
+      "step": 759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015909154663304743,
+      "loss": 1.2054,
+      "step": 760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001584941086944423,
+      "loss": 1.1036,
+      "step": 761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015789727357316425,
+      "loss": 1.263,
+      "step": 762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015730104520100984,
+      "loss": 1.1646,
+      "step": 763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015670542750577848,
+      "loss": 1.1066,
+      "step": 764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015611042441124686,
+      "loss": 1.2196,
+      "step": 765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015551603983714258,
+      "loss": 1.191,
+      "step": 766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001549222776991186,
+      "loss": 1.2003,
+      "step": 767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015432914190872756,
+      "loss": 1.1709,
+      "step": 768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015373663637339587,
+      "loss": 1.2384,
+      "step": 769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001531447649963978,
+      "loss": 1.2677,
+      "step": 770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015255353167683018,
+      "loss": 1.1179,
+      "step": 771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015196294030958639,
+      "loss": 1.1405,
+      "step": 772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015137299478533065,
+      "loss": 1.1502,
+      "step": 773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015078369899047285,
+      "loss": 1.2262,
+      "step": 774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015019505680714232,
+      "loss": 1.111,
+      "step": 775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001496070721131627,
+      "loss": 1.1294,
+      "step": 776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00014901974878202628,
+      "loss": 1.1955,
+      "step": 777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014843309068286847,
+      "loss": 1.0532,
+      "step": 778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014784710168044214,
+      "loss": 1.1781,
+      "step": 779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014726178563509258,
+      "loss": 1.1577,
+      "step": 780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001466771464027316,
+      "loss": 1.1312,
+      "step": 781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014609318783481237,
+      "loss": 1.1241,
+      "step": 782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014550991377830424,
+      "loss": 1.0212,
+      "step": 783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014492732807566672,
+      "loss": 1.0744,
+      "step": 784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001443454345648252,
+      "loss": 1.2294,
+      "step": 785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001437642370791446,
+      "loss": 1.1392,
+      "step": 786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014318373944740485,
+      "loss": 1.3109,
+      "step": 787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014260394549377525,
+      "loss": 1.1492,
+      "step": 788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00014202485903778978,
+      "loss": 1.1972,
+      "step": 789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014144648389432125,
+      "loss": 1.2617,
+      "step": 790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014086882387355658,
+      "loss": 1.0715,
+      "step": 791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00014029188278097198,
+      "loss": 1.2241,
+      "step": 792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013971566441730714,
+      "loss": 1.2166,
+      "step": 793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013914017257854079,
+      "loss": 1.2289,
+      "step": 794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013856541105586545,
+      "loss": 1.1733,
+      "step": 795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001379913836356625,
+      "loss": 1.0924,
+      "step": 796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001374180940994773,
+      "loss": 1.357,
+      "step": 797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013684554622399404,
+      "loss": 1.1423,
+      "step": 798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001362737437810114,
+      "loss": 1.1007,
+      "step": 799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000135702690537417,
+      "loss": 1.207,
+      "step": 800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001351323902551631,
+      "loss": 1.0954,
+      "step": 801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00013456284669124158,
+      "loss": 1.1157,
+      "step": 802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001339940635976592,
+      "loss": 1.0707,
+      "step": 803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013342604472141295,
+      "loss": 1.1977,
+      "step": 804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013285879380446565,
+      "loss": 1.1694,
+      "step": 805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001322923145837206,
+      "loss": 1.1187,
+      "step": 806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001317266107909975,
+      "loss": 1.1737,
+      "step": 807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013116168615300784,
+      "loss": 1.111,
+      "step": 808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00013059754439133004,
+      "loss": 1.1211,
+      "step": 809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001300341892223852,
+      "loss": 1.1414,
+      "step": 810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012947162435741278,
+      "loss": 1.1291,
+      "step": 811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012890985350244563,
+      "loss": 1.1705,
+      "step": 812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00012834888035828597,
+      "loss": 1.1175,
+      "step": 813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001277887086204809,
+      "loss": 1.198,
+      "step": 814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012722934197929804,
+      "loss": 1.0646,
+      "step": 815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012667078411970117,
+      "loss": 1.2132,
+      "step": 816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012611303872132632,
+      "loss": 1.022,
+      "step": 817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012555610945845692,
+      "loss": 1.0786,
+      "step": 818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012500000000000006,
+      "loss": 1.0347,
+      "step": 819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012444471400946207,
+      "loss": 1.1083,
+      "step": 820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001238902551449246,
+      "loss": 1.127,
+      "step": 821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012333662705902017,
+      "loss": 1.254,
+      "step": 822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001227838333989088,
+      "loss": 1.0407,
+      "step": 823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001222318778062531,
+      "loss": 1.0086,
+      "step": 824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012168076391719491,
+      "loss": 1.1618,
+      "step": 825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00012113049536233092,
+      "loss": 1.1283,
+      "step": 826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001205810757666894,
+      "loss": 1.119,
+      "step": 827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00012003250874970548,
+      "loss": 1.193,
+      "step": 828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011948479792519793,
+      "loss": 1.1842,
+      "step": 829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011893794690134499,
+      "loss": 1.1829,
+      "step": 830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011839195928066102,
+      "loss": 1.0626,
+      "step": 831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011784683865997229,
+      "loss": 1.1399,
+      "step": 832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011730258863039347,
+      "loss": 1.2088,
+      "step": 833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011675921277730414,
+      "loss": 1.3295,
+      "step": 834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011621671468032493,
+      "loss": 1.0416,
+      "step": 835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000115675097913294,
+      "loss": 1.287,
+      "step": 836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001151343660442438,
+      "loss": 1.2913,
+      "step": 837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011459452263537695,
+      "loss": 1.1021,
+      "step": 838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011405557124304336,
+      "loss": 1.1548,
+      "step": 839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011351751541771643,
+      "loss": 1.1196,
+      "step": 840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011298035870396985,
+      "loss": 1.2203,
+      "step": 841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011244410464045412,
+      "loss": 1.1008,
+      "step": 842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011190875675987356,
+      "loss": 1.1293,
+      "step": 843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011137431858896247,
+      "loss": 1.2009,
+      "step": 844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011084079364846242,
+      "loss": 1.0981,
+      "step": 845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011030818545309879,
+      "loss": 1.1827,
+      "step": 846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001097764975115576,
+      "loss": 1.3151,
+      "step": 847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010924573332646257,
+      "loss": 1.2196,
+      "step": 848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010871589639435203,
+      "loss": 1.2505,
+      "step": 849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010818699020565564,
+      "loss": 1.272,
+      "step": 850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010765901824467166,
+      "loss": 1.1925,
+      "step": 851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010713198398954382,
+      "loss": 1.0792,
+      "step": 852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010660589091223854,
+      "loss": 1.3603,
+      "step": 853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010608074247852193,
+      "loss": 1.0601,
+      "step": 854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010555654214793722,
+      "loss": 1.272,
+      "step": 855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010503329337378164,
+      "loss": 1.1748,
+      "step": 856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010451099960308375,
+      "loss": 1.2163,
+      "step": 857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010398966427658091,
+      "loss": 1.2352,
+      "step": 858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001034692908286964,
+      "loss": 1.0979,
+      "step": 859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010294988268751693,
+      "loss": 1.1728,
+      "step": 860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010243144327477013,
+      "loss": 1.1381,
+      "step": 861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010191397600580177,
+      "loss": 1.1454,
+      "step": 862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010139748428955334,
+      "loss": 1.0453,
+      "step": 863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001008819715285397,
+      "loss": 1.0805,
+      "step": 864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00010036744111882672,
+      "loss": 1.2874,
+      "step": 865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.985389645000865e-05,
+      "loss": 1.2998,
+      "step": 866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.934134090518593e-05,
+      "loss": 1.1244,
+      "step": 867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.882977786094288e-05,
+      "loss": 1.18,
+      "step": 868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.831921068732572e-05,
+      "loss": 0.981,
+      "step": 869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.780964274781984e-05,
+      "loss": 1.2201,
+      "step": 870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.730107739932806e-05,
+      "loss": 1.2101,
+      "step": 871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.679351799214837e-05,
+      "loss": 1.0001,
+      "step": 872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.628696786995189e-05,
+      "loss": 1.1488,
+      "step": 873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.578143036976073e-05,
+      "loss": 1.0487,
+      "step": 874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.527690882192635e-05,
+      "loss": 1.1362,
+      "step": 875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.477340655010716e-05,
+      "loss": 1.2093,
+      "step": 876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.427092687124691e-05,
+      "loss": 1.2663,
+      "step": 877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.376947309555273e-05,
+      "loss": 1.1629,
+      "step": 878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.326904852647344e-05,
+      "loss": 1.0622,
+      "step": 879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.276965646067753e-05,
+      "loss": 1.2569,
+      "step": 880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.227130018803195e-05,
+      "loss": 1.1841,
+      "step": 881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.177398299157988e-05,
+      "loss": 1.2218,
+      "step": 882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.127770814751932e-05,
+      "loss": 1.2433,
+      "step": 883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.07824789251816e-05,
+      "loss": 1.1107,
+      "step": 884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.028829858700973e-05,
+      "loss": 1.2066,
+      "step": 885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.979517038853685e-05,
+      "loss": 1.1586,
+      "step": 886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.930309757836516e-05,
+      "loss": 1.0483,
+      "step": 887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.881208339814394e-05,
+      "loss": 1.2277,
+      "step": 888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.832213108254864e-05,
+      "loss": 1.1566,
+      "step": 889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.783324385925931e-05,
+      "loss": 1.2038,
+      "step": 890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.734542494893955e-05,
+      "loss": 1.0776,
+      "step": 891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.685867756521501e-05,
+      "loss": 1.3272,
+      "step": 892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.637300491465272e-05,
+      "loss": 1.1463,
+      "step": 893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.588841019673937e-05,
+      "loss": 1.2504,
+      "step": 894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.540489660386064e-05,
+      "loss": 1.1214,
+      "step": 895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.49224673212799e-05,
+      "loss": 1.1741,
+      "step": 896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.444112552711752e-05,
+      "loss": 1.0977,
+      "step": 897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.396087439232966e-05,
+      "loss": 0.9768,
+      "step": 898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.348171708068747e-05,
+      "loss": 1.1513,
+      "step": 899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.300365674875651e-05,
+      "loss": 1.1469,
+      "step": 900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.252669654587549e-05,
+      "loss": 1.2347,
+      "step": 901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.205083961413573e-05,
+      "loss": 1.1628,
+      "step": 902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.15760890883607e-05,
+      "loss": 1.156,
+      "step": 903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.110244809608495e-05,
+      "loss": 1.1286,
+      "step": 904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.062991975753378e-05,
+      "loss": 1.2565,
+      "step": 905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.015850718560256e-05,
+      "loss": 1.2304,
+      "step": 906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.968821348583643e-05,
+      "loss": 1.214,
+      "step": 907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.921904175640957e-05,
+      "loss": 1.1538,
+      "step": 908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.875099508810486e-05,
+      "loss": 1.2075,
+      "step": 909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.828407656429365e-05,
+      "loss": 1.1165,
+      "step": 910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.781828926091536e-05,
+      "loss": 1.1181,
+      "step": 911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.735363624645711e-05,
+      "loss": 1.192,
+      "step": 912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.689012058193385e-05,
+      "loss": 1.2208,
+      "step": 913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.642774532086775e-05,
+      "loss": 1.2446,
+      "step": 914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.596651350926836e-05,
+      "loss": 1.0554,
+      "step": 915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.550642818561249e-05,
+      "loss": 1.0828,
+      "step": 916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.504749238082414e-05,
+      "loss": 1.1136,
+      "step": 917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.458970911825458e-05,
+      "loss": 1.1758,
+      "step": 918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.413308141366254e-05,
+      "loss": 1.2148,
+      "step": 919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.367761227519415e-05,
+      "loss": 1.1484,
+      "step": 920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.322330470336314e-05,
+      "loss": 1.1624,
+      "step": 921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.27701616910312e-05,
+      "loss": 1.098,
+      "step": 922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.231818622338823e-05,
+      "loss": 1.0362,
+      "step": 923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.186738127793257e-05,
+      "loss": 1.2082,
+      "step": 924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.141774982445148e-05,
+      "loss": 1.0154,
+      "step": 925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.096929482500167e-05,
+      "loss": 1.1318,
+      "step": 926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.052201923388955e-05,
+      "loss": 1.0863,
+      "step": 927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.007592599765188e-05,
+      "loss": 1.1101,
+      "step": 928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.963101805503647e-05,
+      "loss": 1.0511,
+      "step": 929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.918729833698259e-05,
+      "loss": 1.0216,
+      "step": 930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.874476976660185e-05,
+      "loss": 1.1037,
+      "step": 931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8303435259159e-05,
+      "loss": 1.2339,
+      "step": 932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.786329772205247e-05,
+      "loss": 1.0232,
+      "step": 933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.742436005479536e-05,
+      "loss": 1.1132,
+      "step": 934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.69866251489964e-05,
+      "loss": 1.2427,
+      "step": 935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.655009588834077e-05,
+      "loss": 1.071,
+      "step": 936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.611477514857114e-05,
+      "loss": 1.0914,
+      "step": 937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.5680665797469e-05,
+      "loss": 1.1928,
+      "step": 938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.524777069483526e-05,
+      "loss": 1.1989,
+      "step": 939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.481609269247171e-05,
+      "loss": 1.2801,
+      "step": 940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.438563463416223e-05,
+      "loss": 1.0442,
+      "step": 941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.395639935565412e-05,
+      "loss": 1.1779,
+      "step": 942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.35283896846392e-05,
+      "loss": 1.1318,
+      "step": 943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.31016084407352e-05,
+      "loss": 1.2038,
+      "step": 944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.267605843546767e-05,
+      "loss": 1.1461,
+      "step": 945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.225174247225066e-05,
+      "loss": 1.327,
+      "step": 946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.182866334636889e-05,
+      "loss": 1.1828,
+      "step": 947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.140682384495902e-05,
+      "loss": 0.9814,
+      "step": 948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.098622674699147e-05,
+      "loss": 1.0257,
+      "step": 949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.0566874823251935e-05,
+      "loss": 1.2595,
+      "step": 950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.01487708363232e-05,
+      "loss": 1.127,
+      "step": 951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9731917540567174e-05,
+      "loss": 1.1989,
+      "step": 952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9316317682106294e-05,
+      "loss": 1.1274,
+      "step": 953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8901973998805814e-05,
+      "loss": 1.0719,
+      "step": 954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.848888922025553e-05,
+      "loss": 1.2299,
+      "step": 955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.807706606775196e-05,
+      "loss": 1.2503,
+      "step": 956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.766650725428027e-05,
+      "loss": 1.1301,
+      "step": 957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.725721548449669e-05,
+      "loss": 1.1631,
+      "step": 958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.684919345471029e-05,
+      "loss": 1.2023,
+      "step": 959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6442443852865485e-05,
+      "loss": 1.1223,
+      "step": 960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.603696935852426e-05,
+      "loss": 1.0044,
+      "step": 961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.56327726428485e-05,
+      "loss": 1.2944,
+      "step": 962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.522985636858238e-05,
+      "loss": 1.0393,
+      "step": 963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4828223190035055e-05,
+      "loss": 1.0775,
+      "step": 964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.442787575306274e-05,
+      "loss": 1.18,
+      "step": 965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.402881669505164e-05,
+      "loss": 1.1477,
+      "step": 966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3631048644900345e-05,
+      "loss": 1.1323,
+      "step": 967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.323457422300271e-05,
+      "loss": 1.2251,
+      "step": 968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.283939604123042e-05,
+      "loss": 1.1488,
+      "step": 969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.244551670291606e-05,
+      "loss": 1.1611,
+      "step": 970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2052938802835515e-05,
+      "loss": 1.2961,
+      "step": 971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1661664927191235e-05,
+      "loss": 1.2177,
+      "step": 972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1271697653595156e-05,
+      "loss": 1.2855,
+      "step": 973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.088303955105153e-05,
+      "loss": 1.1278,
+      "step": 974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.049569317994013e-05,
+      "loss": 1.1728,
+      "step": 975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.01096610919996e-05,
+      "loss": 1.1809,
+      "step": 976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.972494583031015e-05,
+      "loss": 1.166,
+      "step": 977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9341549929277227e-05,
+      "loss": 1.1333,
+      "step": 978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.895947591461455e-05,
+      "loss": 1.0972,
+      "step": 979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.85787263033278e-05,
+      "loss": 1.062,
+      "step": 980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.819930360369762e-05,
+      "loss": 1.1008,
+      "step": 981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.78212103152634e-05,
+      "loss": 1.1797,
+      "step": 982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.744444892880661e-05,
+      "loss": 1.1212,
+      "step": 983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7069021926334696e-05,
+      "loss": 1.065,
+      "step": 984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.669493178106432e-05,
+      "loss": 1.1494,
+      "step": 985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6322180957405366e-05,
+      "loss": 1.255,
+      "step": 986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59507719109446e-05,
+      "loss": 1.0787,
+      "step": 987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.558070708842948e-05,
+      "loss": 1.0694,
+      "step": 988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.521198892775202e-05,
+      "loss": 1.0938,
+      "step": 989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.484461985793298e-05,
+      "loss": 1.1596,
+      "step": 990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.447860229910544e-05,
+      "loss": 1.1586,
+      "step": 991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4113938662499103e-05,
+      "loss": 1.1063,
+      "step": 992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3750631350424455e-05,
+      "loss": 1.2304,
+      "step": 993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.338868275625679e-05,
+      "loss": 1.0505,
+      "step": 994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3028095264420534e-05,
+      "loss": 1.3133,
+      "step": 995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.266887125037364e-05,
+      "loss": 1.1243,
+      "step": 996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.231101308059165e-05,
+      "loss": 1.2433,
+      "step": 997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.195452311255235e-05,
+      "loss": 1.0728,
+      "step": 998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.159940369472015e-05,
+      "loss": 0.9998,
+      "step": 999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.124565716653067e-05,
+      "loss": 1.0315,
+      "step": 1000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.089328585837512e-05,
+      "loss": 1.2038,
+      "step": 1001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.054229209158544e-05,
+      "loss": 1.1679,
+      "step": 1002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.019267817841835e-05,
+      "loss": 1.1028,
+      "step": 1003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.984444642204063e-05,
+      "loss": 1.1775,
+      "step": 1004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.949759911651371e-05,
+      "loss": 1.0999,
+      "step": 1005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9152138546778624e-05,
+      "loss": 1.2327,
+      "step": 1006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.880806698864087e-05,
+      "loss": 1.0221,
+      "step": 1007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.846538670875569e-05,
+      "loss": 1.2356,
+      "step": 1008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.812409996461275e-05,
+      "loss": 1.0937,
+      "step": 1009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7784209004521565e-05,
+      "loss": 1.2011,
+      "step": 1010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7445716067596506e-05,
+      "loss": 1.172,
+      "step": 1011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.710862338374219e-05,
+      "loss": 1.2182,
+      "step": 1012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.677293317363864e-05,
+      "loss": 1.0164,
+      "step": 1013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6438647648726924e-05,
+      "loss": 1.1008,
+      "step": 1014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.610576901119422e-05,
+      "loss": 1.1986,
+      "step": 1015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.57742994539596e-05,
+      "loss": 1.1129,
+      "step": 1016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.544424116065931e-05,
+      "loss": 1.2602,
+      "step": 1017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5115596305632854e-05,
+      "loss": 1.0993,
+      "step": 1018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.478836705390809e-05,
+      "loss": 1.1901,
+      "step": 1019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4462555561187355e-05,
+      "loss": 1.1811,
+      "step": 1020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.41381639738331e-05,
+      "loss": 1.1807,
+      "step": 1021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3815194428853945e-05,
+      "loss": 1.1534,
+      "step": 1022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3493649053890325e-05,
+      "loss": 1.2909,
+      "step": 1023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.317352996720063e-05,
+      "loss": 1.2967,
+      "step": 1024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2854839277647265e-05,
+      "loss": 1.2431,
+      "step": 1025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.253757908468269e-05,
+      "loss": 1.033,
+      "step": 1026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.222175147833556e-05,
+      "loss": 1.2833,
+      "step": 1027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.190735853919724e-05,
+      "loss": 1.0004,
+      "step": 1028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.159440233840763e-05,
+      "loss": 1.1293,
+      "step": 1029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1282884937641865e-05,
+      "loss": 1.0512,
+      "step": 1030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.097280838909664e-05,
+      "loss": 1.0307,
+      "step": 1031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.066417473547667e-05,
+      "loss": 1.0601,
+      "step": 1032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0356986009981212e-05,
+      "loss": 1.1182,
+      "step": 1033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0051244236290854e-05,
+      "loss": 1.1825,
+      "step": 1034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.974695142855388e-05,
+      "loss": 1.0933,
+      "step": 1035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9444109591373196e-05,
+      "loss": 1.1215,
+      "step": 1036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9142720719793124e-05,
+      "loss": 1.2765,
+      "step": 1037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.88427867992862e-05,
+      "loss": 1.1549,
+      "step": 1038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.854430980574002e-05,
+      "loss": 1.2543,
+      "step": 1039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.824729170544457e-05,
+      "loss": 1.1894,
+      "step": 1040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7951734455078786e-05,
+      "loss": 1.1371,
+      "step": 1041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7657640001697986e-05,
+      "loss": 1.2182,
+      "step": 1042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7365010282720953e-05,
+      "loss": 1.2534,
+      "step": 1043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.707384722591716e-05,
+      "loss": 1.0925,
+      "step": 1044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.678415274939408e-05,
+      "loss": 1.1895,
+      "step": 1045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.649592876158463e-05,
+      "loss": 1.1423,
+      "step": 1046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6209177161234443e-05,
+      "loss": 1.1321,
+      "step": 1047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5923899837389465e-05,
+      "loss": 1.2068,
+      "step": 1048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5640098669383494e-05,
+      "loss": 1.146,
+      "step": 1049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5357775526825777e-05,
+      "loss": 1.0929,
+      "step": 1050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5076932269588708e-05,
+      "loss": 1.1261,
+      "step": 1051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4797570747795596e-05,
+      "loss": 1.2947,
+      "step": 1052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.451969280180849e-05,
+      "loss": 1.1558,
+      "step": 1053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.424330026221594e-05,
+      "loss": 1.1588,
+      "step": 1054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3968394949821032e-05,
+      "loss": 1.1148,
+      "step": 1055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3694978675629473e-05,
+      "loss": 1.099,
+      "step": 1056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3423053240837516e-05,
+      "loss": 1.107,
+      "step": 1057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3152620436820092e-05,
+      "loss": 1.039,
+      "step": 1058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2883682045119063e-05,
+      "loss": 1.1043,
+      "step": 1059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2616239837431634e-05,
+      "loss": 1.1579,
+      "step": 1060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2350295575598368e-05,
+      "loss": 1.1572,
+      "step": 1061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.208585101159183e-05,
+      "loss": 1.1346,
+      "step": 1062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1822907887504934e-05,
+      "loss": 1.1404,
+      "step": 1063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.156146793553948e-05,
+      "loss": 1.1443,
+      "step": 1064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1301532877994746e-05,
+      "loss": 1.0973,
+      "step": 1065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1043104427256266e-05,
+      "loss": 1.0978,
+      "step": 1066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.07861842857843e-05,
+      "loss": 1.0767,
+      "step": 1067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0530774146102798e-05,
+      "loss": 1.0728,
+      "step": 1068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0276875690788203e-05,
+      "loss": 1.0472,
+      "step": 1069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0024490592458395e-05,
+      "loss": 1.1073,
+      "step": 1070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.977362051376158e-05,
+      "loss": 1.1251,
+      "step": 1071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9524267107365544e-05,
+      "loss": 1.1315,
+      "step": 1072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.927643201594645e-05,
+      "loss": 1.1705,
+      "step": 1073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9030116872178316e-05,
+      "loss": 1.1808,
+      "step": 1074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8785323298722097e-05,
+      "loss": 1.1152,
+      "step": 1075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.854205290821498e-05,
+      "loss": 1.1495,
+      "step": 1076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8300307303259907e-05,
+      "loss": 1.1684,
+      "step": 1077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.806008807641482e-05,
+      "loss": 1.2384,
+      "step": 1078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7821396810182438e-05,
+      "loss": 1.178,
+      "step": 1079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7584235076999465e-05,
+      "loss": 1.2636,
+      "step": 1080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7348604439226617e-05,
+      "loss": 1.3461,
+      "step": 1081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.711450644913798e-05,
+      "loss": 1.0695,
+      "step": 1082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6881942648911074e-05,
+      "loss": 1.1345,
+      "step": 1083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6650914570616486e-05,
+      "loss": 1.0923,
+      "step": 1084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6421423736208e-05,
+      "loss": 1.0734,
+      "step": 1085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6193471657512253e-05,
+      "loss": 1.009,
+      "step": 1086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5967059836219043e-05,
+      "loss": 1.0944,
+      "step": 1087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5742189763871374e-05,
+      "loss": 1.2493,
+      "step": 1088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5518862921855532e-05,
+      "loss": 1.2198,
+      "step": 1089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5297080781391416e-05,
+      "loss": 1.1519,
+      "step": 1090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5076844803522921e-05,
+      "loss": 1.0305,
+      "step": 1091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4858156439108094e-05,
+      "loss": 1.1734,
+      "step": 1092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4641017128809802e-05,
+      "loss": 1.0773,
+      "step": 1093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.442542830308599e-05,
+      "loss": 1.1516,
+      "step": 1094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4211391382180639e-05,
+      "loss": 1.1512,
+      "step": 1095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3998907776113984e-05,
+      "loss": 1.1634,
+      "step": 1096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3787978884673452e-05,
+      "loss": 1.1719,
+      "step": 1097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.357860609740455e-05,
+      "loss": 1.0838,
+      "step": 1098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3370790793601372e-05,
+      "loss": 1.058,
+      "step": 1099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3164534342297863e-05,
+      "loss": 1.1387,
+      "step": 1100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2959838102258537e-05,
+      "loss": 1.1291,
+      "step": 1101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2756703421969745e-05,
+      "loss": 1.1521,
+      "step": 1102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2555131639630568e-05,
+      "loss": 1.162,
+      "step": 1103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2355124083144182e-05,
+      "loss": 1.2249,
+      "step": 1104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2156682070109087e-05,
+      "loss": 1.1332,
+      "step": 1105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1959806907810289e-05,
+      "loss": 1.2263,
+      "step": 1106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1764499893210877e-05,
+      "loss": 1.1723,
+      "step": 1107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1570762312943296e-05,
+      "loss": 1.1984,
+      "step": 1108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1378595443301e-05,
+      "loss": 1.1274,
+      "step": 1109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1188000550230004e-05,
+      "loss": 1.0725,
+      "step": 1110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0998978889320582e-05,
+      "loss": 1.2412,
+      "step": 1111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0811531705798905e-05,
+      "loss": 1.2438,
+      "step": 1112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0625660234518913e-05,
+      "loss": 1.156,
+      "step": 1113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0441365699954159e-05,
+      "loss": 1.0937,
+      "step": 1114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0258649316189723e-05,
+      "loss": 1.1832,
+      "step": 1115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0077512286914226e-05,
+      "loss": 1.1857,
+      "step": 1116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.897955805412e-06,
+      "loss": 1.0269,
+      "step": 1117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.719981054555017e-06,
+      "loss": 1.1238,
+      "step": 1118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.54358920679524e-06,
+      "loss": 1.0953,
+      "step": 1119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.368781424156924e-06,
+      "loss": 1.0879,
+      "step": 1120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.19555885822887e-06,
+      "loss": 1.1497,
+      "step": 1121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.023922650156863e-06,
+      "loss": 1.074,
+      "step": 1122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.853873930636218e-06,
+      "loss": 1.1549,
+      "step": 1123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.68541381990426e-06,
+      "loss": 1.0667,
+      "step": 1124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.51854342773295e-06,
+      "loss": 1.0716,
+      "step": 1125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.353263853421584e-06,
+      "loss": 1.1075,
+      "step": 1126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.189576185789638e-06,
+      "loss": 0.9856,
+      "step": 1127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.02748150316937e-06,
+      "loss": 1.0225,
+      "step": 1128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.866980873399016e-06,
+      "loss": 1.2488,
+      "step": 1129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.708075353815513e-06,
+      "loss": 1.1536,
+      "step": 1130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.550765991247654e-06,
+      "loss": 1.2557,
+      "step": 1131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.3950538220090846e-06,
+      "loss": 1.2443,
+      "step": 1132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.2409398718916994e-06,
+      "loss": 1.1899,
+      "step": 1133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.088425156158623e-06,
+      "loss": 1.1983,
+      "step": 1134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.937510679537628e-06,
+      "loss": 1.0985,
+      "step": 1135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.78819743621456e-06,
+      "loss": 1.094,
+      "step": 1136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.640486409826785e-06,
+      "loss": 1.138,
+      "step": 1137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.494378573456611e-06,
+      "loss": 1.1446,
+      "step": 1138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.349874889624963e-06,
+      "loss": 1.1106,
+      "step": 1139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.206976310284995e-06,
+      "loss": 1.0965,
+      "step": 1140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.065683776815933e-06,
+      "loss": 1.1614,
+      "step": 1141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.92599822001666e-06,
+      "loss": 1.0881,
+      "step": 1142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.78792056009983e-06,
+      "loss": 1.0994,
+      "step": 1143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.651451706685601e-06,
+      "loss": 1.2027,
+      "step": 1144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.516592558795747e-06,
+      "loss": 1.034,
+      "step": 1145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.383344004847773e-06,
+      "loss": 1.0785,
+      "step": 1146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.251706922648869e-06,
+      "loss": 1.15,
+      "step": 1147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.121682179390408e-06,
+      "loss": 1.1274,
+      "step": 1148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.993270631642038e-06,
+      "loss": 1.1822,
+      "step": 1149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.866473125346049e-06,
+      "loss": 1.1644,
+      "step": 1150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.741290495811873e-06,
+      "loss": 1.2427,
+      "step": 1151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6177235677105636e-06,
+      "loss": 1.0606,
+      "step": 1152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4957731550692995e-06,
+      "loss": 1.0533,
+      "step": 1153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3754400612661115e-06,
+      "loss": 1.0582,
+      "step": 1154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.256725079024554e-06,
+      "loss": 0.9833,
+      "step": 1155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.139628990408456e-06,
+      "loss": 1.2803,
+      "step": 1156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.024152566816791e-06,
+      "loss": 1.0903,
+      "step": 1157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.910296568978622e-06,
+      "loss": 1.1445,
+      "step": 1158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.798061746947995e-06,
+      "loss": 1.1353,
+      "step": 1159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.687448840099139e-06,
+      "loss": 1.198,
+      "step": 1160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.578458577121524e-06,
+      "loss": 1.1087,
+      "step": 1161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.471091676014976e-06,
+      "loss": 1.2356,
+      "step": 1162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3653488440851255e-06,
+      "loss": 1.2982,
+      "step": 1163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2612307779386065e-06,
+      "loss": 1.0539,
+      "step": 1164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.158738163478475e-06,
+      "loss": 1.1077,
+      "step": 1165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.057871675899826e-06,
+      "loss": 1.1119,
+      "step": 1166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9586319796851556e-06,
+      "loss": 1.1195,
+      "step": 1167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.861019728600117e-06,
+      "loss": 1.0234,
+      "step": 1168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7650355656892167e-06,
+      "loss": 1.2179,
+      "step": 1169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6706801232714018e-06,
+      "loss": 1.0648,
+      "step": 1170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5779540229361743e-06,
+      "loss": 1.1121,
+      "step": 1171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.486857875539261e-06,
+      "loss": 1.0745,
+      "step": 1172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.397392281198729e-06,
+      "loss": 1.0614,
+      "step": 1173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.309557829290876e-06,
+      "loss": 1.1138,
+      "step": 1174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2233550984466224e-06,
+      "loss": 1.0559,
+      "step": 1175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1387846565474044e-06,
+      "loss": 0.9898,
+      "step": 1176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.055847060721566e-06,
+      "loss": 1.1394,
+      "step": 1177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9745428573406943e-06,
+      "loss": 1.2373,
+      "step": 1178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8948725820160662e-06,
+      "loss": 1.1048,
+      "step": 1179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8168367595949586e-06,
+      "loss": 1.0287,
+      "step": 1180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7404359041573725e-06,
+      "loss": 1.161,
+      "step": 1181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6656705190125076e-06,
+      "loss": 1.0959,
+      "step": 1182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5925410966955712e-06,
+      "loss": 1.2044,
+      "step": 1183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.521048118964391e-06,
+      "loss": 1.2111,
+      "step": 1184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.451192056796391e-06,
+      "loss": 1.365,
+      "step": 1185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3829733703853154e-06,
+      "loss": 1.1618,
+      "step": 1186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3163925091384533e-06,
+      "loss": 1.2241,
+      "step": 1187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2514499116733636e-06,
+      "loss": 1.0862,
+      "step": 1188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1881460058152383e-06,
+      "loss": 1.2027,
+      "step": 1189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1264812085939602e-06,
+      "loss": 1.1396,
+      "step": 1190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.066455926241383e-06,
+      "loss": 1.3238,
+      "step": 1191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0080705541886392e-06,
+      "loss": 1.1636,
+      "step": 1192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.513254770636137e-07,
+      "loss": 1.2061,
+      "step": 1193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.962210686883076e-07,
+      "loss": 1.138,
+      "step": 1194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.427576920763958e-07,
+      "loss": 1.18,
+      "step": 1195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.909356994308947e-07,
+      "loss": 1.0543,
+      "step": 1196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.407554321417765e-07,
+      "loss": 1.3065,
+      "step": 1197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.922172207838029e-07,
+      "loss": 1.3796,
+      "step": 1198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.453213851142226e-07,
+      "loss": 1.2853,
+      "step": 1199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.00068234070772e-07,
+      "loss": 1.0838,
+      "step": 1200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.564580657695939e-07,
+      "loss": 0.9754,
+      "step": 1201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.144911675032671e-07,
+      "loss": 1.2056,
+      "step": 1202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.741678157389739e-07,
+      "loss": 1.1819,
+      "step": 1203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3548827611661344e-07,
+      "loss": 1.1037,
+      "step": 1204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.984528034470525e-07,
+      "loss": 1.0535,
+      "step": 1205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.630616417104604e-07,
+      "loss": 1.1327,
+      "step": 1206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.293150240547549e-07,
+      "loss": 1.0755,
+      "step": 1207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.972131727939642e-07,
+      "loss": 1.0664,
+      "step": 1208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6675629940689505e-07,
+      "loss": 1.2237,
+      "step": 1209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3794460453555045e-07,
+      "loss": 1.2799,
+      "step": 1210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1077827798404726e-07,
+      "loss": 1.0723,
+      "step": 1211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.852574987171174e-07,
+      "loss": 1.3057,
+      "step": 1212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6138243485910864e-07,
+      "loss": 1.1932,
+      "step": 1213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3915324369284665e-07,
+      "loss": 1.245,
+      "step": 1214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1857007165852473e-07,
+      "loss": 1.0783,
+      "step": 1215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.963305435278791e-08,
+      "loss": 1.0482,
+      "step": 1216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.23423165278725e-08,
+      "loss": 1.1656,
+      "step": 1217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.669797209069017e-08,
+      "loss": 1.0242,
+      "step": 1218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.270012410216185e-08,
+      "loss": 1.2263,
+      "step": 1219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.034886477655153e-08,
+      "loss": 1.1416,
+      "step": 1220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.964427548077242e-08,
+      "loss": 1.1767,
+      "step": 1221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0586426733887332e-08,
+      "loss": 1.0344,
+      "step": 1222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3175378206720102e-08,
+      "loss": 1.1361,
+      "step": 1223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.4111787213282375e-09,
+      "loss": 1.0677,
+      "step": 1224
+    },
+    {
+      "epoch": 1.0,
+      "step": 1224,
+      "total_flos": 1.804319615782748e+17,
+      "train_loss": 1.241359075640931,
+      "train_runtime": 34128.9816,
+      "train_samples_per_second": 0.287,
+      "train_steps_per_second": 0.036
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1224,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 1.804319615782748e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}