diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,132109 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.945705295574747,
+  "eval_steps": 2000,
+  "global_step": 22000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.5365,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.4654,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 1.3935,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 1.2862,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.6635,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999107102995671,
+      "loss": 1.3839,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999821420599134,
+      "loss": 1.379,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999732130898701,
+      "loss": 1.5376,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999642841198268,
+      "loss": 1.6643,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019995535514978348,
+      "loss": 1.4524,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019994642617974018,
+      "loss": 1.5279,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019993749720969688,
+      "loss": 1.4933,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019992856823965356,
+      "loss": 1.3967,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019991963926961026,
+      "loss": 1.4166,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019991071029956697,
+      "loss": 1.4646,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019990178132952365,
+      "loss": 1.4877,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019989285235948035,
+      "loss": 1.3728,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019988392338943705,
+      "loss": 1.2623,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019987499441939373,
+      "loss": 1.5161,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019986606544935043,
+      "loss": 1.5199,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019985713647930714,
+      "loss": 1.3659,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019984820750926382,
+      "loss": 1.2751,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019983927853922052,
+      "loss": 1.4322,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019983034956917722,
+      "loss": 1.5427,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001998214205991339,
+      "loss": 1.5632,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001998124916290906,
+      "loss": 1.2686,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019980356265904728,
+      "loss": 1.4814,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019979463368900399,
+      "loss": 1.3408,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019978570471896066,
+      "loss": 1.5212,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019977677574891737,
+      "loss": 1.6856,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019976784677887407,
+      "loss": 1.3954,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019975891780883075,
+      "loss": 1.3921,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019974998883878745,
+      "loss": 1.3373,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019974105986874413,
+      "loss": 1.3558,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019973213089870083,
+      "loss": 1.4336,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019972320192865754,
+      "loss": 1.4379,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001997142729586142,
+      "loss": 1.4776,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019970534398857092,
+      "loss": 1.6107,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019969641501852762,
+      "loss": 1.3955,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001996874860484843,
+      "loss": 1.4252,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199678557078441,
+      "loss": 1.4612,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001996696281083977,
+      "loss": 1.5756,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019966069913835438,
+      "loss": 1.5048,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001996517701683111,
+      "loss": 1.4496,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001996428411982678,
+      "loss": 1.451,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019963391222822447,
+      "loss": 1.4355,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019962498325818117,
+      "loss": 1.4645,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019961605428813788,
+      "loss": 1.4661,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019960712531809455,
+      "loss": 1.79,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019959819634805126,
+      "loss": 1.4122,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019958926737800796,
+      "loss": 1.808,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019958033840796464,
+      "loss": 1.7562,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019957140943792134,
+      "loss": 1.4916,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019956248046787805,
+      "loss": 1.4898,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019955355149783475,
+      "loss": 1.5679,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019954462252779143,
+      "loss": 1.3971,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019953569355774813,
+      "loss": 1.2793,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019952676458770484,
+      "loss": 1.4984,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001995178356176615,
+      "loss": 1.2821,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019950890664761822,
+      "loss": 1.5591,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019949997767757492,
+      "loss": 1.6317,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001994910487075316,
+      "loss": 1.3388,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001994821197374883,
+      "loss": 1.1273,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199473190767445,
+      "loss": 1.3858,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019946426179740168,
+      "loss": 1.6441,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019945533282735839,
+      "loss": 1.741,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001994464038573151,
+      "loss": 1.5055,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019943747488727177,
+      "loss": 1.474,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019942854591722847,
+      "loss": 1.1919,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019941961694718515,
+      "loss": 1.3789,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019941068797714185,
+      "loss": 1.27,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019940175900709856,
+      "loss": 1.5614,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019939283003705523,
+      "loss": 1.2595,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019938390106701194,
+      "loss": 1.3344,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019937497209696861,
+      "loss": 1.1496,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019936604312692532,
+      "loss": 1.3231,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199357114156882,
+      "loss": 1.2975,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001993481851868387,
+      "loss": 1.3656,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001993392562167954,
+      "loss": 1.3085,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019933032724675208,
+      "loss": 1.3513,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019932139827670878,
+      "loss": 1.2063,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001993124693066655,
+      "loss": 1.3567,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019930354033662217,
+      "loss": 1.1856,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019929461136657887,
+      "loss": 1.3735,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019928568239653557,
+      "loss": 1.2972,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019927675342649225,
+      "loss": 1.3742,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019926782445644895,
+      "loss": 1.4178,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019925889548640566,
+      "loss": 1.3793,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019924996651636234,
+      "loss": 1.3741,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019924103754631904,
+      "loss": 1.4098,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019923210857627574,
+      "loss": 1.4321,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019922317960623242,
+      "loss": 1.3354,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019921425063618912,
+      "loss": 1.3546,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019920532166614583,
+      "loss": 1.4222,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001991963926961025,
+      "loss": 1.3682,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001991874637260592,
+      "loss": 1.3111,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001991785347560159,
+      "loss": 1.4677,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001991696057859726,
+      "loss": 1.4413,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001991606768159293,
+      "loss": 1.325,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199151747845886,
+      "loss": 1.3287,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019914281887584268,
+      "loss": 1.2809,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019913388990579938,
+      "loss": 1.6719,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019912496093575608,
+      "loss": 1.4391,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019911603196571276,
+      "loss": 2.1635,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019910710299566946,
+      "loss": 1.4189,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019909817402562617,
+      "loss": 1.1291,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019908924505558285,
+      "loss": 1.3977,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019908031608553955,
+      "loss": 1.2441,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019907138711549625,
+      "loss": 1.4188,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019906245814545293,
+      "loss": 1.2397,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019905352917540963,
+      "loss": 1.3032,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019904460020536634,
+      "loss": 1.3096,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019903567123532302,
+      "loss": 1.3218,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019902674226527972,
+      "loss": 1.5651,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019901781329523642,
+      "loss": 1.2239,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001990088843251931,
+      "loss": 1.2842,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001989999553551498,
+      "loss": 1.3742,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019899102638510648,
+      "loss": 1.3189,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019898209741506319,
+      "loss": 1.2361,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019897316844501986,
+      "loss": 1.1933,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019896423947497657,
+      "loss": 1.5609,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019895531050493327,
+      "loss": 1.2154,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019894638153488995,
+      "loss": 1.3312,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019893745256484665,
+      "loss": 1.197,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019892852359480333,
+      "loss": 1.2502,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019891959462476003,
+      "loss": 1.2269,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019891066565471674,
+      "loss": 1.2855,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001989017366846734,
+      "loss": 1.3553,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019889280771463012,
+      "loss": 1.3116,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019888387874458682,
+      "loss": 1.1465,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001988749497745435,
+      "loss": 1.2747,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001988660208045002,
+      "loss": 1.2193,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001988570918344569,
+      "loss": 1.33,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019884816286441358,
+      "loss": 1.2168,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001988392338943703,
+      "loss": 1.3909,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000198830304924327,
+      "loss": 1.3214,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019882137595428367,
+      "loss": 1.2504,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019881244698424037,
+      "loss": 1.108,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019880351801419708,
+      "loss": 1.2616,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019879458904415375,
+      "loss": 1.4223,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019878566007411046,
+      "loss": 1.1857,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019877673110406716,
+      "loss": 1.146,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019876780213402384,
+      "loss": 1.324,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019875887316398054,
+      "loss": 1.2094,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019874994419393725,
+      "loss": 1.3253,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019874101522389392,
+      "loss": 1.3224,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019873208625385063,
+      "loss": 1.3011,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019872315728380733,
+      "loss": 1.313,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000198714228313764,
+      "loss": 1.4943,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001987052993437207,
+      "loss": 1.362,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019869637037367742,
+      "loss": 1.351,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019868744140363412,
+      "loss": 1.324,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001986785124335908,
+      "loss": 1.277,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001986695834635475,
+      "loss": 1.2552,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001986606544935042,
+      "loss": 1.3576,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019865172552346088,
+      "loss": 1.1347,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019864279655341759,
+      "loss": 1.2362,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001986338675833743,
+      "loss": 1.2331,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019862493861333097,
+      "loss": 1.2726,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019861600964328767,
+      "loss": 1.1742,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019860708067324435,
+      "loss": 1.206,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019859815170320105,
+      "loss": 1.1353,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019858922273315776,
+      "loss": 1.3699,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019858029376311443,
+      "loss": 1.4025,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019857136479307114,
+      "loss": 1.7966,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019856243582302781,
+      "loss": 1.1683,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019855350685298452,
+      "loss": 1.2148,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001985445778829412,
+      "loss": 1.2705,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001985356489128979,
+      "loss": 1.2243,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001985267199428546,
+      "loss": 1.2995,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019851779097281128,
+      "loss": 1.2397,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019850886200276798,
+      "loss": 1.2142,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001984999330327247,
+      "loss": 1.1,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019849100406268136,
+      "loss": 1.3452,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019848207509263807,
+      "loss": 1.3768,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019847314612259477,
+      "loss": 1.237,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019846421715255145,
+      "loss": 1.1522,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019845528818250815,
+      "loss": 1.1073,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019844635921246486,
+      "loss": 1.1882,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019843743024242153,
+      "loss": 1.0905,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019842850127237824,
+      "loss": 1.2771,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019841957230233494,
+      "loss": 1.2216,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019841064333229162,
+      "loss": 1.3261,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019840171436224832,
+      "loss": 1.2689,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019839278539220503,
+      "loss": 1.2032,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001983838564221617,
+      "loss": 1.2358,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001983749274521184,
+      "loss": 1.1352,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001983659984820751,
+      "loss": 1.2196,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001983570695120318,
+      "loss": 1.2045,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001983481405419885,
+      "loss": 1.2232,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001983392115719452,
+      "loss": 1.1456,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019833028260190187,
+      "loss": 1.3001,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019832135363185858,
+      "loss": 1.4018,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019831242466181528,
+      "loss": 1.1807,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019830349569177196,
+      "loss": 1.1551,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019829456672172866,
+      "loss": 1.3377,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019828563775168537,
+      "loss": 1.1802,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019827670878164204,
+      "loss": 1.1659,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019826777981159875,
+      "loss": 1.0108,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019825885084155545,
+      "loss": 1.1846,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019824992187151213,
+      "loss": 1.1948,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019824099290146883,
+      "loss": 1.3417,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019823206393142554,
+      "loss": 1.2622,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019822313496138221,
+      "loss": 1.3854,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019821420599133892,
+      "loss": 1.2945,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019820527702129562,
+      "loss": 1.3606,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001981963480512523,
+      "loss": 1.3578,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000198187419081209,
+      "loss": 1.3052,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019817849011116568,
+      "loss": 1.2062,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019816956114112238,
+      "loss": 1.169,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019816063217107906,
+      "loss": 1.2455,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019815170320103577,
+      "loss": 1.2185,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019814277423099244,
+      "loss": 1.1758,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019813384526094915,
+      "loss": 1.2067,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019812491629090585,
+      "loss": 1.9907,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019811598732086253,
+      "loss": 1.2641,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019810705835081923,
+      "loss": 1.2698,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019809812938077594,
+      "loss": 1.1846,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001980892004107326,
+      "loss": 1.2078,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019808027144068932,
+      "loss": 1.3202,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019807134247064602,
+      "loss": 1.2477,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001980624135006027,
+      "loss": 1.1811,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001980534845305594,
+      "loss": 1.1719,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001980445555605161,
+      "loss": 1.2005,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019803562659047278,
+      "loss": 1.1854,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001980266976204295,
+      "loss": 1.1134,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001980177686503862,
+      "loss": 1.0676,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019800883968034287,
+      "loss": 1.08,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019799991071029957,
+      "loss": 1.3237,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019799098174025628,
+      "loss": 1.1233,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019798205277021295,
+      "loss": 1.3126,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019797312380016966,
+      "loss": 1.332,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019796419483012636,
+      "loss": 1.0181,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019795526586008304,
+      "loss": 1.0853,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019794633689003974,
+      "loss": 1.167,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019793740791999645,
+      "loss": 1.1485,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019792847894995312,
+      "loss": 1.2031,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019791954997990983,
+      "loss": 1.3194,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019791062100986653,
+      "loss": 1.2322,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001979016920398232,
+      "loss": 1.2196,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001978927630697799,
+      "loss": 1.4948,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019788383409973662,
+      "loss": 1.1226,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001978749051296933,
+      "loss": 1.2536,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019786597615965,
+      "loss": 1.3927,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001978570471896067,
+      "loss": 1.0517,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019784811821956338,
+      "loss": 1.267,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019783918924952008,
+      "loss": 1.1511,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019783026027947679,
+      "loss": 1.0819,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019782133130943346,
+      "loss": 1.2296,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019781240233939017,
+      "loss": 1.2061,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019780347336934687,
+      "loss": 1.1881,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019779454439930355,
+      "loss": 1.1344,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019778561542926025,
+      "loss": 1.2581,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019777668645921696,
+      "loss": 1.2338,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019776775748917363,
+      "loss": 1.2076,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019775882851913034,
+      "loss": 1.1536,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000197749899549087,
+      "loss": 1.2376,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019774097057904372,
+      "loss": 1.2581,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001977320416090004,
+      "loss": 1.0811,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001977231126389571,
+      "loss": 1.1239,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001977141836689138,
+      "loss": 1.1114,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019770525469887048,
+      "loss": 1.0588,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019769632572882718,
+      "loss": 1.1496,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001976873967587839,
+      "loss": 0.9874,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019767846778874056,
+      "loss": 1.1813,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019766953881869727,
+      "loss": 1.0996,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019766060984865397,
+      "loss": 1.0362,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019765168087861065,
+      "loss": 1.1308,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019764275190856735,
+      "loss": 1.3873,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019763382293852406,
+      "loss": 1.1609,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019762489396848073,
+      "loss": 1.2172,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019761596499843744,
+      "loss": 1.1362,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019760703602839414,
+      "loss": 1.43,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019759810705835082,
+      "loss": 1.2557,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019758917808830752,
+      "loss": 1.1606,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019758024911826423,
+      "loss": 1.0473,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001975713201482209,
+      "loss": 1.0507,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001975623911781776,
+      "loss": 1.2542,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001975534622081343,
+      "loss": 1.1049,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000197544533238091,
+      "loss": 1.2446,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001975356042680477,
+      "loss": 1.0338,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001975266752980044,
+      "loss": 1.244,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019751774632796107,
+      "loss": 1.1842,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019750881735791778,
+      "loss": 1.1909,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019749988838787448,
+      "loss": 1.0688,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019749095941783116,
+      "loss": 1.0785,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019748203044778786,
+      "loss": 1.0861,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019747310147774457,
+      "loss": 1.1233,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019746417250770124,
+      "loss": 1.1107,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019745524353765795,
+      "loss": 1.1563,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019744631456761465,
+      "loss": 1.1028,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019743738559757133,
+      "loss": 1.2607,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019742845662752803,
+      "loss": 1.3394,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019741952765748474,
+      "loss": 1.1079,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019741059868744141,
+      "loss": 1.1142,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019740166971739812,
+      "loss": 1.1361,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019739274074735482,
+      "loss": 1.074,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001973838117773115,
+      "loss": 1.1505,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001973748828072682,
+      "loss": 1.2272,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019736595383722488,
+      "loss": 1.2229,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019735702486718158,
+      "loss": 1.084,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019734809589713826,
+      "loss": 1.0664,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019733916692709497,
+      "loss": 1.0594,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019733023795705164,
+      "loss": 1.1002,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019732130898700835,
+      "loss": 1.1297,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019731238001696505,
+      "loss": 1.1208,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019730345104692173,
+      "loss": 1.1279,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019729452207687843,
+      "loss": 1.1164,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019728559310683514,
+      "loss": 1.2423,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001972766641367918,
+      "loss": 1.2391,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019726773516674852,
+      "loss": 1.1111,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019725880619670522,
+      "loss": 1.1029,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001972498772266619,
+      "loss": 1.1651,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001972409482566186,
+      "loss": 0.9969,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001972320192865753,
+      "loss": 1.1234,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019722309031653198,
+      "loss": 1.1982,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019721416134648869,
+      "loss": 1.2783,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001972052323764454,
+      "loss": 1.1882,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019719630340640207,
+      "loss": 1.2359,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019718737443635877,
+      "loss": 1.0954,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019717844546631548,
+      "loss": 1.1638,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019716951649627215,
+      "loss": 1.0712,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019716058752622886,
+      "loss": 1.1334,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019715165855618556,
+      "loss": 1.1838,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019714272958614224,
+      "loss": 1.1988,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019713380061609894,
+      "loss": 1.2043,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019712487164605565,
+      "loss": 1.1277,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019711594267601232,
+      "loss": 1.1669,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019710701370596903,
+      "loss": 1.0529,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019709808473592573,
+      "loss": 1.2401,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001970891557658824,
+      "loss": 1.1179,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001970802267958391,
+      "loss": 1.1525,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019707129782579582,
+      "loss": 1.0563,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001970623688557525,
+      "loss": 1.184,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001970534398857092,
+      "loss": 1.1955,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001970445109156659,
+      "loss": 1.0427,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019703558194562258,
+      "loss": 1.1174,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019702665297557928,
+      "loss": 1.0866,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019701772400553599,
+      "loss": 1.0837,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019700879503549266,
+      "loss": 1.0696,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019699986606544937,
+      "loss": 1.2374,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019699093709540607,
+      "loss": 1.1412,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019698200812536275,
+      "loss": 1.1844,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019697307915531945,
+      "loss": 1.112,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019696415018527616,
+      "loss": 1.1295,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019695522121523283,
+      "loss": 1.3161,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019694629224518954,
+      "loss": 1.0646,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001969373632751462,
+      "loss": 1.2071,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019692843430510292,
+      "loss": 1.1628,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001969195053350596,
+      "loss": 1.0621,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001969105763650163,
+      "loss": 1.1839,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000196901647394973,
+      "loss": 1.1336,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019689271842492968,
+      "loss": 1.1668,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019688378945488638,
+      "loss": 1.121,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001968748604848431,
+      "loss": 1.1588,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019686593151479976,
+      "loss": 1.0446,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019685700254475647,
+      "loss": 1.1181,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019684807357471317,
+      "loss": 1.1623,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019683914460466985,
+      "loss": 1.0846,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019683021563462655,
+      "loss": 1.0669,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019682128666458326,
+      "loss": 1.1641,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019681235769453993,
+      "loss": 1.0303,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019680342872449664,
+      "loss": 1.1685,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019679449975445334,
+      "loss": 1.2225,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019678557078441002,
+      "loss": 1.1388,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019677664181436672,
+      "loss": 1.1797,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019676771284432343,
+      "loss": 1.2363,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001967587838742801,
+      "loss": 1.0235,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001967498549042368,
+      "loss": 1.0956,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001967409259341935,
+      "loss": 1.0656,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001967319969641502,
+      "loss": 1.1828,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001967230679941069,
+      "loss": 1.0719,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001967141390240636,
+      "loss": 1.0824,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019670521005402027,
+      "loss": 1.2191,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019669628108397698,
+      "loss": 1.1139,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019668735211393368,
+      "loss": 1.0743,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019667842314389036,
+      "loss": 1.1616,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019666949417384706,
+      "loss": 1.1258,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019666056520380377,
+      "loss": 1.1187,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019665163623376044,
+      "loss": 1.2394,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019664270726371715,
+      "loss": 1.0104,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019663377829367385,
+      "loss": 1.0444,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019662484932363053,
+      "loss": 1.1419,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019661592035358723,
+      "loss": 1.3552,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019660699138354394,
+      "loss": 1.0891,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019659806241350061,
+      "loss": 1.0988,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019658913344345732,
+      "loss": 1.1478,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019658020447341402,
+      "loss": 1.3217,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001965712755033707,
+      "loss": 1.2317,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001965623465333274,
+      "loss": 1.7696,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019655341756328408,
+      "loss": 1.0012,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019654448859324078,
+      "loss": 1.0514,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019653555962319746,
+      "loss": 1.2735,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019652663065315416,
+      "loss": 1.2504,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019651770168311084,
+      "loss": 1.2017,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019650877271306755,
+      "loss": 1.0927,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019649984374302425,
+      "loss": 1.4297,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019649091477298093,
+      "loss": 1.1735,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019648198580293763,
+      "loss": 1.1585,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019647305683289433,
+      "loss": 1.1252,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000196464127862851,
+      "loss": 1.1459,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019645519889280772,
+      "loss": 1.1837,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019644626992276442,
+      "loss": 1.1319,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001964373409527211,
+      "loss": 1.1483,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001964284119826778,
+      "loss": 1.5299,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001964194830126345,
+      "loss": 1.1348,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019641055404259118,
+      "loss": 1.1769,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019640162507254789,
+      "loss": 1.1574,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001963926961025046,
+      "loss": 1.125,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019638376713246127,
+      "loss": 1.1599,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019637483816241797,
+      "loss": 1.1364,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019636590919237467,
+      "loss": 1.0377,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019635698022233135,
+      "loss": 1.192,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019634805125228806,
+      "loss": 1.1843,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019633912228224476,
+      "loss": 1.0353,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019633019331220144,
+      "loss": 1.1153,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019632126434215814,
+      "loss": 1.2703,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019631233537211484,
+      "loss": 1.1457,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019630340640207152,
+      "loss": 1.0955,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019629447743202823,
+      "loss": 1.1127,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019628554846198493,
+      "loss": 1.0415,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001962766194919416,
+      "loss": 1.0867,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001962676905218983,
+      "loss": 1.0972,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019625876155185501,
+      "loss": 1.1709,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001962498325818117,
+      "loss": 1.4965,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001962409036117684,
+      "loss": 1.2341,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001962319746417251,
+      "loss": 1.1696,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019622304567168178,
+      "loss": 1.0222,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019621411670163848,
+      "loss": 1.1138,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019620518773159518,
+      "loss": 1.2534,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019619625876155186,
+      "loss": 1.1422,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019618732979150857,
+      "loss": 1.1391,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019617840082146527,
+      "loss": 1.4719,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019616947185142195,
+      "loss": 1.0854,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019616054288137865,
+      "loss": 1.2378,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019615161391133533,
+      "loss": 1.3235,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019614268494129203,
+      "loss": 1.0014,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019613375597124874,
+      "loss": 1.102,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001961248270012054,
+      "loss": 0.9959,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019611589803116212,
+      "loss": 1.0624,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001961069690611188,
+      "loss": 1.0772,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001960980400910755,
+      "loss": 1.0754,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019608911112103217,
+      "loss": 1.1764,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019608018215098888,
+      "loss": 1.0594,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019607125318094558,
+      "loss": 1.2613,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001960623242109023,
+      "loss": 1.0508,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019605339524085896,
+      "loss": 1.1224,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019604446627081567,
+      "loss": 1.0945,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019603553730077237,
+      "loss": 1.1767,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019602660833072905,
+      "loss": 0.9944,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019601767936068575,
+      "loss": 1.1861,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019600875039064246,
+      "loss": 1.0779,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019599982142059913,
+      "loss": 1.1314,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019599089245055584,
+      "loss": 1.0306,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019598196348051254,
+      "loss": 1.2181,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019597303451046922,
+      "loss": 1.2946,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019596410554042592,
+      "loss": 1.2343,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019595517657038263,
+      "loss": 1.2072,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001959462476003393,
+      "loss": 1.1122,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000195937318630296,
+      "loss": 1.2741,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001959283896602527,
+      "loss": 1.3947,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001959194606902094,
+      "loss": 1.2408,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001959105317201661,
+      "loss": 1.0867,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001959016027501228,
+      "loss": 1.3034,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019589267378007947,
+      "loss": 1.0743,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019588374481003618,
+      "loss": 1.0712,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019587481583999288,
+      "loss": 1.1586,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019586588686994956,
+      "loss": 1.0743,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019585695789990626,
+      "loss": 1.1477,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019584802892986297,
+      "loss": 1.1001,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019583909995981964,
+      "loss": 1.0671,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019583017098977635,
+      "loss": 1.0631,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019582124201973305,
+      "loss": 1.1401,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019581231304968973,
+      "loss": 1.1799,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019580338407964643,
+      "loss": 1.2198,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019579445510960314,
+      "loss": 1.3011,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019578552613955981,
+      "loss": 1.042,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019577659716951652,
+      "loss": 1.2723,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019576766819947322,
+      "loss": 1.119,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001957587392294299,
+      "loss": 1.082,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001957498102593866,
+      "loss": 1.1045,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019574088128934328,
+      "loss": 1.1984,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019573195231929998,
+      "loss": 1.0039,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019572302334925666,
+      "loss": 1.1994,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019571409437921336,
+      "loss": 1.0953,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019570516540917004,
+      "loss": 2.5092,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019569623643912675,
+      "loss": 1.3729,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019568730746908345,
+      "loss": 1.1361,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019567837849904013,
+      "loss": 1.2131,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019566944952899683,
+      "loss": 1.1004,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019566052055895353,
+      "loss": 1.1764,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001956515915889102,
+      "loss": 1.1193,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019564266261886692,
+      "loss": 1.0498,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019563373364882362,
+      "loss": 1.1479,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001956248046787803,
+      "loss": 1.18,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000195615875708737,
+      "loss": 1.1292,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001956069467386937,
+      "loss": 1.1642,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019559801776865038,
+      "loss": 1.1252,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019558908879860709,
+      "loss": 1.1132,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001955801598285638,
+      "loss": 0.9906,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019557123085852047,
+      "loss": 1.1821,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019556230188847717,
+      "loss": 1.0973,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019555337291843387,
+      "loss": 1.1511,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019554444394839055,
+      "loss": 1.1319,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019553551497834726,
+      "loss": 0.9462,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019552658600830396,
+      "loss": 1.0895,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019551765703826064,
+      "loss": 1.0965,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019550872806821734,
+      "loss": 1.0713,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019549979909817404,
+      "loss": 1.1267,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019549087012813072,
+      "loss": 1.1153,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019548194115808743,
+      "loss": 1.0762,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019547301218804413,
+      "loss": 1.002,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001954640832180008,
+      "loss": 1.0756,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001954551542479575,
+      "loss": 1.2807,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019544622527791421,
+      "loss": 1.0423,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001954372963078709,
+      "loss": 1.1189,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001954283673378276,
+      "loss": 1.0183,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001954194383677843,
+      "loss": 1.0985,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019541050939774098,
+      "loss": 1.0538,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019540158042769768,
+      "loss": 1.0284,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019539265145765438,
+      "loss": 1.1131,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019538372248761106,
+      "loss": 0.9865,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019537479351756777,
+      "loss": 1.1284,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019536586454752447,
+      "loss": 1.1138,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019535693557748115,
+      "loss": 1.282,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019534800660743785,
+      "loss": 1.1201,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019533907763739453,
+      "loss": 1.2077,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019533014866735123,
+      "loss": 1.1236,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019532121969730794,
+      "loss": 1.0631,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001953122907272646,
+      "loss": 1.1995,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019530336175722132,
+      "loss": 1.0107,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000195294432787178,
+      "loss": 1.0997,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001952855038171347,
+      "loss": 1.1324,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019527657484709137,
+      "loss": 1.0508,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019526764587704808,
+      "loss": 1.1395,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019525871690700478,
+      "loss": 1.135,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019524978793696146,
+      "loss": 1.1502,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019524085896691816,
+      "loss": 1.1438,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019523192999687487,
+      "loss": 1.1132,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019522300102683154,
+      "loss": 1.117,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019521407205678825,
+      "loss": 1.1308,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019520514308674495,
+      "loss": 1.0661,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019519621411670166,
+      "loss": 1.0684,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019518728514665833,
+      "loss": 1.2714,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019517835617661504,
+      "loss": 1.1833,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019516942720657174,
+      "loss": 1.0981,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019516049823652842,
+      "loss": 1.1206,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019515156926648512,
+      "loss": 0.9883,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019514264029644183,
+      "loss": 1.1855,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001951337113263985,
+      "loss": 1.2309,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001951247823563552,
+      "loss": 1.1593,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001951158533863119,
+      "loss": 1.0449,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001951069244162686,
+      "loss": 1.2579,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001950979954462253,
+      "loss": 1.0005,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000195089066476182,
+      "loss": 1.1238,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019508013750613867,
+      "loss": 1.0793,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019507120853609538,
+      "loss": 0.9978,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019506227956605208,
+      "loss": 1.0251,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019505335059600876,
+      "loss": 1.2308,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019504442162596546,
+      "loss": 1.1828,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019503549265592217,
+      "loss": 1.0849,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019502656368587884,
+      "loss": 1.2069,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019501763471583555,
+      "loss": 1.1877,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019500870574579225,
+      "loss": 1.055,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019499977677574893,
+      "loss": 1.1044,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019499084780570563,
+      "loss": 1.213,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019498191883566234,
+      "loss": 1.0674,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000194972989865619,
+      "loss": 1.3088,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019496406089557572,
+      "loss": 1.2221,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019495513192553242,
+      "loss": 1.0384,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001949462029554891,
+      "loss": 1.1386,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001949372739854458,
+      "loss": 1.0813,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019492834501540248,
+      "loss": 1.1652,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019491941604535918,
+      "loss": 0.9481,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019491048707531586,
+      "loss": 1.0426,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019490155810527256,
+      "loss": 1.0344,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019489262913522924,
+      "loss": 1.0495,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019488370016518595,
+      "loss": 1.3546,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019487477119514265,
+      "loss": 1.1557,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019486584222509933,
+      "loss": 1.0462,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019485691325505603,
+      "loss": 1.0635,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019484798428501273,
+      "loss": 1.0699,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001948390553149694,
+      "loss": 1.0518,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019483012634492612,
+      "loss": 1.1067,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019482119737488282,
+      "loss": 1.0471,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001948122684048395,
+      "loss": 1.1891,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001948033394347962,
+      "loss": 1.3282,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001947944104647529,
+      "loss": 1.0695,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019478548149470958,
+      "loss": 1.0158,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019477655252466629,
+      "loss": 1.1003,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000194767623554623,
+      "loss": 1.1861,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019475869458457967,
+      "loss": 1.0684,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019474976561453637,
+      "loss": 1.2522,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019474083664449307,
+      "loss": 1.262,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019473190767444975,
+      "loss": 1.1243,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019472297870440646,
+      "loss": 1.124,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019471404973436316,
+      "loss": 1.0224,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019470512076431984,
+      "loss": 1.2053,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019469619179427654,
+      "loss": 1.0592,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019468726282423324,
+      "loss": 1.2127,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019467833385418992,
+      "loss": 1.1606,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019466940488414663,
+      "loss": 1.1481,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019466047591410333,
+      "loss": 1.3502,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019465154694406,
+      "loss": 1.0737,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001946426179740167,
+      "loss": 1.1021,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019463368900397341,
+      "loss": 1.0353,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001946247600339301,
+      "loss": 1.0637,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001946158310638868,
+      "loss": 1.1079,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001946069020938435,
+      "loss": 1.1178,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019459797312380018,
+      "loss": 1.1987,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019458904415375688,
+      "loss": 1.2464,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019458011518371358,
+      "loss": 1.0282,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019457118621367026,
+      "loss": 0.9864,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019456225724362697,
+      "loss": 1.1112,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019455332827358367,
+      "loss": 1.0699,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019454439930354035,
+      "loss": 1.0047,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019453547033349705,
+      "loss": 1.0238,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019452654136345373,
+      "loss": 0.9608,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019451761239341043,
+      "loss": 0.9863,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019450868342336714,
+      "loss": 1.106,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001944997544533238,
+      "loss": 1.0505,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019449082548328052,
+      "loss": 1.2006,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001944818965132372,
+      "loss": 1.0864,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001944729675431939,
+      "loss": 1.1891,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019446403857315057,
+      "loss": 1.0449,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019445510960310728,
+      "loss": 1.0359,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019444618063306398,
+      "loss": 1.1603,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019443725166302066,
+      "loss": 1.1859,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019442832269297736,
+      "loss": 1.1241,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019441939372293407,
+      "loss": 1.0846,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019441046475289074,
+      "loss": 1.1351,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019440153578284745,
+      "loss": 1.0061,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019439260681280415,
+      "loss": 1.032,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019438367784276083,
+      "loss": 1.0888,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019437474887271753,
+      "loss": 1.0898,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019436581990267424,
+      "loss": 0.9749,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019435689093263091,
+      "loss": 1.0526,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019434796196258762,
+      "loss": 1.1788,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019433903299254432,
+      "loss": 1.0457,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000194330104022501,
+      "loss": 1.2757,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001943211750524577,
+      "loss": 1.1383,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001943122460824144,
+      "loss": 1.1204,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001943033171123711,
+      "loss": 1.1796,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001942943881423278,
+      "loss": 1.0128,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001942854591722845,
+      "loss": 1.1011,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001942765302022412,
+      "loss": 1.1136,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019426760123219787,
+      "loss": 1.2267,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019425867226215458,
+      "loss": 0.9373,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019424974329211128,
+      "loss": 1.0932,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019424081432206796,
+      "loss": 1.1325,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019423188535202466,
+      "loss": 1.1729,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019422295638198137,
+      "loss": 0.9857,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019421402741193804,
+      "loss": 1.0478,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019420509844189475,
+      "loss": 1.0538,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019419616947185145,
+      "loss": 1.2884,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019418724050180813,
+      "loss": 1.173,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019417831153176483,
+      "loss": 1.0828,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019416938256172154,
+      "loss": 1.1341,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001941604535916782,
+      "loss": 1.0864,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019415152462163492,
+      "loss": 1.0566,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019414259565159162,
+      "loss": 0.9446,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001941336666815483,
+      "loss": 1.1824,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000194124737711505,
+      "loss": 1.0703,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019411580874146168,
+      "loss": 1.1505,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019410687977141838,
+      "loss": 1.0517,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019409795080137506,
+      "loss": 1.0784,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019408902183133176,
+      "loss": 1.126,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019408009286128844,
+      "loss": 1.1202,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019407116389124514,
+      "loss": 1.0781,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019406223492120185,
+      "loss": 1.3039,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019405330595115853,
+      "loss": 1.0751,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019404437698111523,
+      "loss": 1.1252,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019403544801107193,
+      "loss": 1.1727,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001940265190410286,
+      "loss": 1.2147,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019401759007098531,
+      "loss": 1.0723,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019400866110094202,
+      "loss": 1.0795,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001939997321308987,
+      "loss": 1.1742,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001939908031608554,
+      "loss": 1.0615,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001939818741908121,
+      "loss": 1.0727,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019397294522076878,
+      "loss": 0.9977,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019396401625072548,
+      "loss": 1.0673,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001939550872806822,
+      "loss": 1.1154,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019394615831063887,
+      "loss": 0.9489,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019393722934059557,
+      "loss": 1.0597,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019392830037055227,
+      "loss": 1.2009,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019391937140050895,
+      "loss": 1.0525,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019391044243046565,
+      "loss": 1.054,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019390151346042236,
+      "loss": 1.2021,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019389258449037904,
+      "loss": 1.1478,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019388365552033574,
+      "loss": 1.0629,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019387472655029244,
+      "loss": 1.0262,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019386579758024912,
+      "loss": 0.9365,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019385686861020582,
+      "loss": 1.0423,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019384793964016253,
+      "loss": 1.0943,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001938390106701192,
+      "loss": 1.1029,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001938300817000759,
+      "loss": 1.0302,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019382115273003261,
+      "loss": 1.1665,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001938122237599893,
+      "loss": 1.089,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000193803294789946,
+      "loss": 1.0969,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001937943658199027,
+      "loss": 1.2506,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019378543684985938,
+      "loss": 1.0607,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019377650787981608,
+      "loss": 0.9543,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019376757890977278,
+      "loss": 1.3384,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019375864993972946,
+      "loss": 1.0749,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019374972096968616,
+      "loss": 1.2011,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019374079199964287,
+      "loss": 1.1143,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019373186302959955,
+      "loss": 1.1969,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019372293405955625,
+      "loss": 0.9453,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019371400508951293,
+      "loss": 1.1146,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019370507611946963,
+      "loss": 1.2121,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019369614714942633,
+      "loss": 1.0592,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000193687218179383,
+      "loss": 1.1535,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019367828920933972,
+      "loss": 1.2278,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001936693602392964,
+      "loss": 1.2289,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001936604312692531,
+      "loss": 1.054,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019365150229920977,
+      "loss": 1.1424,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019364257332916648,
+      "loss": 1.0983,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019363364435912318,
+      "loss": 1.0509,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019362471538907986,
+      "loss": 1.1156,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019361578641903656,
+      "loss": 1.1933,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019360685744899327,
+      "loss": 1.3011,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019359792847894994,
+      "loss": 1.0059,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019358899950890665,
+      "loss": 1.0459,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019358007053886335,
+      "loss": 1.2204,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019357114156882003,
+      "loss": 1.1711,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019356221259877673,
+      "loss": 1.0916,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019355328362873344,
+      "loss": 1.0518,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019354435465869011,
+      "loss": 1.2195,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019353542568864682,
+      "loss": 1.0173,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019352649671860352,
+      "loss": 1.1156,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001935175677485602,
+      "loss": 1.0947,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001935086387785169,
+      "loss": 1.1008,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001934997098084736,
+      "loss": 1.1187,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019349078083843028,
+      "loss": 1.068,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000193481851868387,
+      "loss": 1.0084,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001934729228983437,
+      "loss": 1.0099,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019346399392830037,
+      "loss": 1.1626,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019345506495825707,
+      "loss": 1.1454,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019344613598821378,
+      "loss": 1.0593,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019343720701817048,
+      "loss": 1.0752,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019342827804812716,
+      "loss": 1.3326,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019341934907808386,
+      "loss": 1.203,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019341042010804057,
+      "loss": 1.1941,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019340149113799724,
+      "loss": 0.9387,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019339256216795395,
+      "loss": 1.1272,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019338363319791065,
+      "loss": 1.3295,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019337470422786733,
+      "loss": 1.0929,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019336577525782403,
+      "loss": 1.0641,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019335684628778074,
+      "loss": 1.1334,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001933479173177374,
+      "loss": 1.1095,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019333898834769412,
+      "loss": 1.0996,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019333005937765082,
+      "loss": 1.0795,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001933211304076075,
+      "loss": 1.0383,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001933122014375642,
+      "loss": 1.1401,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019330327246752088,
+      "loss": 1.0754,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019329434349747758,
+      "loss": 1.0919,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019328541452743426,
+      "loss": 1.1912,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019327648555739096,
+      "loss": 1.0907,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019326755658734764,
+      "loss": 1.3276,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019325862761730434,
+      "loss": 1.0776,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019324969864726105,
+      "loss": 1.0192,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019324076967721773,
+      "loss": 1.0919,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019323184070717443,
+      "loss": 1.147,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019322291173713113,
+      "loss": 1.005,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001932139827670878,
+      "loss": 1.046,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019320505379704451,
+      "loss": 1.1003,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019319612482700122,
+      "loss": 1.1133,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001931871958569579,
+      "loss": 1.1245,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001931782668869146,
+      "loss": 1.1876,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001931693379168713,
+      "loss": 1.0457,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019316040894682798,
+      "loss": 1.081,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019315147997678468,
+      "loss": 1.1581,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001931425510067414,
+      "loss": 1.152,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019313362203669807,
+      "loss": 1.1013,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019312469306665477,
+      "loss": 1.0642,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019311576409661147,
+      "loss": 1.1292,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019310683512656815,
+      "loss": 1.1414,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019309790615652485,
+      "loss": 1.0949,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019308897718648156,
+      "loss": 1.0952,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019308004821643824,
+      "loss": 1.2136,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019307111924639494,
+      "loss": 1.2545,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019306219027635164,
+      "loss": 1.0853,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019305326130630832,
+      "loss": 1.0653,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019304433233626502,
+      "loss": 1.0563,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019303540336622173,
+      "loss": 0.9844,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001930264743961784,
+      "loss": 1.1967,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001930175454261351,
+      "loss": 1.1218,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001930086164560918,
+      "loss": 1.0536,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001929996874860485,
+      "loss": 1.0549,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001929907585160052,
+      "loss": 1.0852,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001929818295459619,
+      "loss": 1.0306,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019297290057591858,
+      "loss": 1.125,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019296397160587528,
+      "loss": 1.0148,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019295504263583198,
+      "loss": 1.0671,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019294611366578866,
+      "loss": 0.9267,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019293718469574536,
+      "loss": 1.3347,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019292825572570207,
+      "loss": 1.1345,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019291932675565875,
+      "loss": 1.023,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019291039778561545,
+      "loss": 0.9919,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019290146881557213,
+      "loss": 1.0385,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019289253984552883,
+      "loss": 1.0601,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019288361087548553,
+      "loss": 1.0003,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001928746819054422,
+      "loss": 1.1651,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019286575293539892,
+      "loss": 1.0731,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001928568239653556,
+      "loss": 1.0274,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001928478949953123,
+      "loss": 1.082,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019283896602526897,
+      "loss": 1.1182,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019283003705522568,
+      "loss": 1.1256,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019282110808518238,
+      "loss": 1.1411,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019281217911513906,
+      "loss": 1.0931,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019280325014509576,
+      "loss": 1.1314,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019279432117505247,
+      "loss": 1.068,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019278539220500914,
+      "loss": 1.1765,
+      "step": 813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019277646323496585,
+      "loss": 1.1155,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019276753426492255,
+      "loss": 1.086,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019275860529487923,
+      "loss": 1.3056,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019274967632483593,
+      "loss": 1.1102,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019274074735479264,
+      "loss": 1.0566,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001927318183847493,
+      "loss": 1.1534,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019272288941470602,
+      "loss": 1.0511,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019271396044466272,
+      "loss": 1.1644,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001927050314746194,
+      "loss": 1.1442,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001926961025045761,
+      "loss": 1.0733,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001926871735345328,
+      "loss": 1.1633,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019267824456448948,
+      "loss": 1.1345,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001926693155944462,
+      "loss": 1.0429,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001926603866244029,
+      "loss": 1.2584,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019265145765435957,
+      "loss": 1.3028,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019264252868431627,
+      "loss": 1.0353,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019263359971427298,
+      "loss": 1.1123,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019262467074422965,
+      "loss": 1.3639,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019261574177418636,
+      "loss": 1.1185,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019260681280414306,
+      "loss": 1.1174,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019259788383409974,
+      "loss": 1.0137,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019258895486405644,
+      "loss": 1.245,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019258002589401315,
+      "loss": 1.1086,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019257109692396985,
+      "loss": 1.1065,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019256216795392653,
+      "loss": 0.9867,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019255323898388323,
+      "loss": 1.0694,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019254431001383994,
+      "loss": 1.143,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001925353810437966,
+      "loss": 1.1503,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019252645207375332,
+      "loss": 1.0791,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019251752310371002,
+      "loss": 1.172,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001925085941336667,
+      "loss": 1.0786,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001924996651636234,
+      "loss": 1.0481,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019249073619358008,
+      "loss": 1.1642,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019248180722353678,
+      "loss": 1.1728,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019247287825349346,
+      "loss": 1.1532,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019246394928345016,
+      "loss": 1.2802,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019245502031340684,
+      "loss": 1.187,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019244609134336354,
+      "loss": 1.2798,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019243716237332025,
+      "loss": 1.0301,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019242823340327693,
+      "loss": 0.9966,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019241930443323363,
+      "loss": 1.1647,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019241037546319033,
+      "loss": 1.1553,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000192401446493147,
+      "loss": 1.0901,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019239251752310371,
+      "loss": 1.0376,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019238358855306042,
+      "loss": 1.0858,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001923746595830171,
+      "loss": 1.1146,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001923657306129738,
+      "loss": 1.0223,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001923568016429305,
+      "loss": 1.2396,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019234787267288718,
+      "loss": 1.0771,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019233894370284388,
+      "loss": 1.2144,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001923300147328006,
+      "loss": 1.1101,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019232108576275727,
+      "loss": 1.0963,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019231215679271397,
+      "loss": 1.1878,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019230322782267067,
+      "loss": 1.0603,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019229429885262735,
+      "loss": 1.2114,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019228536988258405,
+      "loss": 1.2244,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019227644091254076,
+      "loss": 1.0878,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019226751194249744,
+      "loss": 1.2395,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019225858297245414,
+      "loss": 1.1224,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019224965400241084,
+      "loss": 1.0042,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019224072503236752,
+      "loss": 1.1305,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019223179606232422,
+      "loss": 0.9384,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019222286709228093,
+      "loss": 1.1119,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001922139381222376,
+      "loss": 1.0252,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001922050091521943,
+      "loss": 1.2241,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000192196080182151,
+      "loss": 1.1232,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001921871512121077,
+      "loss": 1.071,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001921782222420644,
+      "loss": 1.0338,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001921692932720211,
+      "loss": 1.1461,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019216036430197777,
+      "loss": 1.0808,
+      "step": 883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019215143533193448,
+      "loss": 1.0964,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019214250636189118,
+      "loss": 1.2022,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019213357739184786,
+      "loss": 1.0956,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019212464842180456,
+      "loss": 1.1831,
+      "step": 887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019211571945176127,
+      "loss": 1.0739,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019210679048171794,
+      "loss": 1.1632,
+      "step": 889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019209786151167465,
+      "loss": 1.2217,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019208893254163133,
+      "loss": 1.2032,
+      "step": 891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019208000357158803,
+      "loss": 1.0836,
+      "step": 892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001920710746015447,
+      "loss": 1.0203,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001920621456315014,
+      "loss": 1.0371,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019205321666145811,
+      "loss": 1.0851,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001920442876914148,
+      "loss": 1.096,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001920353587213715,
+      "loss": 1.0648,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019202642975132817,
+      "loss": 1.0663,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019201750078128488,
+      "loss": 1.1852,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019200857181124158,
+      "loss": 1.1648,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019199964284119826,
+      "loss": 0.9831,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019199071387115496,
+      "loss": 1.1719,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019198178490111167,
+      "loss": 1.0196,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019197285593106834,
+      "loss": 0.9952,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019196392696102505,
+      "loss": 1.2351,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019195499799098175,
+      "loss": 1.0032,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019194606902093843,
+      "loss": 0.9848,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019193714005089513,
+      "loss": 1.0279,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019192821108085184,
+      "loss": 1.0402,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001919192821108085,
+      "loss": 1.1582,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019191035314076522,
+      "loss": 1.3243,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019190142417072192,
+      "loss": 1.1033,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001918924952006786,
+      "loss": 1.0296,
+      "step": 913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001918835662306353,
+      "loss": 1.0527,
+      "step": 914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000191874637260592,
+      "loss": 1.0852,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019186570829054868,
+      "loss": 1.1557,
+      "step": 916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001918567793205054,
+      "loss": 1.1715,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001918478503504621,
+      "loss": 1.1241,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019183892138041877,
+      "loss": 1.0717,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019182999241037547,
+      "loss": 1.1058,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019182106344033218,
+      "loss": 1.2059,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019181213447028885,
+      "loss": 1.1235,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019180320550024556,
+      "loss": 1.1173,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019179427653020226,
+      "loss": 1.0708,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019178534756015894,
+      "loss": 1.091,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019177641859011564,
+      "loss": 1.1283,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019176748962007235,
+      "loss": 1.0125,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019175856065002902,
+      "loss": 1.2409,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019174963167998573,
+      "loss": 1.0,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019174070270994243,
+      "loss": 1.1434,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001917317737398991,
+      "loss": 1.1221,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001917228447698558,
+      "loss": 1.0285,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019171391579981252,
+      "loss": 1.317,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001917049868297692,
+      "loss": 1.1898,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916960578597259,
+      "loss": 0.9788,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916871288896826,
+      "loss": 1.0775,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019167819991963928,
+      "loss": 0.989,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019166927094959598,
+      "loss": 1.1071,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019166034197955266,
+      "loss": 1.0521,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019165141300950936,
+      "loss": 1.1003,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019164248403946604,
+      "loss": 1.0177,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019163355506942274,
+      "loss": 1.1307,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019162462609937945,
+      "loss": 1.1057,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019161569712933612,
+      "loss": 1.1324,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019160676815929283,
+      "loss": 1.2197,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019159783918924953,
+      "loss": 1.105,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915889102192062,
+      "loss": 0.9955,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019157998124916291,
+      "loss": 1.0553,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019157105227911962,
+      "loss": 1.1093,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915621233090763,
+      "loss": 1.0656,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000191553194339033,
+      "loss": 1.1723,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915442653689897,
+      "loss": 1.0499,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019153533639894638,
+      "loss": 1.149,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019152640742890308,
+      "loss": 1.0697,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915174784588598,
+      "loss": 1.1082,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019150854948881646,
+      "loss": 1.0396,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019149962051877317,
+      "loss": 0.974,
+      "step": 957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019149069154872987,
+      "loss": 1.0174,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019148176257868655,
+      "loss": 1.2519,
+      "step": 959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019147283360864325,
+      "loss": 1.1317,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019146390463859996,
+      "loss": 1.1335,
+      "step": 961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019145497566855663,
+      "loss": 1.0879,
+      "step": 962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019144604669851334,
+      "loss": 1.1668,
+      "step": 963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019143711772847004,
+      "loss": 1.1265,
+      "step": 964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019142818875842672,
+      "loss": 1.047,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019141925978838342,
+      "loss": 0.9314,
+      "step": 966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019141033081834013,
+      "loss": 1.0528,
+      "step": 967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914014018482968,
+      "loss": 1.0207,
+      "step": 968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913924728782535,
+      "loss": 1.128,
+      "step": 969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913835439082102,
+      "loss": 0.9644,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913746149381669,
+      "loss": 1.0701,
+      "step": 971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913656859681236,
+      "loss": 1.0736,
+      "step": 972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913567569980803,
+      "loss": 1.1152,
+      "step": 973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019134782802803697,
+      "loss": 1.0245,
+      "step": 974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019133889905799368,
+      "loss": 1.0933,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019132997008795038,
+      "loss": 1.1494,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019132104111790706,
+      "loss": 1.1207,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019131211214786376,
+      "loss": 1.0804,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019130318317782047,
+      "loss": 1.082,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019129425420777714,
+      "loss": 1.0288,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019128532523773385,
+      "loss": 1.0945,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019127639626769053,
+      "loss": 1.1455,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019126746729764723,
+      "loss": 1.0976,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912585383276039,
+      "loss": 1.1266,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912496093575606,
+      "loss": 0.976,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019124068038751731,
+      "loss": 1.0685,
+      "step": 986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000191231751417474,
+      "loss": 1.1012,
+      "step": 987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912228224474307,
+      "loss": 1.2081,
+      "step": 988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019121389347738737,
+      "loss": 1.0067,
+      "step": 989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019120496450734408,
+      "loss": 1.1001,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019119603553730078,
+      "loss": 0.9839,
+      "step": 991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019118710656725746,
+      "loss": 1.2443,
+      "step": 992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019117817759721416,
+      "loss": 1.0098,
+      "step": 993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019116924862717087,
+      "loss": 0.9506,
+      "step": 994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019116031965712754,
+      "loss": 0.9879,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019115139068708425,
+      "loss": 1.0746,
+      "step": 996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019114246171704095,
+      "loss": 0.9337,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019113353274699763,
+      "loss": 1.0634,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019112460377695433,
+      "loss": 0.9779,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019111567480691104,
+      "loss": 1.0939,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001911067458368677,
+      "loss": 1.198,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019109781686682442,
+      "loss": 1.1178,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019108888789678112,
+      "loss": 1.0804,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001910799589267378,
+      "loss": 1.1364,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001910710299566945,
+      "loss": 1.1703,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001910621009866512,
+      "loss": 1.0827,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019105317201660788,
+      "loss": 1.0518,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001910442430465646,
+      "loss": 1.0542,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001910353140765213,
+      "loss": 1.225,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019102638510647797,
+      "loss": 1.1,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019101745613643467,
+      "loss": 1.1671,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019100852716639138,
+      "loss": 1.1653,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019099959819634805,
+      "loss": 1.1043,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019099066922630476,
+      "loss": 1.0544,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019098174025626146,
+      "loss": 1.0538,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019097281128621814,
+      "loss": 1.1474,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019096388231617484,
+      "loss": 1.1026,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019095495334613155,
+      "loss": 1.0914,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019094602437608822,
+      "loss": 1.1087,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019093709540604493,
+      "loss": 1.235,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019092816643600163,
+      "loss": 1.1576,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001909192374659583,
+      "loss": 0.9873,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190910308495915,
+      "loss": 1.0782,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019090137952587172,
+      "loss": 0.9799,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001908924505558284,
+      "loss": 1.1695,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001908835215857851,
+      "loss": 1.0952,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001908745926157418,
+      "loss": 1.1832,
+      "step": 1027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019086566364569848,
+      "loss": 1.2288,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019085673467565518,
+      "loss": 1.136,
+      "step": 1029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019084780570561186,
+      "loss": 1.059,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019083887673556856,
+      "loss": 1.0675,
+      "step": 1031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019082994776552524,
+      "loss": 1.1572,
+      "step": 1032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019082101879548194,
+      "loss": 1.0589,
+      "step": 1033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019081208982543865,
+      "loss": 1.1536,
+      "step": 1034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019080316085539532,
+      "loss": 1.1086,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019079423188535203,
+      "loss": 1.0283,
+      "step": 1036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019078530291530873,
+      "loss": 1.1138,
+      "step": 1037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001907763739452654,
+      "loss": 1.0449,
+      "step": 1038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001907674449752221,
+      "loss": 1.0653,
+      "step": 1039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019075851600517882,
+      "loss": 0.9209,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001907495870351355,
+      "loss": 1.0279,
+      "step": 1041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001907406580650922,
+      "loss": 0.9992,
+      "step": 1042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001907317290950489,
+      "loss": 1.1081,
+      "step": 1043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019072280012500558,
+      "loss": 1.0044,
+      "step": 1044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019071387115496228,
+      "loss": 1.0794,
+      "step": 1045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190704942184919,
+      "loss": 0.9354,
+      "step": 1046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019069601321487566,
+      "loss": 1.1447,
+      "step": 1047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019068708424483237,
+      "loss": 1.155,
+      "step": 1048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019067815527478907,
+      "loss": 1.1168,
+      "step": 1049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019066922630474575,
+      "loss": 0.9854,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019066029733470245,
+      "loss": 1.0458,
+      "step": 1051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019065136836465916,
+      "loss": 1.1326,
+      "step": 1052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019064243939461583,
+      "loss": 1.0759,
+      "step": 1053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019063351042457254,
+      "loss": 1.0238,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019062458145452924,
+      "loss": 0.9852,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019061565248448592,
+      "loss": 1.1411,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019060672351444262,
+      "loss": 0.9758,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019059779454439933,
+      "loss": 1.2076,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190588865574356,
+      "loss": 1.0836,
+      "step": 1059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905799366043127,
+      "loss": 1.0858,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905710076342694,
+      "loss": 1.0517,
+      "step": 1061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905620786642261,
+      "loss": 1.0461,
+      "step": 1062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905531496941828,
+      "loss": 1.0848,
+      "step": 1063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905442207241395,
+      "loss": 0.9242,
+      "step": 1064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019053529175409617,
+      "loss": 1.1632,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019052636278405288,
+      "loss": 0.9846,
+      "step": 1066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019051743381400958,
+      "loss": 1.0942,
+      "step": 1067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019050850484396626,
+      "loss": 1.0968,
+      "step": 1068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019049957587392296,
+      "loss": 1.1346,
+      "step": 1069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019049064690387967,
+      "loss": 1.085,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019048171793383634,
+      "loss": 1.095,
+      "step": 1071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019047278896379305,
+      "loss": 1.0259,
+      "step": 1072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019046385999374973,
+      "loss": 1.1583,
+      "step": 1073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019045493102370643,
+      "loss": 1.0636,
+      "step": 1074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904460020536631,
+      "loss": 1.144,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904370730836198,
+      "loss": 1.2091,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019042814411357651,
+      "loss": 1.2063,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904192151435332,
+      "loss": 0.9964,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904102861734899,
+      "loss": 1.073,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019040135720344657,
+      "loss": 1.0404,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019039242823340328,
+      "loss": 1.2694,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019038349926335998,
+      "loss": 1.209,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019037457029331666,
+      "loss": 1.0819,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019036564132327336,
+      "loss": 0.9755,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019035671235323007,
+      "loss": 1.0776,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019034778338318674,
+      "loss": 1.1396,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019033885441314345,
+      "loss": 1.0682,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019032992544310015,
+      "loss": 1.2221,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019032099647305683,
+      "loss": 1.2487,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019031206750301353,
+      "loss": 1.1666,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019030313853297024,
+      "loss": 1.1214,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001902942095629269,
+      "loss": 0.9444,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019028528059288362,
+      "loss": 1.1715,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019027635162284032,
+      "loss": 1.0756,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000190267422652797,
+      "loss": 0.9242,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001902584936827537,
+      "loss": 1.1869,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001902495647127104,
+      "loss": 1.0688,
+      "step": 1097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019024063574266708,
+      "loss": 1.1351,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019023170677262379,
+      "loss": 1.0496,
+      "step": 1099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001902227778025805,
+      "loss": 0.9788,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019021384883253717,
+      "loss": 1.0222,
+      "step": 1101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019020491986249387,
+      "loss": 1.0427,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019019599089245058,
+      "loss": 1.1749,
+      "step": 1103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019018706192240725,
+      "loss": 1.4029,
+      "step": 1104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019017813295236396,
+      "loss": 1.0572,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019016920398232066,
+      "loss": 1.2115,
+      "step": 1106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019016027501227734,
+      "loss": 0.9934,
+      "step": 1107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019015134604223404,
+      "loss": 1.0956,
+      "step": 1108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019014241707219075,
+      "loss": 1.1054,
+      "step": 1109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019013348810214742,
+      "loss": 1.0308,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019012455913210413,
+      "loss": 1.0738,
+      "step": 1111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019011563016206083,
+      "loss": 1.1356,
+      "step": 1112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001901067011920175,
+      "loss": 0.9825,
+      "step": 1113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001900977722219742,
+      "loss": 1.0208,
+      "step": 1114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019008884325193092,
+      "loss": 0.9612,
+      "step": 1115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001900799142818876,
+      "loss": 1.0412,
+      "step": 1116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001900709853118443,
+      "loss": 1.0304,
+      "step": 1117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000190062056341801,
+      "loss": 1.1187,
+      "step": 1118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019005312737175768,
+      "loss": 1.2028,
+      "step": 1119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019004419840171438,
+      "loss": 1.1061,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019003526943167106,
+      "loss": 0.9646,
+      "step": 1121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019002634046162776,
+      "loss": 1.2105,
+      "step": 1122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019001741149158444,
+      "loss": 0.95,
+      "step": 1123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019000848252154114,
+      "loss": 1.1139,
+      "step": 1124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018999955355149782,
+      "loss": 1.1557,
+      "step": 1125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018999062458145452,
+      "loss": 1.0483,
+      "step": 1126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018998169561141123,
+      "loss": 1.3048,
+      "step": 1127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001899727666413679,
+      "loss": 1.0531,
+      "step": 1128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001899638376713246,
+      "loss": 1.1486,
+      "step": 1129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001899549087012813,
+      "loss": 1.0655,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018994597973123802,
+      "loss": 1.1459,
+      "step": 1131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001899370507611947,
+      "loss": 1.1075,
+      "step": 1132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001899281217911514,
+      "loss": 1.0562,
+      "step": 1133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001899191928211081,
+      "loss": 1.005,
+      "step": 1134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018991026385106478,
+      "loss": 1.3214,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018990133488102148,
+      "loss": 1.0951,
+      "step": 1136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001898924059109782,
+      "loss": 1.1527,
+      "step": 1137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018988347694093486,
+      "loss": 0.9409,
+      "step": 1138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018987454797089157,
+      "loss": 1.1183,
+      "step": 1139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018986561900084827,
+      "loss": 1.2352,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018985669003080495,
+      "loss": 1.0291,
+      "step": 1141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018984776106076165,
+      "loss": 1.0108,
+      "step": 1142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018983883209071836,
+      "loss": 1.0544,
+      "step": 1143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018982990312067503,
+      "loss": 1.2588,
+      "step": 1144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018982097415063174,
+      "loss": 1.0652,
+      "step": 1145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018981204518058844,
+      "loss": 1.0236,
+      "step": 1146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018980311621054512,
+      "loss": 1.2314,
+      "step": 1147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018979418724050182,
+      "loss": 1.2368,
+      "step": 1148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018978525827045853,
+      "loss": 1.2788,
+      "step": 1149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897763293004152,
+      "loss": 1.2507,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897674003303719,
+      "loss": 1.0722,
+      "step": 1151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897584713603286,
+      "loss": 0.9941,
+      "step": 1152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897495423902853,
+      "loss": 1.0029,
+      "step": 1153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000189740613420242,
+      "loss": 1.2551,
+      "step": 1154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897316844501987,
+      "loss": 1.0167,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018972275548015537,
+      "loss": 1.1302,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018971382651011208,
+      "loss": 0.9723,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018970489754006878,
+      "loss": 1.1488,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018969596857002546,
+      "loss": 1.236,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018968703959998216,
+      "loss": 1.0525,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018967811062993887,
+      "loss": 1.0477,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018966918165989554,
+      "loss": 1.1773,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018966025268985225,
+      "loss": 1.0443,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018965132371980892,
+      "loss": 1.0747,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018964239474976563,
+      "loss": 1.0192,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001896334657797223,
+      "loss": 1.0628,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000189624536809679,
+      "loss": 1.1905,
+      "step": 1167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018961560783963571,
+      "loss": 1.0212,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001896066788695924,
+      "loss": 1.0707,
+      "step": 1169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001895977498995491,
+      "loss": 1.2125,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018958882092950577,
+      "loss": 1.1428,
+      "step": 1171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018957989195946248,
+      "loss": 1.0247,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018957096298941918,
+      "loss": 1.119,
+      "step": 1173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018956203401937586,
+      "loss": 1.0945,
+      "step": 1174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018955310504933256,
+      "loss": 1.1672,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018954417607928926,
+      "loss": 1.14,
+      "step": 1176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018953524710924594,
+      "loss": 1.0757,
+      "step": 1177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018952631813920265,
+      "loss": 1.0837,
+      "step": 1178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018951738916915935,
+      "loss": 1.045,
+      "step": 1179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018950846019911603,
+      "loss": 1.1135,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018949953122907273,
+      "loss": 0.9938,
+      "step": 1181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018949060225902943,
+      "loss": 1.1236,
+      "step": 1182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894816732889861,
+      "loss": 1.0867,
+      "step": 1183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018947274431894282,
+      "loss": 1.2641,
+      "step": 1184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018946381534889952,
+      "loss": 1.1798,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894548863788562,
+      "loss": 1.0673,
+      "step": 1186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894459574088129,
+      "loss": 1.0855,
+      "step": 1187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894370284387696,
+      "loss": 0.9809,
+      "step": 1188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018942809946872628,
+      "loss": 0.9894,
+      "step": 1189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018941917049868299,
+      "loss": 1.1187,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001894102415286397,
+      "loss": 1.2227,
+      "step": 1191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018940131255859637,
+      "loss": 1.1124,
+      "step": 1192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018939238358855307,
+      "loss": 1.0829,
+      "step": 1193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018938345461850977,
+      "loss": 1.0616,
+      "step": 1194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018937452564846645,
+      "loss": 1.1673,
+      "step": 1195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018936559667842316,
+      "loss": 1.0209,
+      "step": 1196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018935666770837986,
+      "loss": 1.1692,
+      "step": 1197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018934773873833654,
+      "loss": 1.1308,
+      "step": 1198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018933880976829324,
+      "loss": 0.9718,
+      "step": 1199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018932988079824994,
+      "loss": 1.0407,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018932095182820662,
+      "loss": 1.0468,
+      "step": 1201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018931202285816333,
+      "loss": 1.0689,
+      "step": 1202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018930309388812003,
+      "loss": 1.1136,
+      "step": 1203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001892941649180767,
+      "loss": 1.0456,
+      "step": 1204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001892852359480334,
+      "loss": 1.1538,
+      "step": 1205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018927630697799011,
+      "loss": 1.193,
+      "step": 1206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001892673780079468,
+      "loss": 0.9986,
+      "step": 1207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001892584490379035,
+      "loss": 1.035,
+      "step": 1208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001892495200678602,
+      "loss": 1.1457,
+      "step": 1209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018924059109781688,
+      "loss": 1.0261,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018923166212777358,
+      "loss": 1.0816,
+      "step": 1211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018922273315773026,
+      "loss": 1.1566,
+      "step": 1212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018921380418768696,
+      "loss": 1.0677,
+      "step": 1213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018920487521764364,
+      "loss": 1.0744,
+      "step": 1214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018919594624760034,
+      "loss": 1.1006,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018918701727755702,
+      "loss": 1.1518,
+      "step": 1216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018917808830751372,
+      "loss": 1.0238,
+      "step": 1217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018916915933747043,
+      "loss": 1.1127,
+      "step": 1218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891602303674271,
+      "loss": 1.2207,
+      "step": 1219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891513013973838,
+      "loss": 0.9582,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891423724273405,
+      "loss": 0.9864,
+      "step": 1221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891334434572972,
+      "loss": 1.099,
+      "step": 1222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891245144872539,
+      "loss": 0.9926,
+      "step": 1223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001891155855172106,
+      "loss": 1.1433,
+      "step": 1224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018910665654716727,
+      "loss": 1.0693,
+      "step": 1225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018909772757712398,
+      "loss": 1.0546,
+      "step": 1226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018908879860708068,
+      "loss": 1.009,
+      "step": 1227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001890798696370374,
+      "loss": 0.9932,
+      "step": 1228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018907094066699406,
+      "loss": 1.306,
+      "step": 1229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018906201169695077,
+      "loss": 1.1375,
+      "step": 1230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018905308272690747,
+      "loss": 1.0412,
+      "step": 1231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018904415375686415,
+      "loss": 1.2383,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018903522478682085,
+      "loss": 1.0621,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018902629581677756,
+      "loss": 1.0617,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018901736684673423,
+      "loss": 0.9142,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018900843787669094,
+      "loss": 1.1066,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018899950890664764,
+      "loss": 1.1083,
+      "step": 1237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018899057993660432,
+      "loss": 1.0163,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018898165096656102,
+      "loss": 1.1448,
+      "step": 1239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018897272199651773,
+      "loss": 1.1766,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001889637930264744,
+      "loss": 1.1067,
+      "step": 1241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001889548640564311,
+      "loss": 1.0754,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001889459350863878,
+      "loss": 1.0666,
+      "step": 1243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001889370061163445,
+      "loss": 1.0236,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001889280771463012,
+      "loss": 1.0762,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001889191481762579,
+      "loss": 1.0616,
+      "step": 1246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018891021920621457,
+      "loss": 1.0215,
+      "step": 1247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018890129023617128,
+      "loss": 1.0488,
+      "step": 1248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018889236126612798,
+      "loss": 1.1517,
+      "step": 1249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018888343229608466,
+      "loss": 1.0828,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018887450332604136,
+      "loss": 1.0554,
+      "step": 1251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018886557435599807,
+      "loss": 1.0191,
+      "step": 1252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018885664538595474,
+      "loss": 1.0641,
+      "step": 1253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018884771641591145,
+      "loss": 1.1357,
+      "step": 1254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018883878744586812,
+      "loss": 1.0008,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018882985847582483,
+      "loss": 1.1661,
+      "step": 1256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001888209295057815,
+      "loss": 1.1203,
+      "step": 1257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001888120005357382,
+      "loss": 1.0451,
+      "step": 1258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001888030715656949,
+      "loss": 0.9404,
+      "step": 1259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001887941425956516,
+      "loss": 1.0472,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001887852136256083,
+      "loss": 1.0722,
+      "step": 1261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018877628465556497,
+      "loss": 0.9078,
+      "step": 1262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018876735568552168,
+      "loss": 1.0526,
+      "step": 1263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018875842671547838,
+      "loss": 1.125,
+      "step": 1264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018874949774543506,
+      "loss": 0.9669,
+      "step": 1265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018874056877539176,
+      "loss": 1.1713,
+      "step": 1266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018873163980534846,
+      "loss": 0.9501,
+      "step": 1267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018872271083530514,
+      "loss": 1.0511,
+      "step": 1268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018871378186526185,
+      "loss": 0.985,
+      "step": 1269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018870485289521855,
+      "loss": 1.1112,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018869592392517523,
+      "loss": 1.0708,
+      "step": 1271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018868699495513193,
+      "loss": 1.2408,
+      "step": 1272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018867806598508863,
+      "loss": 1.0954,
+      "step": 1273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001886691370150453,
+      "loss": 1.2186,
+      "step": 1274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018866020804500202,
+      "loss": 1.0672,
+      "step": 1275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018865127907495872,
+      "loss": 1.207,
+      "step": 1276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001886423501049154,
+      "loss": 1.0757,
+      "step": 1277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001886334211348721,
+      "loss": 1.0111,
+      "step": 1278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001886244921648288,
+      "loss": 1.1706,
+      "step": 1279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018861556319478548,
+      "loss": 1.07,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018860663422474219,
+      "loss": 1.1249,
+      "step": 1281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001885977052546989,
+      "loss": 1.0524,
+      "step": 1282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018858877628465557,
+      "loss": 1.1182,
+      "step": 1283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018857984731461227,
+      "loss": 1.0702,
+      "step": 1284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018857091834456897,
+      "loss": 1.0068,
+      "step": 1285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018856198937452565,
+      "loss": 1.0154,
+      "step": 1286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018855306040448236,
+      "loss": 1.0468,
+      "step": 1287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018854413143443906,
+      "loss": 1.0536,
+      "step": 1288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018853520246439574,
+      "loss": 1.1509,
+      "step": 1289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018852627349435244,
+      "loss": 0.9655,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018851734452430914,
+      "loss": 1.0766,
+      "step": 1291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018850841555426582,
+      "loss": 1.1278,
+      "step": 1292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018849948658422253,
+      "loss": 1.0629,
+      "step": 1293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018849055761417923,
+      "loss": 1.0756,
+      "step": 1294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001884816286441359,
+      "loss": 1.1028,
+      "step": 1295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001884726996740926,
+      "loss": 0.9934,
+      "step": 1296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018846377070404931,
+      "loss": 1.1863,
+      "step": 1297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000188454841734006,
+      "loss": 1.0397,
+      "step": 1298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001884459127639627,
+      "loss": 1.0109,
+      "step": 1299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001884369837939194,
+      "loss": 1.0304,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018842805482387608,
+      "loss": 1.156,
+      "step": 1301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018841912585383278,
+      "loss": 1.1066,
+      "step": 1302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018841019688378946,
+      "loss": 1.0275,
+      "step": 1303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018840126791374616,
+      "loss": 1.1184,
+      "step": 1304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018839233894370284,
+      "loss": 1.0778,
+      "step": 1305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018838340997365954,
+      "loss": 1.0603,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018837448100361622,
+      "loss": 1.0736,
+      "step": 1307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018836555203357292,
+      "loss": 1.0057,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018835662306352963,
+      "loss": 1.2999,
+      "step": 1309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883476940934863,
+      "loss": 1.0497,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188338765123443,
+      "loss": 1.0671,
+      "step": 1311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883298361533997,
+      "loss": 1.0169,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883209071833564,
+      "loss": 1.0946,
+      "step": 1313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883119782133131,
+      "loss": 1.1463,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883030492432698,
+      "loss": 1.1367,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018829412027322647,
+      "loss": 1.006,
+      "step": 1316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018828519130318318,
+      "loss": 0.9563,
+      "step": 1317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018827626233313988,
+      "loss": 1.0333,
+      "step": 1318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018826733336309656,
+      "loss": 1.0941,
+      "step": 1319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018825840439305326,
+      "loss": 1.0963,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018824947542300997,
+      "loss": 1.2415,
+      "step": 1321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018824054645296664,
+      "loss": 1.1418,
+      "step": 1322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018823161748292335,
+      "loss": 1.0501,
+      "step": 1323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018822268851288005,
+      "loss": 1.0047,
+      "step": 1324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018821375954283673,
+      "loss": 1.0817,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018820483057279343,
+      "loss": 1.1534,
+      "step": 1326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018819590160275014,
+      "loss": 1.1765,
+      "step": 1327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018818697263270684,
+      "loss": 1.1083,
+      "step": 1328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018817804366266352,
+      "loss": 1.1787,
+      "step": 1329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018816911469262022,
+      "loss": 1.0903,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018816018572257693,
+      "loss": 1.0339,
+      "step": 1331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881512567525336,
+      "loss": 1.0294,
+      "step": 1332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881423277824903,
+      "loss": 1.0924,
+      "step": 1333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188133398812447,
+      "loss": 1.0231,
+      "step": 1334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881244698424037,
+      "loss": 1.0337,
+      "step": 1335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881155408723604,
+      "loss": 1.1063,
+      "step": 1336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881066119023171,
+      "loss": 1.2093,
+      "step": 1337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018809768293227377,
+      "loss": 0.9054,
+      "step": 1338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018808875396223048,
+      "loss": 1.1523,
+      "step": 1339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018807982499218718,
+      "loss": 1.0453,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018807089602214386,
+      "loss": 1.1392,
+      "step": 1341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018806196705210056,
+      "loss": 1.1225,
+      "step": 1342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018805303808205727,
+      "loss": 1.0783,
+      "step": 1343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018804410911201394,
+      "loss": 1.0532,
+      "step": 1344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018803518014197065,
+      "loss": 1.0518,
+      "step": 1345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018802625117192732,
+      "loss": 1.0784,
+      "step": 1346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018801732220188403,
+      "loss": 0.9979,
+      "step": 1347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001880083932318407,
+      "loss": 1.2113,
+      "step": 1348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879994642617974,
+      "loss": 0.9883,
+      "step": 1349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879905352917541,
+      "loss": 0.9692,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879816063217108,
+      "loss": 1.0584,
+      "step": 1351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879726773516675,
+      "loss": 1.1249,
+      "step": 1352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018796374838162417,
+      "loss": 1.0565,
+      "step": 1353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018795481941158088,
+      "loss": 1.0767,
+      "step": 1354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018794589044153758,
+      "loss": 1.1176,
+      "step": 1355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018793696147149426,
+      "loss": 0.9982,
+      "step": 1356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018792803250145096,
+      "loss": 1.0553,
+      "step": 1357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018791910353140766,
+      "loss": 1.1006,
+      "step": 1358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018791017456136434,
+      "loss": 1.074,
+      "step": 1359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018790124559132105,
+      "loss": 1.0416,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018789231662127775,
+      "loss": 1.2348,
+      "step": 1361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018788338765123443,
+      "loss": 1.0253,
+      "step": 1362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018787445868119113,
+      "loss": 1.0426,
+      "step": 1363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018786552971114783,
+      "loss": 1.3454,
+      "step": 1364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878566007411045,
+      "loss": 0.9948,
+      "step": 1365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018784767177106122,
+      "loss": 0.9209,
+      "step": 1366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018783874280101792,
+      "loss": 0.9947,
+      "step": 1367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878298138309746,
+      "loss": 1.2549,
+      "step": 1368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878208848609313,
+      "loss": 1.115,
+      "step": 1369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000187811955890888,
+      "loss": 1.0741,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018780302692084468,
+      "loss": 1.0768,
+      "step": 1371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018779409795080138,
+      "loss": 1.0168,
+      "step": 1372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001877851689807581,
+      "loss": 1.0668,
+      "step": 1373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777624001071477,
+      "loss": 1.1436,
+      "step": 1374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018776731104067147,
+      "loss": 1.0361,
+      "step": 1375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018775838207062817,
+      "loss": 1.0631,
+      "step": 1376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018774945310058485,
+      "loss": 1.146,
+      "step": 1377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018774052413054155,
+      "loss": 0.9633,
+      "step": 1378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018773159516049826,
+      "loss": 1.012,
+      "step": 1379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018772266619045494,
+      "loss": 0.9923,
+      "step": 1380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018771373722041164,
+      "loss": 1.229,
+      "step": 1381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018770480825036834,
+      "loss": 1.1624,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018769587928032502,
+      "loss": 1.1552,
+      "step": 1383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018768695031028172,
+      "loss": 1.0273,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018767802134023843,
+      "loss": 1.1734,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876690923701951,
+      "loss": 1.1243,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876601634001518,
+      "loss": 1.0283,
+      "step": 1387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018765123443010851,
+      "loss": 1.1247,
+      "step": 1388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876423054600652,
+      "loss": 0.9521,
+      "step": 1389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876333764900219,
+      "loss": 1.1918,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001876244475199786,
+      "loss": 1.1237,
+      "step": 1391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018761551854993528,
+      "loss": 1.1316,
+      "step": 1392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018760658957989198,
+      "loss": 1.0863,
+      "step": 1393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018759766060984866,
+      "loss": 1.0202,
+      "step": 1394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018758873163980536,
+      "loss": 1.1077,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757980266976204,
+      "loss": 1.1587,
+      "step": 1396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757087369971874,
+      "loss": 1.0281,
+      "step": 1397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018756194472967542,
+      "loss": 0.9668,
+      "step": 1398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018755301575963212,
+      "loss": 1.0418,
+      "step": 1399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018754408678958883,
+      "loss": 1.0407,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875351578195455,
+      "loss": 1.111,
+      "step": 1401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875262288495022,
+      "loss": 1.116,
+      "step": 1402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875172998794589,
+      "loss": 1.1273,
+      "step": 1403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001875083709094156,
+      "loss": 1.1924,
+      "step": 1404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874994419393723,
+      "loss": 1.1342,
+      "step": 1405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000187490512969329,
+      "loss": 1.1576,
+      "step": 1406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018748158399928567,
+      "loss": 1.0089,
+      "step": 1407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018747265502924238,
+      "loss": 1.1078,
+      "step": 1408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018746372605919908,
+      "loss": 1.1347,
+      "step": 1409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018745479708915576,
+      "loss": 1.0499,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018744586811911246,
+      "loss": 1.1731,
+      "step": 1411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018743693914906917,
+      "loss": 1.05,
+      "step": 1412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018742801017902584,
+      "loss": 0.9599,
+      "step": 1413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018741908120898255,
+      "loss": 0.9938,
+      "step": 1414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018741015223893925,
+      "loss": 1.1477,
+      "step": 1415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018740122326889593,
+      "loss": 1.0554,
+      "step": 1416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018739229429885263,
+      "loss": 1.0434,
+      "step": 1417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018738336532880934,
+      "loss": 1.1444,
+      "step": 1418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018737443635876601,
+      "loss": 1.0466,
+      "step": 1419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018736550738872272,
+      "loss": 1.1227,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018735657841867942,
+      "loss": 1.0072,
+      "step": 1421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873476494486361,
+      "loss": 1.0184,
+      "step": 1422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873387204785928,
+      "loss": 1.1697,
+      "step": 1423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873297915085495,
+      "loss": 1.1721,
+      "step": 1424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873208625385062,
+      "loss": 1.0496,
+      "step": 1425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873119335684629,
+      "loss": 1.1025,
+      "step": 1426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873030045984196,
+      "loss": 1.0169,
+      "step": 1427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872940756283763,
+      "loss": 1.0545,
+      "step": 1428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018728514665833297,
+      "loss": 1.0232,
+      "step": 1429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018727621768828968,
+      "loss": 0.986,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018726728871824638,
+      "loss": 1.1768,
+      "step": 1431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018725835974820306,
+      "loss": 1.0408,
+      "step": 1432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018724943077815976,
+      "loss": 1.5743,
+      "step": 1433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018724050180811647,
+      "loss": 1.1819,
+      "step": 1434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018723157283807314,
+      "loss": 0.9846,
+      "step": 1435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018722264386802985,
+      "loss": 1.1459,
+      "step": 1436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018721371489798652,
+      "loss": 1.0061,
+      "step": 1437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018720478592794323,
+      "loss": 0.9878,
+      "step": 1438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871958569578999,
+      "loss": 0.938,
+      "step": 1439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871869279878566,
+      "loss": 1.1541,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018717799901781329,
+      "loss": 1.048,
+      "step": 1441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018716907004777,
+      "loss": 1.0965,
+      "step": 1442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871601410777267,
+      "loss": 1.0381,
+      "step": 1443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018715121210768337,
+      "loss": 1.053,
+      "step": 1444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018714228313764007,
+      "loss": 1.3908,
+      "step": 1445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018713335416759678,
+      "loss": 0.9843,
+      "step": 1446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018712442519755346,
+      "loss": 0.9544,
+      "step": 1447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018711549622751016,
+      "loss": 1.0935,
+      "step": 1448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018710656725746686,
+      "loss": 1.2562,
+      "step": 1449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018709763828742354,
+      "loss": 1.0169,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018708870931738024,
+      "loss": 1.0384,
+      "step": 1451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018707978034733695,
+      "loss": 1.0541,
+      "step": 1452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018707085137729363,
+      "loss": 1.0372,
+      "step": 1453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018706192240725033,
+      "loss": 0.9951,
+      "step": 1454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018705299343720703,
+      "loss": 1.0302,
+      "step": 1455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001870440644671637,
+      "loss": 1.1465,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018703513549712041,
+      "loss": 1.1227,
+      "step": 1457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018702620652707712,
+      "loss": 1.1227,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001870172775570338,
+      "loss": 0.9544,
+      "step": 1459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001870083485869905,
+      "loss": 1.0302,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001869994196169472,
+      "loss": 1.0344,
+      "step": 1461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018699049064690388,
+      "loss": 1.0121,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018698156167686058,
+      "loss": 0.8804,
+      "step": 1463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001869726327068173,
+      "loss": 1.102,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018696370373677397,
+      "loss": 1.0365,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018695477476673067,
+      "loss": 1.1063,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018694584579668737,
+      "loss": 1.058,
+      "step": 1467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018693691682664405,
+      "loss": 1.0158,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018692798785660075,
+      "loss": 1.007,
+      "step": 1469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018691905888655746,
+      "loss": 1.2399,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018691012991651414,
+      "loss": 1.0882,
+      "step": 1471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018690120094647084,
+      "loss": 1.0313,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018689227197642754,
+      "loss": 1.0309,
+      "step": 1473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018688334300638422,
+      "loss": 1.16,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018687441403634092,
+      "loss": 1.3318,
+      "step": 1475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018686548506629763,
+      "loss": 1.0276,
+      "step": 1476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001868565560962543,
+      "loss": 1.045,
+      "step": 1477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000186847627126211,
+      "loss": 1.0718,
+      "step": 1478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018683869815616771,
+      "loss": 1.0758,
+      "step": 1479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001868297691861244,
+      "loss": 1.0191,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001868208402160811,
+      "loss": 0.9818,
+      "step": 1481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001868119112460378,
+      "loss": 1.1661,
+      "step": 1482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018680298227599448,
+      "loss": 1.0492,
+      "step": 1483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018679405330595118,
+      "loss": 1.09,
+      "step": 1484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018678512433590786,
+      "loss": 1.0173,
+      "step": 1485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018677619536586456,
+      "loss": 1.0618,
+      "step": 1486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018676726639582124,
+      "loss": 1.0324,
+      "step": 1487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018675833742577794,
+      "loss": 1.0797,
+      "step": 1488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018674940845573462,
+      "loss": 1.1785,
+      "step": 1489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018674047948569132,
+      "loss": 0.9766,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018673155051564803,
+      "loss": 1.1384,
+      "step": 1491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001867226215456047,
+      "loss": 1.1461,
+      "step": 1492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001867136925755614,
+      "loss": 1.0575,
+      "step": 1493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001867047636055181,
+      "loss": 0.9805,
+      "step": 1494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001866958346354748,
+      "loss": 1.0414,
+      "step": 1495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001866869056654315,
+      "loss": 1.2702,
+      "step": 1496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001866779766953882,
+      "loss": 1.1452,
+      "step": 1497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018666904772534487,
+      "loss": 1.0002,
+      "step": 1498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018666011875530158,
+      "loss": 1.0133,
+      "step": 1499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018665118978525828,
+      "loss": 1.0914,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018664226081521496,
+      "loss": 1.0108,
+      "step": 1501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018663333184517166,
+      "loss": 1.295,
+      "step": 1502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018662440287512837,
+      "loss": 1.0893,
+      "step": 1503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018661547390508504,
+      "loss": 1.0464,
+      "step": 1504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018660654493504175,
+      "loss": 0.9513,
+      "step": 1505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018659761596499845,
+      "loss": 1.0314,
+      "step": 1506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018658868699495513,
+      "loss": 1.0621,
+      "step": 1507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018657975802491183,
+      "loss": 1.015,
+      "step": 1508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018657082905486854,
+      "loss": 0.9858,
+      "step": 1509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001865619000848252,
+      "loss": 1.1512,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018655297111478192,
+      "loss": 1.0729,
+      "step": 1511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018654404214473862,
+      "loss": 1.1066,
+      "step": 1512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001865351131746953,
+      "loss": 1.0362,
+      "step": 1513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000186526184204652,
+      "loss": 1.1059,
+      "step": 1514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001865172552346087,
+      "loss": 1.1356,
+      "step": 1515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018650832626456538,
+      "loss": 1.2677,
+      "step": 1516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001864993972945221,
+      "loss": 1.0071,
+      "step": 1517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001864904683244788,
+      "loss": 1.0412,
+      "step": 1518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018648153935443547,
+      "loss": 1.1936,
+      "step": 1519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018647261038439217,
+      "loss": 1.0556,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018646368141434888,
+      "loss": 1.1067,
+      "step": 1521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018645475244430558,
+      "loss": 0.9728,
+      "step": 1522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018644582347426226,
+      "loss": 1.1289,
+      "step": 1523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018643689450421896,
+      "loss": 1.0061,
+      "step": 1524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018642796553417567,
+      "loss": 1.0492,
+      "step": 1525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018641903656413234,
+      "loss": 1.1113,
+      "step": 1526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018641010759408905,
+      "loss": 1.0278,
+      "step": 1527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018640117862404572,
+      "loss": 0.9818,
+      "step": 1528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018639224965400243,
+      "loss": 1.1129,
+      "step": 1529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001863833206839591,
+      "loss": 1.1041,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001863743917139158,
+      "loss": 1.1631,
+      "step": 1531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018636546274387249,
+      "loss": 1.1073,
+      "step": 1532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001863565337738292,
+      "loss": 1.0649,
+      "step": 1533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001863476048037859,
+      "loss": 1.0984,
+      "step": 1534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018633867583374257,
+      "loss": 1.5053,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018632974686369927,
+      "loss": 0.9375,
+      "step": 1536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018632081789365598,
+      "loss": 1.0316,
+      "step": 1537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018631188892361266,
+      "loss": 1.1064,
+      "step": 1538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018630295995356936,
+      "loss": 1.0341,
+      "step": 1539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018629403098352606,
+      "loss": 1.0639,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018628510201348274,
+      "loss": 1.2244,
+      "step": 1541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018627617304343944,
+      "loss": 1.05,
+      "step": 1542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018626724407339615,
+      "loss": 0.9464,
+      "step": 1543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018625831510335283,
+      "loss": 1.1463,
+      "step": 1544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018624938613330953,
+      "loss": 1.0709,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018624045716326623,
+      "loss": 1.1373,
+      "step": 1546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001862315281932229,
+      "loss": 1.1586,
+      "step": 1547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018622259922317961,
+      "loss": 1.0676,
+      "step": 1548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018621367025313632,
+      "loss": 1.0451,
+      "step": 1549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000186204741283093,
+      "loss": 1.0301,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001861958123130497,
+      "loss": 1.0536,
+      "step": 1551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001861868833430064,
+      "loss": 1.0525,
+      "step": 1552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018617795437296308,
+      "loss": 1.0663,
+      "step": 1553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018616902540291978,
+      "loss": 1.2023,
+      "step": 1554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001861600964328765,
+      "loss": 1.1625,
+      "step": 1555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018615116746283317,
+      "loss": 1.0068,
+      "step": 1556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018614223849278987,
+      "loss": 1.1424,
+      "step": 1557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018613330952274657,
+      "loss": 1.0951,
+      "step": 1558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018612438055270325,
+      "loss": 1.0148,
+      "step": 1559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018611545158265995,
+      "loss": 1.0272,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018610652261261666,
+      "loss": 1.0627,
+      "step": 1561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018609759364257334,
+      "loss": 1.0298,
+      "step": 1562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018608866467253004,
+      "loss": 0.9675,
+      "step": 1563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018607973570248674,
+      "loss": 0.9991,
+      "step": 1564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018607080673244342,
+      "loss": 1.0618,
+      "step": 1565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018606187776240012,
+      "loss": 1.0234,
+      "step": 1566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018605294879235683,
+      "loss": 0.9942,
+      "step": 1567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001860440198223135,
+      "loss": 1.0781,
+      "step": 1568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001860350908522702,
+      "loss": 1.1406,
+      "step": 1569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001860261618822269,
+      "loss": 1.0565,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001860172329121836,
+      "loss": 1.0861,
+      "step": 1571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001860083039421403,
+      "loss": 1.1018,
+      "step": 1572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000185999374972097,
+      "loss": 1.0133,
+      "step": 1573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018599044600205368,
+      "loss": 0.9912,
+      "step": 1574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018598151703201038,
+      "loss": 1.1694,
+      "step": 1575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018597258806196706,
+      "loss": 1.0664,
+      "step": 1576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018596365909192376,
+      "loss": 1.0808,
+      "step": 1577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018595473012188044,
+      "loss": 1.0,
+      "step": 1578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018594580115183714,
+      "loss": 1.0456,
+      "step": 1579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018593687218179382,
+      "loss": 1.1397,
+      "step": 1580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018592794321175052,
+      "loss": 1.0211,
+      "step": 1581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018591901424170723,
+      "loss": 1.1422,
+      "step": 1582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001859100852716639,
+      "loss": 1.0713,
+      "step": 1583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001859011563016206,
+      "loss": 1.0797,
+      "step": 1584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001858922273315773,
+      "loss": 1.0805,
+      "step": 1585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000185883298361534,
+      "loss": 1.0744,
+      "step": 1586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001858743693914907,
+      "loss": 1.0017,
+      "step": 1587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001858654404214474,
+      "loss": 1.136,
+      "step": 1588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018585651145140407,
+      "loss": 0.9684,
+      "step": 1589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018584758248136078,
+      "loss": 1.0508,
+      "step": 1590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018583865351131748,
+      "loss": 1.0087,
+      "step": 1591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018582972454127416,
+      "loss": 1.0959,
+      "step": 1592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018582079557123086,
+      "loss": 1.1806,
+      "step": 1593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018581186660118757,
+      "loss": 1.0351,
+      "step": 1594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018580293763114424,
+      "loss": 0.9876,
+      "step": 1595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018579400866110095,
+      "loss": 1.1329,
+      "step": 1596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018578507969105765,
+      "loss": 1.0748,
+      "step": 1597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018577615072101433,
+      "loss": 1.0375,
+      "step": 1598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018576722175097103,
+      "loss": 1.0515,
+      "step": 1599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018575829278092774,
+      "loss": 0.9829,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001857493638108844,
+      "loss": 1.1851,
+      "step": 1601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018574043484084112,
+      "loss": 1.0008,
+      "step": 1602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018573150587079782,
+      "loss": 1.1271,
+      "step": 1603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001857225769007545,
+      "loss": 0.9361,
+      "step": 1604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001857136479307112,
+      "loss": 0.9979,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001857047189606679,
+      "loss": 0.9797,
+      "step": 1606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018569578999062458,
+      "loss": 1.1336,
+      "step": 1607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001856868610205813,
+      "loss": 1.0965,
+      "step": 1608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000185677932050538,
+      "loss": 1.0713,
+      "step": 1609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018566900308049467,
+      "loss": 1.1097,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018566007411045137,
+      "loss": 1.044,
+      "step": 1611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018565114514040808,
+      "loss": 1.0906,
+      "step": 1612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018564221617036475,
+      "loss": 1.0575,
+      "step": 1613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018563328720032146,
+      "loss": 1.1872,
+      "step": 1614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018562435823027816,
+      "loss": 1.1261,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018561542926023484,
+      "loss": 0.9946,
+      "step": 1616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018560650029019154,
+      "loss": 1.0852,
+      "step": 1617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018559757132014825,
+      "loss": 1.1481,
+      "step": 1618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018558864235010492,
+      "loss": 1.0011,
+      "step": 1619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018557971338006163,
+      "loss": 1.2889,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001855707844100183,
+      "loss": 1.1271,
+      "step": 1621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000185561855439975,
+      "loss": 1.0213,
+      "step": 1622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018555292646993168,
+      "loss": 1.0781,
+      "step": 1623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001855439974998884,
+      "loss": 0.9934,
+      "step": 1624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001855350685298451,
+      "loss": 0.9832,
+      "step": 1625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018552613955980177,
+      "loss": 1.0256,
+      "step": 1626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018551721058975847,
+      "loss": 0.9833,
+      "step": 1627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018550828161971518,
+      "loss": 0.9637,
+      "step": 1628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018549935264967185,
+      "loss": 1.0241,
+      "step": 1629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018549042367962856,
+      "loss": 0.972,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018548149470958526,
+      "loss": 1.0453,
+      "step": 1631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018547256573954194,
+      "loss": 0.928,
+      "step": 1632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018546363676949864,
+      "loss": 1.1013,
+      "step": 1633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018545470779945535,
+      "loss": 1.1393,
+      "step": 1634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018544577882941202,
+      "loss": 1.0661,
+      "step": 1635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018543684985936873,
+      "loss": 1.0373,
+      "step": 1636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018542792088932543,
+      "loss": 0.9664,
+      "step": 1637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001854189919192821,
+      "loss": 1.096,
+      "step": 1638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018541006294923881,
+      "loss": 1.0014,
+      "step": 1639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018540113397919552,
+      "loss": 1.2012,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001853922050091522,
+      "loss": 1.0172,
+      "step": 1641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001853832760391089,
+      "loss": 1.0779,
+      "step": 1642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001853743470690656,
+      "loss": 1.1127,
+      "step": 1643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018536541809902228,
+      "loss": 1.0821,
+      "step": 1644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018535648912897898,
+      "loss": 1.0838,
+      "step": 1645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001853475601589357,
+      "loss": 1.1379,
+      "step": 1646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018533863118889236,
+      "loss": 1.0111,
+      "step": 1647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018532970221884907,
+      "loss": 1.1389,
+      "step": 1648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018532077324880577,
+      "loss": 1.0581,
+      "step": 1649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018531184427876245,
+      "loss": 1.0235,
+      "step": 1650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018530291530871915,
+      "loss": 0.9824,
+      "step": 1651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018529398633867586,
+      "loss": 1.0212,
+      "step": 1652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018528505736863253,
+      "loss": 1.0749,
+      "step": 1653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018527612839858924,
+      "loss": 1.095,
+      "step": 1654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018526719942854594,
+      "loss": 1.0998,
+      "step": 1655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018525827045850262,
+      "loss": 0.9436,
+      "step": 1656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018524934148845932,
+      "loss": 1.113,
+      "step": 1657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018524041251841603,
+      "loss": 0.9912,
+      "step": 1658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001852314835483727,
+      "loss": 1.1418,
+      "step": 1659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001852225545783294,
+      "loss": 0.8849,
+      "step": 1660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001852136256082861,
+      "loss": 1.203,
+      "step": 1661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001852046966382428,
+      "loss": 1.037,
+      "step": 1662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001851957676681995,
+      "loss": 0.9476,
+      "step": 1663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018518683869815617,
+      "loss": 1.1035,
+      "step": 1664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018517790972811287,
+      "loss": 1.1199,
+      "step": 1665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018516898075806958,
+      "loss": 1.0626,
+      "step": 1666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018516005178802626,
+      "loss": 1.0551,
+      "step": 1667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018515112281798296,
+      "loss": 0.9796,
+      "step": 1668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018514219384793964,
+      "loss": 1.0083,
+      "step": 1669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018513326487789634,
+      "loss": 1.0179,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018512433590785302,
+      "loss": 1.1261,
+      "step": 1671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018511540693780972,
+      "loss": 1.1913,
+      "step": 1672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018510647796776643,
+      "loss": 1.105,
+      "step": 1673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850975489977231,
+      "loss": 1.1396,
+      "step": 1674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850886200276798,
+      "loss": 1.4619,
+      "step": 1675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850796910576365,
+      "loss": 1.2086,
+      "step": 1676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850707620875932,
+      "loss": 1.2421,
+      "step": 1677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850618331175499,
+      "loss": 1.0328,
+      "step": 1678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001850529041475066,
+      "loss": 0.9737,
+      "step": 1679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018504397517746327,
+      "loss": 0.9744,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018503504620741998,
+      "loss": 1.0207,
+      "step": 1681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018502611723737668,
+      "loss": 1.0264,
+      "step": 1682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018501718826733336,
+      "loss": 1.2143,
+      "step": 1683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018500825929729006,
+      "loss": 0.9747,
+      "step": 1684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018499933032724677,
+      "loss": 1.0667,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018499040135720344,
+      "loss": 1.0692,
+      "step": 1686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018498147238716015,
+      "loss": 1.1936,
+      "step": 1687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018497254341711685,
+      "loss": 1.1037,
+      "step": 1688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018496361444707353,
+      "loss": 1.0077,
+      "step": 1689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018495468547703023,
+      "loss": 1.0689,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018494575650698694,
+      "loss": 0.9529,
+      "step": 1691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001849368275369436,
+      "loss": 1.2459,
+      "step": 1692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018492789856690032,
+      "loss": 1.1353,
+      "step": 1693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018491896959685702,
+      "loss": 1.7448,
+      "step": 1694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001849100406268137,
+      "loss": 1.1087,
+      "step": 1695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001849011116567704,
+      "loss": 1.0373,
+      "step": 1696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001848921826867271,
+      "loss": 1.0381,
+      "step": 1697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018488325371668378,
+      "loss": 1.2556,
+      "step": 1698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001848743247466405,
+      "loss": 1.0103,
+      "step": 1699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001848653957765972,
+      "loss": 1.2515,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018485646680655387,
+      "loss": 1.1582,
+      "step": 1701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018484753783651057,
+      "loss": 1.1455,
+      "step": 1702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018483860886646728,
+      "loss": 1.0411,
+      "step": 1703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018482967989642395,
+      "loss": 1.1482,
+      "step": 1704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018482075092638066,
+      "loss": 1.0846,
+      "step": 1705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018481182195633736,
+      "loss": 1.1384,
+      "step": 1706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018480289298629404,
+      "loss": 0.9887,
+      "step": 1707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018479396401625074,
+      "loss": 0.9721,
+      "step": 1708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018478503504620745,
+      "loss": 1.0799,
+      "step": 1709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018477610607616412,
+      "loss": 1.1912,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018476717710612083,
+      "loss": 1.1064,
+      "step": 1711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001847582481360775,
+      "loss": 1.0567,
+      "step": 1712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001847493191660342,
+      "loss": 1.0358,
+      "step": 1713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018474039019599088,
+      "loss": 0.9926,
+      "step": 1714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001847314612259476,
+      "loss": 1.1429,
+      "step": 1715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001847225322559043,
+      "loss": 0.9884,
+      "step": 1716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018471360328586097,
+      "loss": 1.1251,
+      "step": 1717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018470467431581767,
+      "loss": 1.2298,
+      "step": 1718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018469574534577438,
+      "loss": 0.9441,
+      "step": 1719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018468681637573105,
+      "loss": 1.0601,
+      "step": 1720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018467788740568776,
+      "loss": 1.2639,
+      "step": 1721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018466895843564446,
+      "loss": 1.027,
+      "step": 1722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018466002946560114,
+      "loss": 1.1004,
+      "step": 1723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018465110049555784,
+      "loss": 1.5856,
+      "step": 1724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018464217152551455,
+      "loss": 1.0444,
+      "step": 1725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018463324255547122,
+      "loss": 1.1469,
+      "step": 1726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018462431358542793,
+      "loss": 0.982,
+      "step": 1727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 1.0165,
+      "step": 1728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001846064556453413,
+      "loss": 1.0213,
+      "step": 1729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000184597526675298,
+      "loss": 1.0604,
+      "step": 1730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018458859770525472,
+      "loss": 1.0314,
+      "step": 1731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001845796687352114,
+      "loss": 1.0512,
+      "step": 1732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001845707397651681,
+      "loss": 0.9812,
+      "step": 1733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001845618107951248,
+      "loss": 1.0388,
+      "step": 1734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018455288182508148,
+      "loss": 0.988,
+      "step": 1735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018454395285503818,
+      "loss": 0.9841,
+      "step": 1736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001845350238849949,
+      "loss": 1.0934,
+      "step": 1737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018452609491495156,
+      "loss": 1.0994,
+      "step": 1738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018451716594490827,
+      "loss": 1.0094,
+      "step": 1739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018450823697486497,
+      "loss": 1.0783,
+      "step": 1740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018449930800482165,
+      "loss": 1.1115,
+      "step": 1741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018449037903477835,
+      "loss": 1.0723,
+      "step": 1742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018448145006473506,
+      "loss": 1.0366,
+      "step": 1743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018447252109469173,
+      "loss": 1.0791,
+      "step": 1744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018446359212464844,
+      "loss": 1.0585,
+      "step": 1745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018445466315460514,
+      "loss": 1.0455,
+      "step": 1746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018444573418456182,
+      "loss": 1.1649,
+      "step": 1747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018443680521451852,
+      "loss": 1.0762,
+      "step": 1748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018442787624447523,
+      "loss": 0.994,
+      "step": 1749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001844189472744319,
+      "loss": 1.0356,
+      "step": 1750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001844100183043886,
+      "loss": 1.1489,
+      "step": 1751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001844010893343453,
+      "loss": 1.2123,
+      "step": 1752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000184392160364302,
+      "loss": 1.1136,
+      "step": 1753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001843832313942587,
+      "loss": 1.0716,
+      "step": 1754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018437430242421537,
+      "loss": 1.0628,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018436537345417207,
+      "loss": 1.0801,
+      "step": 1756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018435644448412878,
+      "loss": 1.1043,
+      "step": 1757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018434751551408546,
+      "loss": 1.0244,
+      "step": 1758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018433858654404216,
+      "loss": 0.9965,
+      "step": 1759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018432965757399884,
+      "loss": 1.0575,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018432072860395554,
+      "loss": 0.9764,
+      "step": 1761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018431179963391222,
+      "loss": 1.0036,
+      "step": 1762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018430287066386892,
+      "loss": 1.0221,
+      "step": 1763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018429394169382563,
+      "loss": 1.0532,
+      "step": 1764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001842850127237823,
+      "loss": 1.0868,
+      "step": 1765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000184276083753739,
+      "loss": 1.1934,
+      "step": 1766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001842671547836957,
+      "loss": 1.0618,
+      "step": 1767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001842582258136524,
+      "loss": 1.195,
+      "step": 1768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001842492968436091,
+      "loss": 1.0492,
+      "step": 1769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001842403678735658,
+      "loss": 0.943,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018423143890352247,
+      "loss": 1.1933,
+      "step": 1771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018422250993347918,
+      "loss": 1.1218,
+      "step": 1772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018421358096343588,
+      "loss": 1.0716,
+      "step": 1773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018420465199339256,
+      "loss": 1.0716,
+      "step": 1774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018419572302334926,
+      "loss": 1.2525,
+      "step": 1775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018418679405330597,
+      "loss": 1.0613,
+      "step": 1776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018417786508326264,
+      "loss": 1.3372,
+      "step": 1777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018416893611321935,
+      "loss": 1.1575,
+      "step": 1778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018416000714317605,
+      "loss": 1.0814,
+      "step": 1779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018415107817313273,
+      "loss": 1.0617,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018414214920308943,
+      "loss": 1.1694,
+      "step": 1781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018413322023304614,
+      "loss": 0.9575,
+      "step": 1782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001841242912630028,
+      "loss": 1.0314,
+      "step": 1783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018411536229295952,
+      "loss": 1.0617,
+      "step": 1784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018410643332291622,
+      "loss": 0.9485,
+      "step": 1785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001840975043528729,
+      "loss": 1.0543,
+      "step": 1786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001840885753828296,
+      "loss": 1.1696,
+      "step": 1787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001840796464127863,
+      "loss": 1.1356,
+      "step": 1788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018407071744274298,
+      "loss": 1.1986,
+      "step": 1789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018406178847269969,
+      "loss": 1.0712,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001840528595026564,
+      "loss": 1.0935,
+      "step": 1791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018404393053261307,
+      "loss": 1.1531,
+      "step": 1792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018403500156256977,
+      "loss": 0.9706,
+      "step": 1793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018402607259252648,
+      "loss": 1.1714,
+      "step": 1794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018401714362248315,
+      "loss": 1.0947,
+      "step": 1795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018400821465243986,
+      "loss": 1.1008,
+      "step": 1796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018399928568239656,
+      "loss": 1.1955,
+      "step": 1797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018399035671235324,
+      "loss": 1.1133,
+      "step": 1798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018398142774230994,
+      "loss": 0.9657,
+      "step": 1799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018397249877226665,
+      "loss": 1.092,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018396356980222332,
+      "loss": 1.2063,
+      "step": 1801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018395464083218003,
+      "loss": 0.9813,
+      "step": 1802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001839457118621367,
+      "loss": 0.9617,
+      "step": 1803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001839367828920934,
+      "loss": 1.0096,
+      "step": 1804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018392785392205008,
+      "loss": 0.9983,
+      "step": 1805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001839189249520068,
+      "loss": 1.0955,
+      "step": 1806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001839099959819635,
+      "loss": 0.9657,
+      "step": 1807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018390106701192017,
+      "loss": 1.1688,
+      "step": 1808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018389213804187687,
+      "loss": 1.0209,
+      "step": 1809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018388320907183355,
+      "loss": 1.0536,
+      "step": 1810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018387428010179025,
+      "loss": 1.0384,
+      "step": 1811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018386535113174696,
+      "loss": 1.0647,
+      "step": 1812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018385642216170364,
+      "loss": 1.1364,
+      "step": 1813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018384749319166034,
+      "loss": 1.2248,
+      "step": 1814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018383856422161704,
+      "loss": 1.1453,
+      "step": 1815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018382963525157375,
+      "loss": 1.1405,
+      "step": 1816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018382070628153042,
+      "loss": 1.1088,
+      "step": 1817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018381177731148713,
+      "loss": 1.1738,
+      "step": 1818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018380284834144383,
+      "loss": 1.1548,
+      "step": 1819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001837939193714005,
+      "loss": 1.0404,
+      "step": 1820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001837849904013572,
+      "loss": 1.0587,
+      "step": 1821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018377606143131392,
+      "loss": 1.0459,
+      "step": 1822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001837671324612706,
+      "loss": 1.0517,
+      "step": 1823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001837582034912273,
+      "loss": 1.0345,
+      "step": 1824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000183749274521184,
+      "loss": 1.0249,
+      "step": 1825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018374034555114068,
+      "loss": 1.0216,
+      "step": 1826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018373141658109738,
+      "loss": 1.0184,
+      "step": 1827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001837224876110541,
+      "loss": 1.2546,
+      "step": 1828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018371355864101076,
+      "loss": 1.0633,
+      "step": 1829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018370462967096747,
+      "loss": 1.095,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018369570070092417,
+      "loss": 1.0629,
+      "step": 1831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018368677173088085,
+      "loss": 1.0698,
+      "step": 1832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018367784276083755,
+      "loss": 1.0977,
+      "step": 1833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018366891379079426,
+      "loss": 1.5774,
+      "step": 1834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018365998482075093,
+      "loss": 1.1412,
+      "step": 1835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018365105585070764,
+      "loss": 1.0875,
+      "step": 1836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018364212688066434,
+      "loss": 1.0995,
+      "step": 1837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018363319791062102,
+      "loss": 1.1377,
+      "step": 1838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018362426894057772,
+      "loss": 1.0293,
+      "step": 1839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018361533997053443,
+      "loss": 1.0225,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001836064110004911,
+      "loss": 1.0317,
+      "step": 1841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001835974820304478,
+      "loss": 1.0578,
+      "step": 1842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001835885530604045,
+      "loss": 1.0272,
+      "step": 1843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001835796240903612,
+      "loss": 1.0555,
+      "step": 1844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001835706951203179,
+      "loss": 1.0364,
+      "step": 1845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018356176615027457,
+      "loss": 1.179,
+      "step": 1846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018355283718023127,
+      "loss": 1.2465,
+      "step": 1847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018354390821018798,
+      "loss": 1.0301,
+      "step": 1848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018353497924014466,
+      "loss": 1.1525,
+      "step": 1849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018352605027010136,
+      "loss": 1.0871,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018351712130005804,
+      "loss": 1.0726,
+      "step": 1851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018350819233001474,
+      "loss": 1.1438,
+      "step": 1852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018349926335997142,
+      "loss": 1.0922,
+      "step": 1853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018349033438992812,
+      "loss": 1.12,
+      "step": 1854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018348140541988483,
+      "loss": 1.1298,
+      "step": 1855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001834724764498415,
+      "loss": 1.046,
+      "step": 1856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001834635474797982,
+      "loss": 1.1275,
+      "step": 1857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001834546185097549,
+      "loss": 1.229,
+      "step": 1858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001834456895397116,
+      "loss": 1.0913,
+      "step": 1859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001834367605696683,
+      "loss": 1.0127,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000183427831599625,
+      "loss": 1.0852,
+      "step": 1861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018341890262958167,
+      "loss": 1.1443,
+      "step": 1862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018340997365953838,
+      "loss": 1.0531,
+      "step": 1863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018340104468949508,
+      "loss": 1.075,
+      "step": 1864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018339211571945176,
+      "loss": 1.0243,
+      "step": 1865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018338318674940846,
+      "loss": 1.2768,
+      "step": 1866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018337425777936516,
+      "loss": 0.8675,
+      "step": 1867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018336532880932184,
+      "loss": 1.0369,
+      "step": 1868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018335639983927855,
+      "loss": 1.0459,
+      "step": 1869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018334747086923525,
+      "loss": 1.2085,
+      "step": 1870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018333854189919193,
+      "loss": 1.1838,
+      "step": 1871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018332961292914863,
+      "loss": 1.0454,
+      "step": 1872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018332068395910533,
+      "loss": 0.9615,
+      "step": 1873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000183311754989062,
+      "loss": 1.1627,
+      "step": 1874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018330282601901872,
+      "loss": 1.0129,
+      "step": 1875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018329389704897542,
+      "loss": 1.0788,
+      "step": 1876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001832849680789321,
+      "loss": 1.2665,
+      "step": 1877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001832760391088888,
+      "loss": 1.0483,
+      "step": 1878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001832671101388455,
+      "loss": 0.9462,
+      "step": 1879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018325818116880218,
+      "loss": 1.0679,
+      "step": 1880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018324925219875889,
+      "loss": 1.0541,
+      "step": 1881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001832403232287156,
+      "loss": 1.0616,
+      "step": 1882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018323139425867227,
+      "loss": 0.9863,
+      "step": 1883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018322246528862897,
+      "loss": 1.1059,
+      "step": 1884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018321353631858567,
+      "loss": 1.0579,
+      "step": 1885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018320460734854235,
+      "loss": 1.0714,
+      "step": 1886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018319567837849906,
+      "loss": 0.9985,
+      "step": 1887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018318674940845576,
+      "loss": 1.0524,
+      "step": 1888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018317782043841244,
+      "loss": 1.1261,
+      "step": 1889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018316889146836914,
+      "loss": 1.1378,
+      "step": 1890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018315996249832584,
+      "loss": 1.1694,
+      "step": 1891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018315103352828252,
+      "loss": 1.0169,
+      "step": 1892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018314210455823923,
+      "loss": 1.2681,
+      "step": 1893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001831331755881959,
+      "loss": 1.0667,
+      "step": 1894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001831242466181526,
+      "loss": 1.1488,
+      "step": 1895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018311531764810928,
+      "loss": 1.0299,
+      "step": 1896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000183106388678066,
+      "loss": 1.0502,
+      "step": 1897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001830974597080227,
+      "loss": 1.2168,
+      "step": 1898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018308853073797937,
+      "loss": 1.1217,
+      "step": 1899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018307960176793607,
+      "loss": 1.3061,
+      "step": 1900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018307067279789275,
+      "loss": 1.1802,
+      "step": 1901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018306174382784945,
+      "loss": 0.9916,
+      "step": 1902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018305281485780616,
+      "loss": 1.1532,
+      "step": 1903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018304388588776283,
+      "loss": 1.1172,
+      "step": 1904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018303495691771954,
+      "loss": 1.062,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018302602794767624,
+      "loss": 1.0682,
+      "step": 1906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018301709897763292,
+      "loss": 1.058,
+      "step": 1907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018300817000758962,
+      "loss": 1.1285,
+      "step": 1908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018299924103754633,
+      "loss": 1.0894,
+      "step": 1909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000182990312067503,
+      "loss": 1.1141,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001829813830974597,
+      "loss": 0.9708,
+      "step": 1911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001829724541274164,
+      "loss": 1.139,
+      "step": 1912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018296352515737312,
+      "loss": 1.0685,
+      "step": 1913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001829545961873298,
+      "loss": 1.0243,
+      "step": 1914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001829456672172865,
+      "loss": 1.0667,
+      "step": 1915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001829367382472432,
+      "loss": 1.0895,
+      "step": 1916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018292780927719988,
+      "loss": 1.1327,
+      "step": 1917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018291888030715658,
+      "loss": 1.091,
+      "step": 1918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001829099513371133,
+      "loss": 0.9679,
+      "step": 1919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018290102236706996,
+      "loss": 1.0657,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018289209339702667,
+      "loss": 1.0222,
+      "step": 1921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018288316442698337,
+      "loss": 0.9945,
+      "step": 1922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018287423545694005,
+      "loss": 1.0854,
+      "step": 1923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018286530648689675,
+      "loss": 1.2367,
+      "step": 1924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018285637751685346,
+      "loss": 0.9329,
+      "step": 1925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018284744854681013,
+      "loss": 1.1042,
+      "step": 1926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018283851957676684,
+      "loss": 1.0933,
+      "step": 1927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018282959060672354,
+      "loss": 1.0188,
+      "step": 1928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018282066163668022,
+      "loss": 1.0762,
+      "step": 1929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018281173266663692,
+      "loss": 0.9756,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018280280369659363,
+      "loss": 0.9965,
+      "step": 1931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001827938747265503,
+      "loss": 1.0291,
+      "step": 1932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000182784945756507,
+      "loss": 1.1409,
+      "step": 1933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001827760167864637,
+      "loss": 1.1238,
+      "step": 1934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001827670878164204,
+      "loss": 1.132,
+      "step": 1935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001827581588463771,
+      "loss": 1.1575,
+      "step": 1936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018274922987633377,
+      "loss": 1.1194,
+      "step": 1937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018274030090629047,
+      "loss": 1.1639,
+      "step": 1938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018273137193624718,
+      "loss": 1.1928,
+      "step": 1939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018272244296620385,
+      "loss": 1.2246,
+      "step": 1940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018271351399616056,
+      "loss": 0.9974,
+      "step": 1941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018270458502611724,
+      "loss": 1.1452,
+      "step": 1942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018269565605607394,
+      "loss": 0.9942,
+      "step": 1943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018268672708603062,
+      "loss": 1.1957,
+      "step": 1944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018267779811598732,
+      "loss": 1.1581,
+      "step": 1945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018266886914594402,
+      "loss": 1.0314,
+      "step": 1946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826599401759007,
+      "loss": 1.0292,
+      "step": 1947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826510112058574,
+      "loss": 1.0342,
+      "step": 1948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826420822358141,
+      "loss": 1.0617,
+      "step": 1949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826331532657708,
+      "loss": 1.0489,
+      "step": 1950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826242242957275,
+      "loss": 1.1801,
+      "step": 1951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001826152953256842,
+      "loss": 1.1269,
+      "step": 1952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018260636635564087,
+      "loss": 1.0677,
+      "step": 1953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018259743738559758,
+      "loss": 1.051,
+      "step": 1954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018258850841555428,
+      "loss": 1.1583,
+      "step": 1955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018257957944551096,
+      "loss": 1.089,
+      "step": 1956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018257065047546766,
+      "loss": 1.0066,
+      "step": 1957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018256172150542436,
+      "loss": 1.0393,
+      "step": 1958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018255279253538104,
+      "loss": 0.9768,
+      "step": 1959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018254386356533775,
+      "loss": 1.1638,
+      "step": 1960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018253493459529445,
+      "loss": 1.0835,
+      "step": 1961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018252600562525113,
+      "loss": 1.0035,
+      "step": 1962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018251707665520783,
+      "loss": 1.1693,
+      "step": 1963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018250814768516453,
+      "loss": 1.036,
+      "step": 1964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001824992187151212,
+      "loss": 1.0331,
+      "step": 1965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018249028974507792,
+      "loss": 1.0171,
+      "step": 1966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018248136077503462,
+      "loss": 1.0106,
+      "step": 1967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001824724318049913,
+      "loss": 1.0225,
+      "step": 1968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000182463502834948,
+      "loss": 1.0632,
+      "step": 1969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001824545738649047,
+      "loss": 1.2258,
+      "step": 1970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018244564489486138,
+      "loss": 1.1847,
+      "step": 1971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018243671592481809,
+      "loss": 1.0827,
+      "step": 1972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001824277869547748,
+      "loss": 1.0046,
+      "step": 1973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018241885798473147,
+      "loss": 1.134,
+      "step": 1974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018240992901468817,
+      "loss": 1.097,
+      "step": 1975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018240100004464487,
+      "loss": 1.003,
+      "step": 1976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018239207107460155,
+      "loss": 1.0945,
+      "step": 1977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018238314210455826,
+      "loss": 1.0766,
+      "step": 1978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018237421313451496,
+      "loss": 1.0345,
+      "step": 1979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018236528416447164,
+      "loss": 0.9725,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018235635519442834,
+      "loss": 1.0923,
+      "step": 1981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018234742622438504,
+      "loss": 1.0705,
+      "step": 1982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018233849725434172,
+      "loss": 1.0071,
+      "step": 1983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018232956828429843,
+      "loss": 1.0313,
+      "step": 1984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001823206393142551,
+      "loss": 1.1902,
+      "step": 1985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001823117103442118,
+      "loss": 1.0023,
+      "step": 1986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018230278137416848,
+      "loss": 1.0335,
+      "step": 1987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001822938524041252,
+      "loss": 1.0348,
+      "step": 1988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018228492343408186,
+      "loss": 0.9389,
+      "step": 1989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018227599446403857,
+      "loss": 1.1809,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018226706549399527,
+      "loss": 1.0315,
+      "step": 1991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018225813652395195,
+      "loss": 1.0722,
+      "step": 1992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018224920755390865,
+      "loss": 1.0616,
+      "step": 1993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018224027858386536,
+      "loss": 1.0474,
+      "step": 1994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018223134961382203,
+      "loss": 1.109,
+      "step": 1995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018222242064377874,
+      "loss": 1.1311,
+      "step": 1996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018221349167373544,
+      "loss": 1.0582,
+      "step": 1997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018220456270369212,
+      "loss": 1.0948,
+      "step": 1998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018219563373364882,
+      "loss": 1.1062,
+      "step": 1999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018218670476360553,
+      "loss": 1.1327,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.0601091384887695,
+      "eval_runtime": 3050.234,
+      "eval_samples_per_second": 6.165,
+      "eval_steps_per_second": 0.771,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001821777757935622,
+      "loss": 1.144,
+      "step": 2001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001821688468235189,
+      "loss": 1.0321,
+      "step": 2002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001821599178534756,
+      "loss": 1.1966,
+      "step": 2003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001821509888834323,
+      "loss": 1.0295,
+      "step": 2004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000182142059913389,
+      "loss": 1.053,
+      "step": 2005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001821331309433457,
+      "loss": 1.0186,
+      "step": 2006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018212420197330237,
+      "loss": 0.978,
+      "step": 2007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018211527300325908,
+      "loss": 0.944,
+      "step": 2008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018210634403321578,
+      "loss": 1.1236,
+      "step": 2009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018209741506317246,
+      "loss": 0.9878,
+      "step": 2010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018208848609312916,
+      "loss": 1.0638,
+      "step": 2011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018207955712308587,
+      "loss": 1.0124,
+      "step": 2012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018207062815304257,
+      "loss": 0.979,
+      "step": 2013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018206169918299925,
+      "loss": 1.0508,
+      "step": 2014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018205277021295595,
+      "loss": 0.9766,
+      "step": 2015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018204384124291266,
+      "loss": 1.1069,
+      "step": 2016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018203491227286933,
+      "loss": 1.0258,
+      "step": 2017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018202598330282604,
+      "loss": 1.0674,
+      "step": 2018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018201705433278274,
+      "loss": 1.1152,
+      "step": 2019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018200812536273942,
+      "loss": 1.1104,
+      "step": 2020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018199919639269612,
+      "loss": 0.9912,
+      "step": 2021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018199026742265283,
+      "loss": 0.9985,
+      "step": 2022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001819813384526095,
+      "loss": 1.0668,
+      "step": 2023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001819724094825662,
+      "loss": 1.2558,
+      "step": 2024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001819634805125229,
+      "loss": 1.0814,
+      "step": 2025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001819545515424796,
+      "loss": 1.0679,
+      "step": 2026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001819456225724363,
+      "loss": 1.0255,
+      "step": 2027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018193669360239297,
+      "loss": 1.1638,
+      "step": 2028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018192776463234967,
+      "loss": 1.0582,
+      "step": 2029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018191883566230638,
+      "loss": 1.0502,
+      "step": 2030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018190990669226305,
+      "loss": 1.3059,
+      "step": 2031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018190097772221976,
+      "loss": 1.0947,
+      "step": 2032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018189204875217644,
+      "loss": 0.929,
+      "step": 2033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018188311978213314,
+      "loss": 1.0153,
+      "step": 2034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018187419081208982,
+      "loss": 1.0047,
+      "step": 2035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018186526184204652,
+      "loss": 1.1374,
+      "step": 2036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018185633287200322,
+      "loss": 1.1771,
+      "step": 2037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001818474039019599,
+      "loss": 1.1643,
+      "step": 2038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001818384749319166,
+      "loss": 1.0422,
+      "step": 2039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001818295459618733,
+      "loss": 1.0966,
+      "step": 2040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018182061699182999,
+      "loss": 1.0966,
+      "step": 2041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001818116880217867,
+      "loss": 1.1433,
+      "step": 2042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001818027590517434,
+      "loss": 1.0038,
+      "step": 2043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018179383008170007,
+      "loss": 1.059,
+      "step": 2044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018178490111165678,
+      "loss": 1.0756,
+      "step": 2045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018177597214161348,
+      "loss": 0.9704,
+      "step": 2046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018176704317157016,
+      "loss": 1.0134,
+      "step": 2047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018175811420152686,
+      "loss": 1.0631,
+      "step": 2048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018174918523148356,
+      "loss": 1.0509,
+      "step": 2049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018174025626144024,
+      "loss": 0.9465,
+      "step": 2050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018173132729139695,
+      "loss": 1.0484,
+      "step": 2051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018172239832135365,
+      "loss": 1.0212,
+      "step": 2052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018171346935131033,
+      "loss": 1.0089,
+      "step": 2053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018170454038126703,
+      "loss": 1.1989,
+      "step": 2054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018169561141122373,
+      "loss": 1.1784,
+      "step": 2055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001816866824411804,
+      "loss": 1.0015,
+      "step": 2056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018167775347113712,
+      "loss": 0.9982,
+      "step": 2057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018166882450109382,
+      "loss": 1.1742,
+      "step": 2058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001816598955310505,
+      "loss": 1.088,
+      "step": 2059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001816509665610072,
+      "loss": 1.0501,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001816420375909639,
+      "loss": 1.0364,
+      "step": 2061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018163310862092058,
+      "loss": 1.024,
+      "step": 2062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018162417965087729,
+      "loss": 1.0108,
+      "step": 2063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000181615250680834,
+      "loss": 1.0771,
+      "step": 2064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018160632171079067,
+      "loss": 0.9791,
+      "step": 2065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018159739274074737,
+      "loss": 1.0687,
+      "step": 2066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018158846377070407,
+      "loss": 1.1596,
+      "step": 2067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018157953480066075,
+      "loss": 1.1474,
+      "step": 2068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018157060583061746,
+      "loss": 1.2268,
+      "step": 2069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018156167686057416,
+      "loss": 1.1574,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018155274789053084,
+      "loss": 1.0302,
+      "step": 2071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018154381892048754,
+      "loss": 1.0869,
+      "step": 2072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018153488995044424,
+      "loss": 1.1487,
+      "step": 2073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018152596098040092,
+      "loss": 1.0502,
+      "step": 2074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018151703201035763,
+      "loss": 1.0069,
+      "step": 2075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001815081030403143,
+      "loss": 1.0135,
+      "step": 2076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000181499174070271,
+      "loss": 1.0682,
+      "step": 2077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018149024510022768,
+      "loss": 1.0723,
+      "step": 2078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001814813161301844,
+      "loss": 0.9748,
+      "step": 2079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018147238716014106,
+      "loss": 0.9842,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018146345819009777,
+      "loss": 1.0026,
+      "step": 2081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018145452922005447,
+      "loss": 1.168,
+      "step": 2082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018144560025001115,
+      "loss": 1.116,
+      "step": 2083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018143667127996785,
+      "loss": 1.261,
+      "step": 2084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018142774230992456,
+      "loss": 1.2286,
+      "step": 2085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018141881333988123,
+      "loss": 1.1819,
+      "step": 2086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018140988436983794,
+      "loss": 0.931,
+      "step": 2087
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018140095539979464,
+      "loss": 1.0272,
+      "step": 2088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018139202642975132,
+      "loss": 1.1541,
+      "step": 2089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018138309745970802,
+      "loss": 1.0681,
+      "step": 2090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018137416848966473,
+      "loss": 1.0008,
+      "step": 2091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813652395196214,
+      "loss": 0.9926,
+      "step": 2092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813563105495781,
+      "loss": 1.0661,
+      "step": 2093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813473815795348,
+      "loss": 1.0996,
+      "step": 2094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813384526094915,
+      "loss": 0.9192,
+      "step": 2095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813295236394482,
+      "loss": 1.0146,
+      "step": 2096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001813205946694049,
+      "loss": 1.1024,
+      "step": 2097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018131166569936157,
+      "loss": 0.9786,
+      "step": 2098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018130273672931828,
+      "loss": 1.0254,
+      "step": 2099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018129380775927498,
+      "loss": 0.9423,
+      "step": 2100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018128487878923166,
+      "loss": 1.0792,
+      "step": 2101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018127594981918836,
+      "loss": 1.0386,
+      "step": 2102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018126702084914507,
+      "loss": 0.9679,
+      "step": 2103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018125809187910174,
+      "loss": 0.9738,
+      "step": 2104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018124916290905845,
+      "loss": 1.0065,
+      "step": 2105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018124023393901515,
+      "loss": 1.0593,
+      "step": 2106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018123130496897183,
+      "loss": 0.9858,
+      "step": 2107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018122237599892853,
+      "loss": 1.1266,
+      "step": 2108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018121344702888524,
+      "loss": 1.1783,
+      "step": 2109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018120451805884194,
+      "loss": 1.21,
+      "step": 2110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018119558908879862,
+      "loss": 1.0003,
+      "step": 2111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018118666011875532,
+      "loss": 1.1047,
+      "step": 2112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018117773114871203,
+      "loss": 1.0435,
+      "step": 2113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001811688021786687,
+      "loss": 1.0997,
+      "step": 2114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001811598732086254,
+      "loss": 0.9699,
+      "step": 2115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001811509442385821,
+      "loss": 1.0297,
+      "step": 2116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001811420152685388,
+      "loss": 0.9904,
+      "step": 2117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001811330862984955,
+      "loss": 0.9589,
+      "step": 2118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018112415732845217,
+      "loss": 1.0851,
+      "step": 2119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018111522835840887,
+      "loss": 1.0857,
+      "step": 2120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018110629938836555,
+      "loss": 1.1529,
+      "step": 2121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018109737041832225,
+      "loss": 1.1082,
+      "step": 2122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018108844144827896,
+      "loss": 1.0631,
+      "step": 2123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018107951247823563,
+      "loss": 1.0202,
+      "step": 2124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018107058350819234,
+      "loss": 1.0703,
+      "step": 2125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018106165453814902,
+      "loss": 1.1933,
+      "step": 2126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018105272556810572,
+      "loss": 0.9797,
+      "step": 2127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018104379659806242,
+      "loss": 1.0258,
+      "step": 2128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001810348676280191,
+      "loss": 1.1202,
+      "step": 2129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001810259386579758,
+      "loss": 1.0571,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001810170096879325,
+      "loss": 1.1889,
+      "step": 2131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018100808071788919,
+      "loss": 1.0141,
+      "step": 2132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001809991517478459,
+      "loss": 1.0684,
+      "step": 2133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001809902227778026,
+      "loss": 1.0768,
+      "step": 2134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018098129380775927,
+      "loss": 1.1331,
+      "step": 2135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018097236483771597,
+      "loss": 1.1002,
+      "step": 2136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018096343586767268,
+      "loss": 1.3032,
+      "step": 2137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018095450689762936,
+      "loss": 1.0432,
+      "step": 2138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018094557792758606,
+      "loss": 1.0508,
+      "step": 2139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018093664895754276,
+      "loss": 1.0858,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018092771998749944,
+      "loss": 1.0596,
+      "step": 2141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018091879101745614,
+      "loss": 1.2157,
+      "step": 2142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018090986204741285,
+      "loss": 1.0115,
+      "step": 2143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018090093307736953,
+      "loss": 1.0689,
+      "step": 2144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018089200410732623,
+      "loss": 1.0243,
+      "step": 2145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018088307513728293,
+      "loss": 1.0473,
+      "step": 2146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001808741461672396,
+      "loss": 1.1715,
+      "step": 2147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018086521719719631,
+      "loss": 1.089,
+      "step": 2148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018085628822715302,
+      "loss": 0.9685,
+      "step": 2149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001808473592571097,
+      "loss": 0.9365,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001808384302870664,
+      "loss": 0.9921,
+      "step": 2151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001808295013170231,
+      "loss": 1.0733,
+      "step": 2152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018082057234697978,
+      "loss": 1.0601,
+      "step": 2153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018081164337693648,
+      "loss": 1.1628,
+      "step": 2154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001808027144068932,
+      "loss": 0.9967,
+      "step": 2155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018079378543684987,
+      "loss": 0.9916,
+      "step": 2156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018078485646680657,
+      "loss": 1.0196,
+      "step": 2157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018077592749676327,
+      "loss": 1.0441,
+      "step": 2158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018076699852671995,
+      "loss": 0.9,
+      "step": 2159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018075806955667665,
+      "loss": 1.0716,
+      "step": 2160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018074914058663336,
+      "loss": 1.1836,
+      "step": 2161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018074021161659004,
+      "loss": 1.1212,
+      "step": 2162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018073128264654674,
+      "loss": 1.0037,
+      "step": 2163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018072235367650344,
+      "loss": 1.0618,
+      "step": 2164
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018071342470646012,
+      "loss": 1.0797,
+      "step": 2165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018070449573641682,
+      "loss": 1.0652,
+      "step": 2166
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001806955667663735,
+      "loss": 1.1232,
+      "step": 2167
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001806866377963302,
+      "loss": 1.1683,
+      "step": 2168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018067770882628688,
+      "loss": 1.0668,
+      "step": 2169
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001806687798562436,
+      "loss": 1.0658,
+      "step": 2170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018065985088620026,
+      "loss": 1.2018,
+      "step": 2171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018065092191615697,
+      "loss": 0.9799,
+      "step": 2172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018064199294611367,
+      "loss": 1.0597,
+      "step": 2173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018063306397607035,
+      "loss": 0.9952,
+      "step": 2174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018062413500602705,
+      "loss": 1.0059,
+      "step": 2175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018061520603598376,
+      "loss": 1.08,
+      "step": 2176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018060627706594043,
+      "loss": 1.0763,
+      "step": 2177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018059734809589714,
+      "loss": 1.0092,
+      "step": 2178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018058841912585384,
+      "loss": 1.0732,
+      "step": 2179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018057949015581052,
+      "loss": 1.0738,
+      "step": 2180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018057056118576722,
+      "loss": 1.0288,
+      "step": 2181
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018056163221572393,
+      "loss": 1.0362,
+      "step": 2182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001805527032456806,
+      "loss": 1.0797,
+      "step": 2183
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001805437742756373,
+      "loss": 1.0339,
+      "step": 2184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000180534845305594,
+      "loss": 0.8663,
+      "step": 2185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001805259163355507,
+      "loss": 1.1487,
+      "step": 2186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001805169873655074,
+      "loss": 1.1065,
+      "step": 2187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001805080583954641,
+      "loss": 1.1376,
+      "step": 2188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018049912942542077,
+      "loss": 1.2466,
+      "step": 2189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018049020045537748,
+      "loss": 1.1652,
+      "step": 2190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018048127148533418,
+      "loss": 1.0712,
+      "step": 2191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018047234251529086,
+      "loss": 1.0837,
+      "step": 2192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018046341354524756,
+      "loss": 1.0917,
+      "step": 2193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018045448457520427,
+      "loss": 1.2048,
+      "step": 2194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018044555560516094,
+      "loss": 1.0576,
+      "step": 2195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018043662663511765,
+      "loss": 0.9327,
+      "step": 2196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018042769766507435,
+      "loss": 1.0311,
+      "step": 2197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018041876869503103,
+      "loss": 1.181,
+      "step": 2198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018040983972498773,
+      "loss": 1.0182,
+      "step": 2199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018040091075494444,
+      "loss": 1.1716,
+      "step": 2200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018039198178490111,
+      "loss": 1.042,
+      "step": 2201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018038305281485782,
+      "loss": 1.06,
+      "step": 2202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018037412384481452,
+      "loss": 1.0862,
+      "step": 2203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001803651948747712,
+      "loss": 1.1704,
+      "step": 2204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001803562659047279,
+      "loss": 1.0578,
+      "step": 2205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001803473369346846,
+      "loss": 0.9815,
+      "step": 2206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001803384079646413,
+      "loss": 0.9857,
+      "step": 2207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000180329478994598,
+      "loss": 1.0843,
+      "step": 2208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001803205500245547,
+      "loss": 1.0126,
+      "step": 2209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018031162105451137,
+      "loss": 1.0332,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018030269208446807,
+      "loss": 1.2749,
+      "step": 2211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018029376311442475,
+      "loss": 0.973,
+      "step": 2212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018028483414438145,
+      "loss": 1.0201,
+      "step": 2213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018027590517433816,
+      "loss": 0.9091,
+      "step": 2214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018026697620429483,
+      "loss": 1.0899,
+      "step": 2215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018025804723425154,
+      "loss": 1.0827,
+      "step": 2216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018024911826420822,
+      "loss": 1.0112,
+      "step": 2217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018024018929416492,
+      "loss": 1.0511,
+      "step": 2218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018023126032412162,
+      "loss": 1.063,
+      "step": 2219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001802223313540783,
+      "loss": 1.0576,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000180213402384035,
+      "loss": 0.9405,
+      "step": 2221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001802044734139917,
+      "loss": 1.1669,
+      "step": 2222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018019554444394839,
+      "loss": 0.9422,
+      "step": 2223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001801866154739051,
+      "loss": 0.9801,
+      "step": 2224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001801776865038618,
+      "loss": 1.0866,
+      "step": 2225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018016875753381847,
+      "loss": 0.9609,
+      "step": 2226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018015982856377517,
+      "loss": 1.0073,
+      "step": 2227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018015089959373188,
+      "loss": 1.0651,
+      "step": 2228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018014197062368856,
+      "loss": 1.05,
+      "step": 2229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018013304165364526,
+      "loss": 1.0371,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018012411268360196,
+      "loss": 0.9654,
+      "step": 2231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018011518371355864,
+      "loss": 1.045,
+      "step": 2232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018010625474351534,
+      "loss": 1.0197,
+      "step": 2233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018009732577347205,
+      "loss": 0.9551,
+      "step": 2234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018008839680342873,
+      "loss": 1.0584,
+      "step": 2235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018007946783338543,
+      "loss": 0.9244,
+      "step": 2236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018007053886334213,
+      "loss": 0.9666,
+      "step": 2237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001800616098932988,
+      "loss": 1.1056,
+      "step": 2238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018005268092325551,
+      "loss": 1.06,
+      "step": 2239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018004375195321222,
+      "loss": 0.9662,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001800348229831689,
+      "loss": 1.1101,
+      "step": 2241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001800258940131256,
+      "loss": 1.1176,
+      "step": 2242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001800169650430823,
+      "loss": 1.1028,
+      "step": 2243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018000803607303898,
+      "loss": 1.156,
+      "step": 2244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017999910710299568,
+      "loss": 0.9684,
+      "step": 2245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001799901781329524,
+      "loss": 0.9753,
+      "step": 2246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017998124916290907,
+      "loss": 1.014,
+      "step": 2247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017997232019286577,
+      "loss": 1.1256,
+      "step": 2248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017996339122282247,
+      "loss": 1.1301,
+      "step": 2249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017995446225277915,
+      "loss": 1.0625,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017994553328273585,
+      "loss": 1.0627,
+      "step": 2251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017993660431269256,
+      "loss": 1.0631,
+      "step": 2252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017992767534264924,
+      "loss": 1.0473,
+      "step": 2253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017991874637260594,
+      "loss": 1.1637,
+      "step": 2254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017990981740256264,
+      "loss": 1.0139,
+      "step": 2255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017990088843251932,
+      "loss": 0.9659,
+      "step": 2256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017989195946247602,
+      "loss": 1.0213,
+      "step": 2257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001798830304924327,
+      "loss": 1.1162,
+      "step": 2258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001798741015223894,
+      "loss": 1.0927,
+      "step": 2259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017986517255234608,
+      "loss": 1.0821,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017985624358230279,
+      "loss": 0.9288,
+      "step": 2261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017984731461225946,
+      "loss": 1.243,
+      "step": 2262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017983838564221617,
+      "loss": 0.9537,
+      "step": 2263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017982945667217287,
+      "loss": 1.0961,
+      "step": 2264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017982052770212955,
+      "loss": 1.0579,
+      "step": 2265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017981159873208625,
+      "loss": 1.0928,
+      "step": 2266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017980266976204296,
+      "loss": 1.0265,
+      "step": 2267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017979374079199963,
+      "loss": 1.1961,
+      "step": 2268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017978481182195634,
+      "loss": 1.1744,
+      "step": 2269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017977588285191304,
+      "loss": 1.2687,
+      "step": 2270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017976695388186972,
+      "loss": 1.0292,
+      "step": 2271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017975802491182642,
+      "loss": 1.0157,
+      "step": 2272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017974909594178313,
+      "loss": 1.1742,
+      "step": 2273
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001797401669717398,
+      "loss": 1.0506,
+      "step": 2274
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001797312380016965,
+      "loss": 0.9741,
+      "step": 2275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001797223090316532,
+      "loss": 1.0009,
+      "step": 2276
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001797133800616099,
+      "loss": 1.3836,
+      "step": 2277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001797044510915666,
+      "loss": 0.9863,
+      "step": 2278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001796955221215233,
+      "loss": 1.0912,
+      "step": 2279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017968659315147997,
+      "loss": 1.0034,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017967766418143668,
+      "loss": 1.0663,
+      "step": 2281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017966873521139338,
+      "loss": 1.1535,
+      "step": 2282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017965980624135006,
+      "loss": 1.0543,
+      "step": 2283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017965087727130676,
+      "loss": 1.0748,
+      "step": 2284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017964194830126347,
+      "loss": 1.1719,
+      "step": 2285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017963301933122014,
+      "loss": 1.013,
+      "step": 2286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017962409036117685,
+      "loss": 1.0109,
+      "step": 2287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017961516139113355,
+      "loss": 1.0539,
+      "step": 2288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017960623242109023,
+      "loss": 1.0666,
+      "step": 2289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017959730345104693,
+      "loss": 1.0296,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017958837448100364,
+      "loss": 1.0312,
+      "step": 2291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001795794455109603,
+      "loss": 0.9545,
+      "step": 2292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017957051654091702,
+      "loss": 1.0118,
+      "step": 2293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017956158757087372,
+      "loss": 1.0887,
+      "step": 2294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001795526586008304,
+      "loss": 1.1024,
+      "step": 2295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001795437296307871,
+      "loss": 1.0433,
+      "step": 2296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001795348006607438,
+      "loss": 0.9886,
+      "step": 2297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017952587169070048,
+      "loss": 0.9475,
+      "step": 2298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001795169427206572,
+      "loss": 1.0197,
+      "step": 2299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001795080137506139,
+      "loss": 1.0162,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017949908478057057,
+      "loss": 1.0925,
+      "step": 2301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017949015581052727,
+      "loss": 1.0401,
+      "step": 2302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017948122684048395,
+      "loss": 0.9839,
+      "step": 2303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017947229787044065,
+      "loss": 1.0374,
+      "step": 2304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017946336890039736,
+      "loss": 1.045,
+      "step": 2305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017945443993035403,
+      "loss": 1.0518,
+      "step": 2306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017944551096031074,
+      "loss": 1.167,
+      "step": 2307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017943658199026742,
+      "loss": 0.9648,
+      "step": 2308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017942765302022412,
+      "loss": 1.1017,
+      "step": 2309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017941872405018082,
+      "loss": 1.1451,
+      "step": 2310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001794097950801375,
+      "loss": 1.0791,
+      "step": 2311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001794008661100942,
+      "loss": 0.9185,
+      "step": 2312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001793919371400509,
+      "loss": 1.1217,
+      "step": 2313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017938300817000759,
+      "loss": 1.0299,
+      "step": 2314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001793740791999643,
+      "loss": 0.9327,
+      "step": 2315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000179365150229921,
+      "loss": 1.0707,
+      "step": 2316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017935622125987767,
+      "loss": 1.0431,
+      "step": 2317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017934729228983437,
+      "loss": 1.0472,
+      "step": 2318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017933836331979108,
+      "loss": 1.1149,
+      "step": 2319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017932943434974776,
+      "loss": 1.051,
+      "step": 2320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017932050537970446,
+      "loss": 1.1329,
+      "step": 2321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017931157640966116,
+      "loss": 1.0861,
+      "step": 2322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017930264743961784,
+      "loss": 1.0347,
+      "step": 2323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017929371846957454,
+      "loss": 1.0238,
+      "step": 2324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017928478949953125,
+      "loss": 1.0245,
+      "step": 2325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017927586052948793,
+      "loss": 1.0189,
+      "step": 2326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017926693155944463,
+      "loss": 1.1172,
+      "step": 2327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017925800258940133,
+      "loss": 1.094,
+      "step": 2328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000179249073619358,
+      "loss": 1.0578,
+      "step": 2329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017924014464931471,
+      "loss": 1.1893,
+      "step": 2330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017923121567927142,
+      "loss": 1.0808,
+      "step": 2331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001792222867092281,
+      "loss": 1.1517,
+      "step": 2332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001792133577391848,
+      "loss": 1.015,
+      "step": 2333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001792044287691415,
+      "loss": 1.0598,
+      "step": 2334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017919549979909818,
+      "loss": 1.1832,
+      "step": 2335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017918657082905488,
+      "loss": 1.0635,
+      "step": 2336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001791776418590116,
+      "loss": 1.1807,
+      "step": 2337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017916871288896827,
+      "loss": 1.0339,
+      "step": 2338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017915978391892497,
+      "loss": 1.0198,
+      "step": 2339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017915085494888167,
+      "loss": 0.9298,
+      "step": 2340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017914192597883835,
+      "loss": 0.9864,
+      "step": 2341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017913299700879505,
+      "loss": 1.0468,
+      "step": 2342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017912406803875176,
+      "loss": 0.9955,
+      "step": 2343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017911513906870844,
+      "loss": 1.0816,
+      "step": 2344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017910621009866514,
+      "loss": 1.2672,
+      "step": 2345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017909728112862184,
+      "loss": 1.0913,
+      "step": 2346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017908835215857852,
+      "loss": 1.059,
+      "step": 2347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017907942318853522,
+      "loss": 1.172,
+      "step": 2348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001790704942184919,
+      "loss": 1.1198,
+      "step": 2349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001790615652484486,
+      "loss": 1.0641,
+      "step": 2350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017905263627840528,
+      "loss": 1.0699,
+      "step": 2351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017904370730836199,
+      "loss": 1.0103,
+      "step": 2352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017903477833831866,
+      "loss": 1.0968,
+      "step": 2353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017902584936827537,
+      "loss": 0.857,
+      "step": 2354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017901692039823207,
+      "loss": 1.0789,
+      "step": 2355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017900799142818875,
+      "loss": 1.1029,
+      "step": 2356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017899906245814545,
+      "loss": 0.9693,
+      "step": 2357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017899013348810216,
+      "loss": 1.0719,
+      "step": 2358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017898120451805883,
+      "loss": 1.0695,
+      "step": 2359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017897227554801554,
+      "loss": 1.0692,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017896334657797224,
+      "loss": 0.9962,
+      "step": 2361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017895441760792892,
+      "loss": 1.0625,
+      "step": 2362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017894548863788562,
+      "loss": 1.0702,
+      "step": 2363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017893655966784233,
+      "loss": 1.0092,
+      "step": 2364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000178927630697799,
+      "loss": 0.9343,
+      "step": 2365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001789187017277557,
+      "loss": 1.0193,
+      "step": 2366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001789097727577124,
+      "loss": 1.0708,
+      "step": 2367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001789008437876691,
+      "loss": 0.976,
+      "step": 2368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001788919148176258,
+      "loss": 1.0255,
+      "step": 2369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001788829858475825,
+      "loss": 0.9977,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017887405687753917,
+      "loss": 1.0431,
+      "step": 2371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017886512790749588,
+      "loss": 1.1388,
+      "step": 2372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017885619893745258,
+      "loss": 1.1623,
+      "step": 2373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017884726996740926,
+      "loss": 1.0501,
+      "step": 2374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017883834099736596,
+      "loss": 0.9617,
+      "step": 2375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017882941202732267,
+      "loss": 1.0274,
+      "step": 2376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017882048305727934,
+      "loss": 1.0097,
+      "step": 2377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017881155408723605,
+      "loss": 1.0188,
+      "step": 2378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017880262511719275,
+      "loss": 1.1153,
+      "step": 2379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017879369614714943,
+      "loss": 1.1955,
+      "step": 2380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017878476717710613,
+      "loss": 1.0096,
+      "step": 2381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017877583820706284,
+      "loss": 1.0617,
+      "step": 2382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001787669092370195,
+      "loss": 1.1052,
+      "step": 2383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017875798026697622,
+      "loss": 1.1185,
+      "step": 2384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017874905129693292,
+      "loss": 1.0061,
+      "step": 2385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001787401223268896,
+      "loss": 1.0022,
+      "step": 2386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001787311933568463,
+      "loss": 1.001,
+      "step": 2387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000178722264386803,
+      "loss": 1.0352,
+      "step": 2388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017871333541675968,
+      "loss": 1.0578,
+      "step": 2389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001787044064467164,
+      "loss": 1.0721,
+      "step": 2390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001786954774766731,
+      "loss": 1.0525,
+      "step": 2391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017868654850662977,
+      "loss": 1.0004,
+      "step": 2392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017867761953658647,
+      "loss": 0.938,
+      "step": 2393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017866869056654315,
+      "loss": 1.0186,
+      "step": 2394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017865976159649985,
+      "loss": 0.9697,
+      "step": 2395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017865083262645656,
+      "loss": 1.0947,
+      "step": 2396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017864190365641323,
+      "loss": 1.0219,
+      "step": 2397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017863297468636994,
+      "loss": 0.8768,
+      "step": 2398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017862404571632661,
+      "loss": 1.0824,
+      "step": 2399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017861511674628332,
+      "loss": 0.9156,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017860618777624,
+      "loss": 1.06,
+      "step": 2401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001785972588061967,
+      "loss": 1.0946,
+      "step": 2402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001785883298361534,
+      "loss": 0.9566,
+      "step": 2403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001785794008661101,
+      "loss": 0.9838,
+      "step": 2404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017857047189606678,
+      "loss": 1.1125,
+      "step": 2405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001785615429260235,
+      "loss": 1.1222,
+      "step": 2406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001785526139559802,
+      "loss": 1.0708,
+      "step": 2407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017854368498593687,
+      "loss": 0.9837,
+      "step": 2408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017853475601589357,
+      "loss": 1.0098,
+      "step": 2409
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017852582704585028,
+      "loss": 1.1353,
+      "step": 2410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017851689807580695,
+      "loss": 1.0797,
+      "step": 2411
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017850796910576366,
+      "loss": 1.0189,
+      "step": 2412
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017849904013572036,
+      "loss": 1.0727,
+      "step": 2413
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017849011116567704,
+      "loss": 1.0602,
+      "step": 2414
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017848118219563374,
+      "loss": 1.1491,
+      "step": 2415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017847225322559045,
+      "loss": 1.0302,
+      "step": 2416
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017846332425554712,
+      "loss": 1.0943,
+      "step": 2417
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017845439528550383,
+      "loss": 1.1815,
+      "step": 2418
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017844546631546053,
+      "loss": 1.0031,
+      "step": 2419
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001784365373454172,
+      "loss": 1.0858,
+      "step": 2420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017842760837537391,
+      "loss": 1.2985,
+      "step": 2421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017841867940533062,
+      "loss": 1.0596,
+      "step": 2422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001784097504352873,
+      "loss": 1.2339,
+      "step": 2423
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000178400821465244,
+      "loss": 0.9665,
+      "step": 2424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001783918924952007,
+      "loss": 1.0368,
+      "step": 2425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017838296352515738,
+      "loss": 1.0071,
+      "step": 2426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017837403455511408,
+      "loss": 1.0092,
+      "step": 2427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001783651055850708,
+      "loss": 1.0403,
+      "step": 2428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017835617661502746,
+      "loss": 0.9769,
+      "step": 2429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017834724764498417,
+      "loss": 1.0328,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017833831867494087,
+      "loss": 1.193,
+      "step": 2431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017832938970489755,
+      "loss": 1.0162,
+      "step": 2432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017832046073485425,
+      "loss": 0.9922,
+      "step": 2433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017831153176481096,
+      "loss": 1.0615,
+      "step": 2434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017830260279476763,
+      "loss": 1.0531,
+      "step": 2435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017829367382472434,
+      "loss": 0.9764,
+      "step": 2436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017828474485468104,
+      "loss": 1.1475,
+      "step": 2437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017827581588463772,
+      "loss": 1.1297,
+      "step": 2438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017826688691459442,
+      "loss": 1.1668,
+      "step": 2439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001782579579445511,
+      "loss": 1.2011,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001782490289745078,
+      "loss": 0.9575,
+      "step": 2441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017824010000446448,
+      "loss": 0.9808,
+      "step": 2442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017823117103442119,
+      "loss": 1.1557,
+      "step": 2443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017822224206437786,
+      "loss": 1.0635,
+      "step": 2444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017821331309433457,
+      "loss": 1.0543,
+      "step": 2445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017820438412429127,
+      "loss": 0.9676,
+      "step": 2446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017819545515424795,
+      "loss": 1.0778,
+      "step": 2447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017818652618420465,
+      "loss": 0.9925,
+      "step": 2448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017817759721416136,
+      "loss": 1.196,
+      "step": 2449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017816866824411803,
+      "loss": 1.1077,
+      "step": 2450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017815973927407474,
+      "loss": 1.079,
+      "step": 2451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017815081030403144,
+      "loss": 0.9717,
+      "step": 2452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017814188133398812,
+      "loss": 1.1207,
+      "step": 2453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017813295236394482,
+      "loss": 1.1186,
+      "step": 2454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017812402339390153,
+      "loss": 1.1679,
+      "step": 2455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001781150944238582,
+      "loss": 0.9123,
+      "step": 2456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001781061654538149,
+      "loss": 1.1561,
+      "step": 2457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001780972364837716,
+      "loss": 1.044,
+      "step": 2458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001780883075137283,
+      "loss": 1.1654,
+      "step": 2459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000178079378543685,
+      "loss": 1.0475,
+      "step": 2460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001780704495736417,
+      "loss": 1.118,
+      "step": 2461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017806152060359837,
+      "loss": 1.0105,
+      "step": 2462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017805259163355508,
+      "loss": 1.1161,
+      "step": 2463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017804366266351178,
+      "loss": 1.1491,
+      "step": 2464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017803473369346846,
+      "loss": 1.1494,
+      "step": 2465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017802580472342516,
+      "loss": 1.0586,
+      "step": 2466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017801687575338187,
+      "loss": 1.0315,
+      "step": 2467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017800794678333854,
+      "loss": 1.1554,
+      "step": 2468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017799901781329525,
+      "loss": 1.0398,
+      "step": 2469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017799008884325195,
+      "loss": 1.0106,
+      "step": 2470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017798115987320863,
+      "loss": 1.0091,
+      "step": 2471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017797223090316533,
+      "loss": 0.9503,
+      "step": 2472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017796330193312204,
+      "loss": 0.9697,
+      "step": 2473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001779543729630787,
+      "loss": 1.099,
+      "step": 2474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017794544399303542,
+      "loss": 1.008,
+      "step": 2475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017793651502299212,
+      "loss": 1.1045,
+      "step": 2476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001779275860529488,
+      "loss": 1.1133,
+      "step": 2477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001779186570829055,
+      "loss": 1.0535,
+      "step": 2478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001779097281128622,
+      "loss": 1.0621,
+      "step": 2479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017790079914281888,
+      "loss": 1.1661,
+      "step": 2480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001778918701727756,
+      "loss": 1.0365,
+      "step": 2481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001778829412027323,
+      "loss": 1.0166,
+      "step": 2482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017787401223268897,
+      "loss": 1.1469,
+      "step": 2483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017786508326264567,
+      "loss": 1.191,
+      "step": 2484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017785615429260235,
+      "loss": 1.0013,
+      "step": 2485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017784722532255905,
+      "loss": 0.969,
+      "step": 2486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017783829635251576,
+      "loss": 1.1311,
+      "step": 2487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017782936738247243,
+      "loss": 1.0211,
+      "step": 2488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017782043841242914,
+      "loss": 0.8986,
+      "step": 2489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017781150944238581,
+      "loss": 0.9125,
+      "step": 2490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017780258047234252,
+      "loss": 1.1801,
+      "step": 2491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001777936515022992,
+      "loss": 1.0733,
+      "step": 2492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001777847225322559,
+      "loss": 0.9811,
+      "step": 2493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001777757935622126,
+      "loss": 0.9334,
+      "step": 2494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017776686459216928,
+      "loss": 1.0806,
+      "step": 2495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017775793562212598,
+      "loss": 1.0482,
+      "step": 2496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001777490066520827,
+      "loss": 1.0656,
+      "step": 2497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017774007768203937,
+      "loss": 0.9485,
+      "step": 2498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017773114871199607,
+      "loss": 1.012,
+      "step": 2499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017772221974195277,
+      "loss": 1.0184,
+      "step": 2500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017771329077190948,
+      "loss": 0.992,
+      "step": 2501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017770436180186615,
+      "loss": 0.899,
+      "step": 2502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017769543283182286,
+      "loss": 1.079,
+      "step": 2503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017768650386177956,
+      "loss": 1.1188,
+      "step": 2504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017767757489173624,
+      "loss": 1.1737,
+      "step": 2505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017766864592169294,
+      "loss": 0.9366,
+      "step": 2506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017765971695164965,
+      "loss": 1.0167,
+      "step": 2507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017765078798160632,
+      "loss": 1.0813,
+      "step": 2508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017764185901156303,
+      "loss": 1.3384,
+      "step": 2509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017763293004151973,
+      "loss": 1.0566,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001776240010714764,
+      "loss": 1.0454,
+      "step": 2511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001776150721014331,
+      "loss": 1.3052,
+      "step": 2512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017760614313138982,
+      "loss": 1.0211,
+      "step": 2513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001775972141613465,
+      "loss": 1.1392,
+      "step": 2514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001775882851913032,
+      "loss": 0.9667,
+      "step": 2515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001775793562212599,
+      "loss": 1.0053,
+      "step": 2516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017757042725121658,
+      "loss": 1.132,
+      "step": 2517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017756149828117328,
+      "loss": 1.0892,
+      "step": 2518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017755256931113,
+      "loss": 1.0892,
+      "step": 2519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017754364034108666,
+      "loss": 1.0962,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017753471137104337,
+      "loss": 1.0757,
+      "step": 2521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017752578240100007,
+      "loss": 0.9728,
+      "step": 2522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017751685343095675,
+      "loss": 1.0945,
+      "step": 2523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017750792446091345,
+      "loss": 1.0534,
+      "step": 2524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017749899549087016,
+      "loss": 1.0948,
+      "step": 2525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017749006652082683,
+      "loss": 0.9448,
+      "step": 2526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017748113755078354,
+      "loss": 1.0561,
+      "step": 2527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017747220858074024,
+      "loss": 1.1164,
+      "step": 2528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017746327961069692,
+      "loss": 1.05,
+      "step": 2529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017745435064065362,
+      "loss": 1.0046,
+      "step": 2530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001774454216706103,
+      "loss": 1.0086,
+      "step": 2531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000177436492700567,
+      "loss": 1.0734,
+      "step": 2532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017742756373052368,
+      "loss": 1.1003,
+      "step": 2533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017741863476048039,
+      "loss": 1.0072,
+      "step": 2534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017740970579043706,
+      "loss": 1.1359,
+      "step": 2535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017740077682039377,
+      "loss": 1.2929,
+      "step": 2536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017739184785035047,
+      "loss": 1.0129,
+      "step": 2537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017738291888030715,
+      "loss": 1.0344,
+      "step": 2538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017737398991026385,
+      "loss": 1.0322,
+      "step": 2539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017736506094022056,
+      "loss": 1.0922,
+      "step": 2540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017735613197017723,
+      "loss": 1.0309,
+      "step": 2541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017734720300013394,
+      "loss": 1.0755,
+      "step": 2542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017733827403009064,
+      "loss": 0.9995,
+      "step": 2543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017732934506004732,
+      "loss": 1.1505,
+      "step": 2544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017732041609000402,
+      "loss": 0.9514,
+      "step": 2545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017731148711996073,
+      "loss": 1.0017,
+      "step": 2546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001773025581499174,
+      "loss": 1.0621,
+      "step": 2547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001772936291798741,
+      "loss": 0.9481,
+      "step": 2548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001772847002098308,
+      "loss": 1.1017,
+      "step": 2549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001772757712397875,
+      "loss": 1.0505,
+      "step": 2550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001772668422697442,
+      "loss": 1.0244,
+      "step": 2551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001772579132997009,
+      "loss": 1.0312,
+      "step": 2552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017724898432965757,
+      "loss": 1.0069,
+      "step": 2553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017724005535961428,
+      "loss": 1.0074,
+      "step": 2554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017723112638957098,
+      "loss": 1.134,
+      "step": 2555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017722219741952766,
+      "loss": 0.9801,
+      "step": 2556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017721326844948436,
+      "loss": 1.0871,
+      "step": 2557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017720433947944107,
+      "loss": 1.1791,
+      "step": 2558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017719541050939774,
+      "loss": 1.1058,
+      "step": 2559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017718648153935445,
+      "loss": 1.0248,
+      "step": 2560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017717755256931115,
+      "loss": 1.15,
+      "step": 2561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017716862359926783,
+      "loss": 1.1201,
+      "step": 2562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017715969462922453,
+      "loss": 1.0675,
+      "step": 2563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017715076565918124,
+      "loss": 0.9793,
+      "step": 2564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001771418366891379,
+      "loss": 1.111,
+      "step": 2565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017713290771909462,
+      "loss": 1.1012,
+      "step": 2566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017712397874905132,
+      "loss": 1.1454,
+      "step": 2567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000177115049779008,
+      "loss": 1.0771,
+      "step": 2568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001771061208089647,
+      "loss": 1.0802,
+      "step": 2569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001770971918389214,
+      "loss": 1.1636,
+      "step": 2570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017708826286887808,
+      "loss": 1.2627,
+      "step": 2571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017707933389883479,
+      "loss": 1.166,
+      "step": 2572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001770704049287915,
+      "loss": 0.9395,
+      "step": 2573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017706147595874817,
+      "loss": 0.9627,
+      "step": 2574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017705254698870487,
+      "loss": 0.9584,
+      "step": 2575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00017704361801866155,
+      "loss": 1.0165,
+      "step": 2576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017703468904861825,
+      "loss": 1.0772,
+      "step": 2577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017702576007857496,
+      "loss": 1.0784,
+      "step": 2578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017701683110853163,
+      "loss": 0.9676,
+      "step": 2579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017700790213848834,
+      "loss": 1.0489,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017699897316844501,
+      "loss": 1.0624,
+      "step": 2581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017699004419840172,
+      "loss": 0.9362,
+      "step": 2582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001769811152283584,
+      "loss": 1.0105,
+      "step": 2583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001769721862583151,
+      "loss": 1.211,
+      "step": 2584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001769632572882718,
+      "loss": 1.1339,
+      "step": 2585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017695432831822848,
+      "loss": 1.0281,
+      "step": 2586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017694539934818518,
+      "loss": 1.1055,
+      "step": 2587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001769364703781419,
+      "loss": 0.9595,
+      "step": 2588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017692754140809856,
+      "loss": 1.0413,
+      "step": 2589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017691861243805527,
+      "loss": 1.0874,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017690968346801197,
+      "loss": 1.0632,
+      "step": 2591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017690075449796865,
+      "loss": 1.1291,
+      "step": 2592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017689182552792535,
+      "loss": 1.0852,
+      "step": 2593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017688289655788206,
+      "loss": 1.0858,
+      "step": 2594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017687396758783873,
+      "loss": 1.1382,
+      "step": 2595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017686503861779544,
+      "loss": 1.1534,
+      "step": 2596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017685610964775214,
+      "loss": 1.1027,
+      "step": 2597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017684718067770885,
+      "loss": 1.0101,
+      "step": 2598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017683825170766552,
+      "loss": 1.0126,
+      "step": 2599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017682932273762223,
+      "loss": 0.9993,
+      "step": 2600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017682039376757893,
+      "loss": 1.1005,
+      "step": 2601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001768114647975356,
+      "loss": 1.0585,
+      "step": 2602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001768025358274923,
+      "loss": 1.0556,
+      "step": 2603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017679360685744902,
+      "loss": 1.0358,
+      "step": 2604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001767846778874057,
+      "loss": 1.0916,
+      "step": 2605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001767757489173624,
+      "loss": 1.1355,
+      "step": 2606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001767668199473191,
+      "loss": 1.0852,
+      "step": 2607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017675789097727578,
+      "loss": 0.9857,
+      "step": 2608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017674896200723248,
+      "loss": 0.9385,
+      "step": 2609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001767400330371892,
+      "loss": 1.1005,
+      "step": 2610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017673110406714586,
+      "loss": 1.0952,
+      "step": 2611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017672217509710257,
+      "loss": 1.1132,
+      "step": 2612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017671324612705927,
+      "loss": 1.1633,
+      "step": 2613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017670431715701595,
+      "loss": 1.2049,
+      "step": 2614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017669538818697265,
+      "loss": 1.0287,
+      "step": 2615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017668645921692936,
+      "loss": 1.0132,
+      "step": 2616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017667753024688603,
+      "loss": 1.0802,
+      "step": 2617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017666860127684274,
+      "loss": 1.1541,
+      "step": 2618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017665967230679944,
+      "loss": 1.06,
+      "step": 2619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017665074333675612,
+      "loss": 0.93,
+      "step": 2620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017664181436671282,
+      "loss": 1.1098,
+      "step": 2621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001766328853966695,
+      "loss": 1.0087,
+      "step": 2622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001766239564266262,
+      "loss": 1.0993,
+      "step": 2623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017661502745658288,
+      "loss": 1.0459,
+      "step": 2624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017660609848653958,
+      "loss": 1.0587,
+      "step": 2625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017659716951649626,
+      "loss": 1.1556,
+      "step": 2626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017658824054645297,
+      "loss": 1.1658,
+      "step": 2627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017657931157640967,
+      "loss": 0.968,
+      "step": 2628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017657038260636635,
+      "loss": 1.0337,
+      "step": 2629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017656145363632305,
+      "loss": 1.0259,
+      "step": 2630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017655252466627975,
+      "loss": 1.0066,
+      "step": 2631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017654359569623643,
+      "loss": 1.1156,
+      "step": 2632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017653466672619314,
+      "loss": 0.9972,
+      "step": 2633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017652573775614984,
+      "loss": 1.0249,
+      "step": 2634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017651680878610652,
+      "loss": 1.0766,
+      "step": 2635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017650787981606322,
+      "loss": 1.1599,
+      "step": 2636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017649895084601992,
+      "loss": 1.0378,
+      "step": 2637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001764900218759766,
+      "loss": 1.1074,
+      "step": 2638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001764810929059333,
+      "loss": 0.9542,
+      "step": 2639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017647216393589,
+      "loss": 1.0437,
+      "step": 2640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001764632349658467,
+      "loss": 1.0903,
+      "step": 2641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001764543059958034,
+      "loss": 1.0074,
+      "step": 2642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001764453770257601,
+      "loss": 1.0002,
+      "step": 2643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017643644805571677,
+      "loss": 1.0271,
+      "step": 2644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017642751908567348,
+      "loss": 1.2232,
+      "step": 2645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017641859011563018,
+      "loss": 1.0935,
+      "step": 2646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017640966114558686,
+      "loss": 1.0636,
+      "step": 2647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017640073217554356,
+      "loss": 1.0973,
+      "step": 2648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017639180320550026,
+      "loss": 1.1175,
+      "step": 2649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017638287423545694,
+      "loss": 1.1047,
+      "step": 2650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00017637394526541365,
+      "loss": 1.0277,
+      "step": 2651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017636501629537035,
+      "loss": 0.9741,
+      "step": 2652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017635608732532703,
+      "loss": 1.0477,
+      "step": 2653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017634715835528373,
+      "loss": 1.1169,
+      "step": 2654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017633822938524043,
+      "loss": 0.948,
+      "step": 2655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001763293004151971,
+      "loss": 1.0301,
+      "step": 2656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017632037144515382,
+      "loss": 1.0701,
+      "step": 2657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017631144247511052,
+      "loss": 1.175,
+      "step": 2658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001763025135050672,
+      "loss": 0.9557,
+      "step": 2659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001762935845350239,
+      "loss": 1.0926,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001762846555649806,
+      "loss": 1.2059,
+      "step": 2661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017627572659493728,
+      "loss": 0.9906,
+      "step": 2662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017626679762489399,
+      "loss": 1.1107,
+      "step": 2663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001762578686548507,
+      "loss": 0.9941,
+      "step": 2664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017624893968480737,
+      "loss": 1.0145,
+      "step": 2665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017624001071476407,
+      "loss": 1.0808,
+      "step": 2666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017623108174472075,
+      "loss": 1.0352,
+      "step": 2667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017622215277467745,
+      "loss": 1.0251,
+      "step": 2668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017621322380463413,
+      "loss": 1.1378,
+      "step": 2669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017620429483459083,
+      "loss": 0.9702,
+      "step": 2670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017619536586454754,
+      "loss": 0.9353,
+      "step": 2671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017618643689450421,
+      "loss": 1.2424,
+      "step": 2672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017617750792446092,
+      "loss": 1.052,
+      "step": 2673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001761685789544176,
+      "loss": 1.0361,
+      "step": 2674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001761596499843743,
+      "loss": 1.142,
+      "step": 2675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000176150721014331,
+      "loss": 0.93,
+      "step": 2676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017614179204428768,
+      "loss": 1.0436,
+      "step": 2677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017613286307424438,
+      "loss": 0.9719,
+      "step": 2678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001761239341042011,
+      "loss": 1.0459,
+      "step": 2679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017611500513415776,
+      "loss": 1.0425,
+      "step": 2680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017610607616411447,
+      "loss": 1.091,
+      "step": 2681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017609714719407117,
+      "loss": 1.1332,
+      "step": 2682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017608821822402785,
+      "loss": 0.9202,
+      "step": 2683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017607928925398455,
+      "loss": 1.0796,
+      "step": 2684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017607036028394126,
+      "loss": 0.9457,
+      "step": 2685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017606143131389793,
+      "loss": 1.0752,
+      "step": 2686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017605250234385464,
+      "loss": 0.9803,
+      "step": 2687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017604357337381134,
+      "loss": 1.0226,
+      "step": 2688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017603464440376802,
+      "loss": 0.9629,
+      "step": 2689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017602571543372472,
+      "loss": 1.1396,
+      "step": 2690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017601678646368143,
+      "loss": 1.0227,
+      "step": 2691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001760078574936381,
+      "loss": 0.924,
+      "step": 2692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001759989285235948,
+      "loss": 1.084,
+      "step": 2693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001759899995535515,
+      "loss": 1.0738,
+      "step": 2694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017598107058350822,
+      "loss": 0.965,
+      "step": 2695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001759721416134649,
+      "loss": 1.1126,
+      "step": 2696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001759632126434216,
+      "loss": 1.1553,
+      "step": 2697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001759542836733783,
+      "loss": 1.0162,
+      "step": 2698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017594535470333498,
+      "loss": 1.03,
+      "step": 2699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017593642573329168,
+      "loss": 1.0582,
+      "step": 2700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001759274967632484,
+      "loss": 1.0526,
+      "step": 2701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017591856779320506,
+      "loss": 0.9844,
+      "step": 2702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017590963882316177,
+      "loss": 1.0372,
+      "step": 2703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017590070985311847,
+      "loss": 1.0384,
+      "step": 2704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017589178088307515,
+      "loss": 0.9668,
+      "step": 2705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017588285191303185,
+      "loss": 1.2301,
+      "step": 2706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017587392294298856,
+      "loss": 1.0174,
+      "step": 2707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017586499397294523,
+      "loss": 1.2366,
+      "step": 2708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017585606500290194,
+      "loss": 1.1125,
+      "step": 2709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017584713603285864,
+      "loss": 1.1201,
+      "step": 2710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017583820706281532,
+      "loss": 1.1758,
+      "step": 2711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017582927809277202,
+      "loss": 1.0759,
+      "step": 2712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001758203491227287,
+      "loss": 1.0733,
+      "step": 2713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001758114201526854,
+      "loss": 0.882,
+      "step": 2714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017580249118264208,
+      "loss": 1.0334,
+      "step": 2715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017579356221259878,
+      "loss": 1.0391,
+      "step": 2716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017578463324255546,
+      "loss": 1.1405,
+      "step": 2717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017577570427251217,
+      "loss": 1.2199,
+      "step": 2718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017576677530246887,
+      "loss": 1.1562,
+      "step": 2719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017575784633242555,
+      "loss": 0.9304,
+      "step": 2720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017574891736238225,
+      "loss": 1.055,
+      "step": 2721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017573998839233895,
+      "loss": 1.0449,
+      "step": 2722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017573105942229563,
+      "loss": 0.9671,
+      "step": 2723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017572213045225234,
+      "loss": 0.9892,
+      "step": 2724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017571320148220904,
+      "loss": 1.092,
+      "step": 2725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00017570427251216572,
+      "loss": 1.038,
+      "step": 2726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017569534354212242,
+      "loss": 1.0516,
+      "step": 2727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017568641457207912,
+      "loss": 1.0104,
+      "step": 2728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001756774856020358,
+      "loss": 1.1347,
+      "step": 2729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001756685566319925,
+      "loss": 0.9285,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001756596276619492,
+      "loss": 1.0814,
+      "step": 2731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017565069869190589,
+      "loss": 1.0787,
+      "step": 2732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001756417697218626,
+      "loss": 0.9353,
+      "step": 2733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001756328407518193,
+      "loss": 1.06,
+      "step": 2734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017562391178177597,
+      "loss": 1.2371,
+      "step": 2735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017561498281173268,
+      "loss": 0.98,
+      "step": 2736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017560605384168938,
+      "loss": 1.0917,
+      "step": 2737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017559712487164606,
+      "loss": 1.0919,
+      "step": 2738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017558819590160276,
+      "loss": 1.0154,
+      "step": 2739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017557926693155946,
+      "loss": 1.0277,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017557033796151614,
+      "loss": 0.9793,
+      "step": 2741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017556140899147285,
+      "loss": 0.9854,
+      "step": 2742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017555248002142955,
+      "loss": 1.0365,
+      "step": 2743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017554355105138623,
+      "loss": 1.0684,
+      "step": 2744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017553462208134293,
+      "loss": 1.0708,
+      "step": 2745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017552569311129963,
+      "loss": 1.0349,
+      "step": 2746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001755167641412563,
+      "loss": 1.218,
+      "step": 2747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017550783517121302,
+      "loss": 0.9883,
+      "step": 2748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017549890620116972,
+      "loss": 1.0777,
+      "step": 2749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001754899772311264,
+      "loss": 1.0348,
+      "step": 2750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001754810482610831,
+      "loss": 1.1138,
+      "step": 2751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001754721192910398,
+      "loss": 1.0193,
+      "step": 2752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017546319032099648,
+      "loss": 1.0331,
+      "step": 2753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017545426135095319,
+      "loss": 0.929,
+      "step": 2754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001754453323809099,
+      "loss": 0.9307,
+      "step": 2755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017543640341086657,
+      "loss": 1.0833,
+      "step": 2756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017542747444082327,
+      "loss": 1.0426,
+      "step": 2757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017541854547077995,
+      "loss": 1.0812,
+      "step": 2758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017540961650073665,
+      "loss": 1.0818,
+      "step": 2759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017540068753069333,
+      "loss": 1.0154,
+      "step": 2760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017539175856065003,
+      "loss": 1.1555,
+      "step": 2761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017538282959060674,
+      "loss": 1.0412,
+      "step": 2762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001753739006205634,
+      "loss": 1.0006,
+      "step": 2763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017536497165052012,
+      "loss": 0.9777,
+      "step": 2764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001753560426804768,
+      "loss": 0.9924,
+      "step": 2765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001753471137104335,
+      "loss": 1.0058,
+      "step": 2766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001753381847403902,
+      "loss": 1.0275,
+      "step": 2767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017532925577034688,
+      "loss": 0.9408,
+      "step": 2768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017532032680030358,
+      "loss": 1.0298,
+      "step": 2769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001753113978302603,
+      "loss": 0.9584,
+      "step": 2770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017530246886021696,
+      "loss": 1.0601,
+      "step": 2771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017529353989017367,
+      "loss": 0.9585,
+      "step": 2772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017528461092013037,
+      "loss": 1.1005,
+      "step": 2773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017527568195008705,
+      "loss": 1.0727,
+      "step": 2774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017526675298004375,
+      "loss": 0.938,
+      "step": 2775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017525782401000046,
+      "loss": 1.0845,
+      "step": 2776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017524889503995713,
+      "loss": 1.0344,
+      "step": 2777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017523996606991384,
+      "loss": 1.0861,
+      "step": 2778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017523103709987054,
+      "loss": 1.1058,
+      "step": 2779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017522210812982722,
+      "loss": 1.0512,
+      "step": 2780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017521317915978392,
+      "loss": 1.1187,
+      "step": 2781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017520425018974063,
+      "loss": 1.0196,
+      "step": 2782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001751953212196973,
+      "loss": 0.8631,
+      "step": 2783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000175186392249654,
+      "loss": 1.0425,
+      "step": 2784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001751774632796107,
+      "loss": 1.1017,
+      "step": 2785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001751685343095674,
+      "loss": 1.2328,
+      "step": 2786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001751596053395241,
+      "loss": 1.0973,
+      "step": 2787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001751506763694808,
+      "loss": 0.9235,
+      "step": 2788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017514174739943747,
+      "loss": 0.9768,
+      "step": 2789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017513281842939418,
+      "loss": 0.9377,
+      "step": 2790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017512388945935088,
+      "loss": 1.0734,
+      "step": 2791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017511496048930756,
+      "loss": 1.0668,
+      "step": 2792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017510603151926426,
+      "loss": 1.0144,
+      "step": 2793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017509710254922097,
+      "loss": 1.1108,
+      "step": 2794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017508817357917767,
+      "loss": 1.0992,
+      "step": 2795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017507924460913435,
+      "loss": 0.9549,
+      "step": 2796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017507031563909105,
+      "loss": 1.0998,
+      "step": 2797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017506138666904776,
+      "loss": 1.2333,
+      "step": 2798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017505245769900443,
+      "loss": 1.0272,
+      "step": 2799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017504352872896114,
+      "loss": 0.9986,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017503459975891781,
+      "loss": 1.198,
+      "step": 2801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017502567078887452,
+      "loss": 1.2873,
+      "step": 2802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017501674181883122,
+      "loss": 0.9589,
+      "step": 2803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001750078128487879,
+      "loss": 1.1052,
+      "step": 2804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001749988838787446,
+      "loss": 0.9954,
+      "step": 2805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017498995490870128,
+      "loss": 0.9577,
+      "step": 2806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017498102593865798,
+      "loss": 1.0672,
+      "step": 2807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017497209696861466,
+      "loss": 1.141,
+      "step": 2808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017496316799857137,
+      "loss": 0.9888,
+      "step": 2809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017495423902852807,
+      "loss": 1.0679,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017494531005848475,
+      "loss": 1.1781,
+      "step": 2811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017493638108844145,
+      "loss": 0.9814,
+      "step": 2812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017492745211839815,
+      "loss": 1.0125,
+      "step": 2813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017491852314835483,
+      "loss": 1.1458,
+      "step": 2814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017490959417831154,
+      "loss": 1.0846,
+      "step": 2815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017490066520826824,
+      "loss": 1.0313,
+      "step": 2816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017489173623822492,
+      "loss": 1.031,
+      "step": 2817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017488280726818162,
+      "loss": 1.1267,
+      "step": 2818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017487387829813832,
+      "loss": 0.9717,
+      "step": 2819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000174864949328095,
+      "loss": 1.0906,
+      "step": 2820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001748560203580517,
+      "loss": 1.1334,
+      "step": 2821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001748470913880084,
+      "loss": 1.0133,
+      "step": 2822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017483816241796509,
+      "loss": 1.1185,
+      "step": 2823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001748292334479218,
+      "loss": 1.0654,
+      "step": 2824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001748203044778785,
+      "loss": 1.0355,
+      "step": 2825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017481137550783517,
+      "loss": 1.001,
+      "step": 2826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017480244653779188,
+      "loss": 1.0971,
+      "step": 2827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017479351756774858,
+      "loss": 1.0465,
+      "step": 2828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017478458859770526,
+      "loss": 1.0594,
+      "step": 2829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017477565962766196,
+      "loss": 1.0558,
+      "step": 2830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017476673065761866,
+      "loss": 1.0096,
+      "step": 2831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017475780168757534,
+      "loss": 0.9812,
+      "step": 2832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017474887271753205,
+      "loss": 1.0485,
+      "step": 2833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017473994374748875,
+      "loss": 1.1488,
+      "step": 2834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017473101477744543,
+      "loss": 0.9948,
+      "step": 2835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017472208580740213,
+      "loss": 1.0937,
+      "step": 2836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017471315683735883,
+      "loss": 0.9713,
+      "step": 2837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001747042278673155,
+      "loss": 1.2105,
+      "step": 2838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017469529889727222,
+      "loss": 0.8731,
+      "step": 2839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017468636992722892,
+      "loss": 1.04,
+      "step": 2840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001746774409571856,
+      "loss": 1.009,
+      "step": 2841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001746685119871423,
+      "loss": 0.9781,
+      "step": 2842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000174659583017099,
+      "loss": 0.9724,
+      "step": 2843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017465065404705568,
+      "loss": 1.2281,
+      "step": 2844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017464172507701239,
+      "loss": 0.9611,
+      "step": 2845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001746327961069691,
+      "loss": 1.0514,
+      "step": 2846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017462386713692577,
+      "loss": 1.1473,
+      "step": 2847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017461493816688247,
+      "loss": 0.9466,
+      "step": 2848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017460600919683915,
+      "loss": 1.024,
+      "step": 2849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017459708022679585,
+      "loss": 1.173,
+      "step": 2850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017458815125675253,
+      "loss": 1.1232,
+      "step": 2851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017457922228670923,
+      "loss": 1.1499,
+      "step": 2852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017457029331666594,
+      "loss": 1.207,
+      "step": 2853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001745613643466226,
+      "loss": 1.0206,
+      "step": 2854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017455243537657932,
+      "loss": 1.0112,
+      "step": 2855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000174543506406536,
+      "loss": 1.2888,
+      "step": 2856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001745345774364927,
+      "loss": 1.1178,
+      "step": 2857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001745256484664494,
+      "loss": 0.9793,
+      "step": 2858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017451671949640608,
+      "loss": 1.1357,
+      "step": 2859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017450779052636278,
+      "loss": 1.0873,
+      "step": 2860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001744988615563195,
+      "loss": 1.0417,
+      "step": 2861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017448993258627616,
+      "loss": 0.9523,
+      "step": 2862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017448100361623287,
+      "loss": 1.1031,
+      "step": 2863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017447207464618957,
+      "loss": 1.1569,
+      "step": 2864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017446314567614625,
+      "loss": 1.3712,
+      "step": 2865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017445421670610295,
+      "loss": 1.0087,
+      "step": 2866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017444528773605966,
+      "loss": 1.0554,
+      "step": 2867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017443635876601633,
+      "loss": 1.0423,
+      "step": 2868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017442742979597304,
+      "loss": 1.0809,
+      "step": 2869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017441850082592974,
+      "loss": 1.1294,
+      "step": 2870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017440957185588642,
+      "loss": 1.0953,
+      "step": 2871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017440064288584312,
+      "loss": 0.9583,
+      "step": 2872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00017439171391579983,
+      "loss": 1.0183,
+      "step": 2873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001743827849457565,
+      "loss": 0.9742,
+      "step": 2874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001743738559757132,
+      "loss": 1.055,
+      "step": 2875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001743649270056699,
+      "loss": 1.0396,
+      "step": 2876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001743559980356266,
+      "loss": 1.1055,
+      "step": 2877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001743470690655833,
+      "loss": 1.0261,
+      "step": 2878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017433814009554,
+      "loss": 0.999,
+      "step": 2879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017432921112549667,
+      "loss": 1.0622,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017432028215545338,
+      "loss": 1.0802,
+      "step": 2881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017431135318541008,
+      "loss": 1.0194,
+      "step": 2882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017430242421536676,
+      "loss": 1.09,
+      "step": 2883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017429349524532346,
+      "loss": 1.0003,
+      "step": 2884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017428456627528017,
+      "loss": 1.2805,
+      "step": 2885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017427563730523684,
+      "loss": 1.0565,
+      "step": 2886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017426670833519355,
+      "loss": 1.0656,
+      "step": 2887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017425777936515025,
+      "loss": 1.0784,
+      "step": 2888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017424885039510693,
+      "loss": 1.1604,
+      "step": 2889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017423992142506363,
+      "loss": 1.043,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017423099245502034,
+      "loss": 0.9081,
+      "step": 2891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017422206348497701,
+      "loss": 1.1469,
+      "step": 2892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017421313451493372,
+      "loss": 1.0869,
+      "step": 2893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017420420554489042,
+      "loss": 1.2408,
+      "step": 2894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001741952765748471,
+      "loss": 0.9875,
+      "step": 2895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001741863476048038,
+      "loss": 0.9436,
+      "step": 2896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017417741863476048,
+      "loss": 0.9973,
+      "step": 2897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017416848966471718,
+      "loss": 0.9702,
+      "step": 2898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017415956069467386,
+      "loss": 1.111,
+      "step": 2899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017415063172463056,
+      "loss": 1.1305,
+      "step": 2900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017414170275458727,
+      "loss": 1.0683,
+      "step": 2901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017413277378454395,
+      "loss": 1.0753,
+      "step": 2902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017412384481450065,
+      "loss": 1.1573,
+      "step": 2903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017411491584445735,
+      "loss": 1.045,
+      "step": 2904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017410598687441403,
+      "loss": 1.1811,
+      "step": 2905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017409705790437073,
+      "loss": 1.0671,
+      "step": 2906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017408812893432744,
+      "loss": 1.6098,
+      "step": 2907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017407919996428412,
+      "loss": 1.0571,
+      "step": 2908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017407027099424082,
+      "loss": 1.0695,
+      "step": 2909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017406134202419752,
+      "loss": 1.1327,
+      "step": 2910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001740524130541542,
+      "loss": 1.0708,
+      "step": 2911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001740434840841109,
+      "loss": 0.9917,
+      "step": 2912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001740345551140676,
+      "loss": 0.9324,
+      "step": 2913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017402562614402429,
+      "loss": 1.0802,
+      "step": 2914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000174016697173981,
+      "loss": 1.1753,
+      "step": 2915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001740077682039377,
+      "loss": 1.1665,
+      "step": 2916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017399883923389437,
+      "loss": 0.9917,
+      "step": 2917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017398991026385107,
+      "loss": 1.0328,
+      "step": 2918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017398098129380778,
+      "loss": 1.0662,
+      "step": 2919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017397205232376446,
+      "loss": 0.9285,
+      "step": 2920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017396312335372116,
+      "loss": 1.0426,
+      "step": 2921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017395419438367786,
+      "loss": 1.0465,
+      "step": 2922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017394526541363454,
+      "loss": 1.0275,
+      "step": 2923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017393633644359124,
+      "loss": 1.2087,
+      "step": 2924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017392740747354795,
+      "loss": 1.0917,
+      "step": 2925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017391847850350463,
+      "loss": 1.0881,
+      "step": 2926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017390954953346133,
+      "loss": 1.0188,
+      "step": 2927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017390062056341803,
+      "loss": 1.0114,
+      "step": 2928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001738916915933747,
+      "loss": 0.9468,
+      "step": 2929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017388276262333141,
+      "loss": 1.1344,
+      "step": 2930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017387383365328812,
+      "loss": 1.0102,
+      "step": 2931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001738649046832448,
+      "loss": 1.0523,
+      "step": 2932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001738559757132015,
+      "loss": 1.0756,
+      "step": 2933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001738470467431582,
+      "loss": 0.9967,
+      "step": 2934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017383811777311488,
+      "loss": 0.9467,
+      "step": 2935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017382918880307158,
+      "loss": 0.9552,
+      "step": 2936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001738202598330283,
+      "loss": 0.9947,
+      "step": 2937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017381133086298497,
+      "loss": 1.2416,
+      "step": 2938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017380240189294167,
+      "loss": 1.0185,
+      "step": 2939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017379347292289835,
+      "loss": 1.1483,
+      "step": 2940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017378454395285505,
+      "loss": 1.1489,
+      "step": 2941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017377561498281173,
+      "loss": 0.9906,
+      "step": 2942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017376668601276843,
+      "loss": 1.0717,
+      "step": 2943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017375775704272514,
+      "loss": 1.0417,
+      "step": 2944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001737488280726818,
+      "loss": 1.1956,
+      "step": 2945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017373989910263852,
+      "loss": 1.0926,
+      "step": 2946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001737309701325952,
+      "loss": 0.9911,
+      "step": 2947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001737220411625519,
+      "loss": 1.1003,
+      "step": 2948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001737131121925086,
+      "loss": 1.0975,
+      "step": 2949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00017370418322246528,
+      "loss": 1.1523,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017369525425242198,
+      "loss": 1.1219,
+      "step": 2951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001736863252823787,
+      "loss": 0.9588,
+      "step": 2952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017367739631233536,
+      "loss": 1.1312,
+      "step": 2953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017366846734229207,
+      "loss": 1.0495,
+      "step": 2954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017365953837224877,
+      "loss": 1.0201,
+      "step": 2955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017365060940220545,
+      "loss": 1.0535,
+      "step": 2956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017364168043216215,
+      "loss": 1.0277,
+      "step": 2957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017363275146211886,
+      "loss": 1.0791,
+      "step": 2958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017362382249207553,
+      "loss": 1.0075,
+      "step": 2959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017361489352203224,
+      "loss": 0.9743,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017360596455198894,
+      "loss": 0.9759,
+      "step": 2961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017359703558194562,
+      "loss": 1.0862,
+      "step": 2962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017358810661190232,
+      "loss": 1.0629,
+      "step": 2963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017357917764185903,
+      "loss": 1.056,
+      "step": 2964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001735702486718157,
+      "loss": 1.0221,
+      "step": 2965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001735613197017724,
+      "loss": 0.9537,
+      "step": 2966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001735523907317291,
+      "loss": 1.0135,
+      "step": 2967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001735434617616858,
+      "loss": 1.0198,
+      "step": 2968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001735345327916425,
+      "loss": 1.002,
+      "step": 2969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001735256038215992,
+      "loss": 1.1164,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017351667485155587,
+      "loss": 1.0585,
+      "step": 2971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017350774588151258,
+      "loss": 1.0842,
+      "step": 2972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017349881691146928,
+      "loss": 1.0199,
+      "step": 2973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017348988794142596,
+      "loss": 1.118,
+      "step": 2974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017348095897138266,
+      "loss": 0.9497,
+      "step": 2975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017347203000133937,
+      "loss": 1.0268,
+      "step": 2976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017346310103129604,
+      "loss": 0.8917,
+      "step": 2977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017345417206125275,
+      "loss": 1.0029,
+      "step": 2978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017344524309120945,
+      "loss": 1.0507,
+      "step": 2979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017343631412116613,
+      "loss": 1.0804,
+      "step": 2980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017342738515112283,
+      "loss": 1.0206,
+      "step": 2981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017341845618107954,
+      "loss": 1.0161,
+      "step": 2982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001734095272110362,
+      "loss": 1.0612,
+      "step": 2983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017340059824099292,
+      "loss": 0.9689,
+      "step": 2984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017339166927094962,
+      "loss": 1.0142,
+      "step": 2985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001733827403009063,
+      "loss": 1.0335,
+      "step": 2986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000173373811330863,
+      "loss": 1.0505,
+      "step": 2987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017336488236081968,
+      "loss": 1.1162,
+      "step": 2988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017335595339077638,
+      "loss": 1.0128,
+      "step": 2989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017334702442073306,
+      "loss": 0.8915,
+      "step": 2990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017333809545068976,
+      "loss": 1.0578,
+      "step": 2991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017332916648064647,
+      "loss": 1.0835,
+      "step": 2992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017332023751060315,
+      "loss": 1.0526,
+      "step": 2993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017331130854055985,
+      "loss": 1.0065,
+      "step": 2994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017330237957051655,
+      "loss": 1.0154,
+      "step": 2995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017329345060047323,
+      "loss": 1.0564,
+      "step": 2996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017328452163042993,
+      "loss": 1.0363,
+      "step": 2997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017327559266038664,
+      "loss": 1.017,
+      "step": 2998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017326666369034332,
+      "loss": 1.1061,
+      "step": 2999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017325773472030002,
+      "loss": 1.0343,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017324880575025672,
+      "loss": 0.9717,
+      "step": 3001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001732398767802134,
+      "loss": 0.9687,
+      "step": 3002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001732309478101701,
+      "loss": 1.0773,
+      "step": 3003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001732220188401268,
+      "loss": 1.1275,
+      "step": 3004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017321308987008349,
+      "loss": 1.035,
+      "step": 3005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001732041609000402,
+      "loss": 1.0254,
+      "step": 3006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001731952319299969,
+      "loss": 0.9925,
+      "step": 3007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017318630295995357,
+      "loss": 1.2076,
+      "step": 3008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017317737398991027,
+      "loss": 1.0121,
+      "step": 3009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017316844501986698,
+      "loss": 0.9257,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017315951604982366,
+      "loss": 0.961,
+      "step": 3011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017315058707978036,
+      "loss": 1.1246,
+      "step": 3012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017314165810973706,
+      "loss": 0.9865,
+      "step": 3013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017313272913969374,
+      "loss": 1.0311,
+      "step": 3014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017312380016965044,
+      "loss": 1.1249,
+      "step": 3015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017311487119960715,
+      "loss": 1.1769,
+      "step": 3016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017310594222956383,
+      "loss": 1.1129,
+      "step": 3017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017309701325952053,
+      "loss": 1.1296,
+      "step": 3018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017308808428947723,
+      "loss": 1.0178,
+      "step": 3019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001730791553194339,
+      "loss": 0.9522,
+      "step": 3020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017307022634939061,
+      "loss": 1.026,
+      "step": 3021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00017306129737934732,
+      "loss": 1.0156,
+      "step": 3022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000173052368409304,
+      "loss": 1.0564,
+      "step": 3023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001730434394392607,
+      "loss": 0.9361,
+      "step": 3024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001730345104692174,
+      "loss": 1.0548,
+      "step": 3025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017302558149917408,
+      "loss": 1.1101,
+      "step": 3026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017301665252913078,
+      "loss": 1.0181,
+      "step": 3027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001730077235590875,
+      "loss": 0.9538,
+      "step": 3028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017299879458904417,
+      "loss": 0.9938,
+      "step": 3029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017298986561900087,
+      "loss": 1.1248,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017298093664895755,
+      "loss": 1.1705,
+      "step": 3031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017297200767891425,
+      "loss": 1.0256,
+      "step": 3032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017296307870887093,
+      "loss": 0.9883,
+      "step": 3033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017295414973882763,
+      "loss": 1.0367,
+      "step": 3034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017294522076878434,
+      "loss": 1.0803,
+      "step": 3035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000172936291798741,
+      "loss": 1.0563,
+      "step": 3036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017292736282869772,
+      "loss": 1.0852,
+      "step": 3037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001729184338586544,
+      "loss": 1.0341,
+      "step": 3038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001729095048886111,
+      "loss": 1.2311,
+      "step": 3039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001729005759185678,
+      "loss": 1.1449,
+      "step": 3040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017289164694852448,
+      "loss": 1.0992,
+      "step": 3041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017288271797848118,
+      "loss": 0.9727,
+      "step": 3042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017287378900843789,
+      "loss": 1.0608,
+      "step": 3043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017286486003839456,
+      "loss": 0.9755,
+      "step": 3044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017285593106835127,
+      "loss": 1.1119,
+      "step": 3045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017284700209830797,
+      "loss": 0.8531,
+      "step": 3046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017283807312826465,
+      "loss": 0.9139,
+      "step": 3047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017282914415822135,
+      "loss": 0.9694,
+      "step": 3048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017282021518817806,
+      "loss": 1.0211,
+      "step": 3049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017281128621813473,
+      "loss": 0.9788,
+      "step": 3050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017280235724809144,
+      "loss": 1.0754,
+      "step": 3051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017279342827804814,
+      "loss": 1.0252,
+      "step": 3052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017278449930800482,
+      "loss": 1.0626,
+      "step": 3053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017277557033796152,
+      "loss": 1.1383,
+      "step": 3054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017276664136791823,
+      "loss": 1.0045,
+      "step": 3055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001727577123978749,
+      "loss": 1.1201,
+      "step": 3056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001727487834278316,
+      "loss": 1.0576,
+      "step": 3057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001727398544577883,
+      "loss": 1.0724,
+      "step": 3058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000172730925487745,
+      "loss": 1.1246,
+      "step": 3059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001727219965177017,
+      "loss": 0.9873,
+      "step": 3060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001727130675476584,
+      "loss": 1.0281,
+      "step": 3061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017270413857761507,
+      "loss": 1.1292,
+      "step": 3062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017269520960757178,
+      "loss": 1.0442,
+      "step": 3063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017268628063752848,
+      "loss": 1.059,
+      "step": 3064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017267735166748516,
+      "loss": 1.027,
+      "step": 3065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017266842269744186,
+      "loss": 1.0749,
+      "step": 3066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017265949372739857,
+      "loss": 1.0608,
+      "step": 3067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017265056475735524,
+      "loss": 1.0259,
+      "step": 3068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017264163578731195,
+      "loss": 1.1027,
+      "step": 3069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017263270681726865,
+      "loss": 1.0432,
+      "step": 3070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017262377784722533,
+      "loss": 1.1514,
+      "step": 3071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017261484887718203,
+      "loss": 1.1682,
+      "step": 3072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017260591990713874,
+      "loss": 1.2739,
+      "step": 3073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001725969909370954,
+      "loss": 1.0975,
+      "step": 3074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017258806196705212,
+      "loss": 1.0319,
+      "step": 3075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017257913299700882,
+      "loss": 1.0534,
+      "step": 3076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001725702040269655,
+      "loss": 1.0225,
+      "step": 3077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001725612750569222,
+      "loss": 1.2001,
+      "step": 3078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017255234608687888,
+      "loss": 1.1026,
+      "step": 3079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017254341711683558,
+      "loss": 1.0114,
+      "step": 3080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017253448814679226,
+      "loss": 0.9826,
+      "step": 3081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017252555917674896,
+      "loss": 1.1355,
+      "step": 3082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017251663020670564,
+      "loss": 1.0645,
+      "step": 3083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017250770123666234,
+      "loss": 0.9212,
+      "step": 3084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017249877226661905,
+      "loss": 1.044,
+      "step": 3085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017248984329657573,
+      "loss": 0.9326,
+      "step": 3086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017248091432653243,
+      "loss": 1.0576,
+      "step": 3087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017247198535648913,
+      "loss": 1.1016,
+      "step": 3088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017246305638644584,
+      "loss": 0.943,
+      "step": 3089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017245412741640251,
+      "loss": 1.068,
+      "step": 3090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017244519844635922,
+      "loss": 1.1139,
+      "step": 3091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017243626947631592,
+      "loss": 0.9949,
+      "step": 3092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001724273405062726,
+      "loss": 0.9347,
+      "step": 3093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001724184115362293,
+      "loss": 1.1339,
+      "step": 3094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000172409482566186,
+      "loss": 1.0094,
+      "step": 3095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017240055359614268,
+      "loss": 0.988,
+      "step": 3096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001723916246260994,
+      "loss": 1.0488,
+      "step": 3097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001723826956560561,
+      "loss": 1.1418,
+      "step": 3098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00017237376668601277,
+      "loss": 1.1178,
+      "step": 3099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017236483771596947,
+      "loss": 1.1313,
+      "step": 3100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017235590874592618,
+      "loss": 1.1597,
+      "step": 3101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017234697977588285,
+      "loss": 1.0195,
+      "step": 3102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017233805080583956,
+      "loss": 1.0368,
+      "step": 3103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017232912183579626,
+      "loss": 1.1879,
+      "step": 3104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017232019286575294,
+      "loss": 1.0857,
+      "step": 3105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017231126389570964,
+      "loss": 0.9759,
+      "step": 3106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017230233492566635,
+      "loss": 1.0868,
+      "step": 3107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017229340595562302,
+      "loss": 0.9819,
+      "step": 3108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017228447698557973,
+      "loss": 1.0965,
+      "step": 3109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017227554801553643,
+      "loss": 1.0873,
+      "step": 3110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001722666190454931,
+      "loss": 1.2627,
+      "step": 3111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017225769007544981,
+      "loss": 1.0032,
+      "step": 3112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017224876110540652,
+      "loss": 1.1457,
+      "step": 3113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001722398321353632,
+      "loss": 0.9847,
+      "step": 3114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001722309031653199,
+      "loss": 1.0451,
+      "step": 3115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001722219741952766,
+      "loss": 1.0571,
+      "step": 3116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017221304522523328,
+      "loss": 1.1298,
+      "step": 3117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017220411625518998,
+      "loss": 1.0145,
+      "step": 3118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001721951872851467,
+      "loss": 0.9685,
+      "step": 3119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017218625831510336,
+      "loss": 1.1083,
+      "step": 3120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017217732934506007,
+      "loss": 1.1937,
+      "step": 3121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017216840037501675,
+      "loss": 1.0118,
+      "step": 3122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017215947140497345,
+      "loss": 1.0078,
+      "step": 3123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017215054243493013,
+      "loss": 1.0614,
+      "step": 3124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017214161346488683,
+      "loss": 1.0106,
+      "step": 3125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017213268449484353,
+      "loss": 1.1568,
+      "step": 3126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001721237555248002,
+      "loss": 0.9111,
+      "step": 3127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017211482655475692,
+      "loss": 1.2001,
+      "step": 3128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001721058975847136,
+      "loss": 1.0034,
+      "step": 3129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001720969686146703,
+      "loss": 1.1689,
+      "step": 3130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000172088039644627,
+      "loss": 1.0996,
+      "step": 3131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017207911067458368,
+      "loss": 1.2356,
+      "step": 3132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017207018170454038,
+      "loss": 1.0109,
+      "step": 3133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017206125273449709,
+      "loss": 0.9999,
+      "step": 3134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017205232376445376,
+      "loss": 1.1305,
+      "step": 3135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017204339479441047,
+      "loss": 1.0562,
+      "step": 3136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017203446582436717,
+      "loss": 0.9073,
+      "step": 3137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017202553685432385,
+      "loss": 1.049,
+      "step": 3138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017201660788428055,
+      "loss": 0.9644,
+      "step": 3139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017200767891423726,
+      "loss": 1.1274,
+      "step": 3140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017199874994419393,
+      "loss": 1.1088,
+      "step": 3141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017198982097415064,
+      "loss": 1.0103,
+      "step": 3142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017198089200410734,
+      "loss": 1.0742,
+      "step": 3143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017197196303406402,
+      "loss": 1.0247,
+      "step": 3144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017196303406402072,
+      "loss": 1.1226,
+      "step": 3145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017195410509397743,
+      "loss": 0.9588,
+      "step": 3146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001719451761239341,
+      "loss": 1.1394,
+      "step": 3147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001719362471538908,
+      "loss": 0.9717,
+      "step": 3148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001719273181838475,
+      "loss": 1.0947,
+      "step": 3149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001719183892138042,
+      "loss": 1.0223,
+      "step": 3150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001719094602437609,
+      "loss": 1.0389,
+      "step": 3151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001719005312737176,
+      "loss": 1.2193,
+      "step": 3152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017189160230367427,
+      "loss": 0.9828,
+      "step": 3153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017188267333363098,
+      "loss": 1.1503,
+      "step": 3154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017187374436358768,
+      "loss": 1.0293,
+      "step": 3155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017186481539354436,
+      "loss": 1.0626,
+      "step": 3156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017185588642350106,
+      "loss": 1.1022,
+      "step": 3157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017184695745345777,
+      "loss": 1.0063,
+      "step": 3158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017183802848341444,
+      "loss": 0.9868,
+      "step": 3159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017182909951337115,
+      "loss": 1.1015,
+      "step": 3160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017182017054332785,
+      "loss": 1.207,
+      "step": 3161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017181124157328453,
+      "loss": 1.0684,
+      "step": 3162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017180231260324123,
+      "loss": 1.0256,
+      "step": 3163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017179338363319794,
+      "loss": 1.1179,
+      "step": 3164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001717844546631546,
+      "loss": 1.0222,
+      "step": 3165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017177552569311132,
+      "loss": 0.9926,
+      "step": 3166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017176659672306802,
+      "loss": 1.0309,
+      "step": 3167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001717576677530247,
+      "loss": 1.1839,
+      "step": 3168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001717487387829814,
+      "loss": 1.0907,
+      "step": 3169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017173980981293808,
+      "loss": 1.0552,
+      "step": 3170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017173088084289478,
+      "loss": 0.996,
+      "step": 3171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017172195187285146,
+      "loss": 1.2014,
+      "step": 3172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017171302290280816,
+      "loss": 0.9362,
+      "step": 3173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00017170409393276484,
+      "loss": 1.0311,
+      "step": 3174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017169516496272154,
+      "loss": 1.0997,
+      "step": 3175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017168623599267825,
+      "loss": 1.103,
+      "step": 3176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017167730702263493,
+      "loss": 1.0247,
+      "step": 3177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017166837805259163,
+      "loss": 1.049,
+      "step": 3178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017165944908254833,
+      "loss": 0.9538,
+      "step": 3179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000171650520112505,
+      "loss": 1.1299,
+      "step": 3180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017164159114246171,
+      "loss": 1.0087,
+      "step": 3181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017163266217241842,
+      "loss": 1.0439,
+      "step": 3182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001716237332023751,
+      "loss": 0.9861,
+      "step": 3183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001716148042323318,
+      "loss": 1.1764,
+      "step": 3184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001716058752622885,
+      "loss": 1.0578,
+      "step": 3185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001715969462922452,
+      "loss": 1.0367,
+      "step": 3186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017158801732220188,
+      "loss": 1.0203,
+      "step": 3187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001715790883521586,
+      "loss": 1.1186,
+      "step": 3188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001715701593821153,
+      "loss": 1.0793,
+      "step": 3189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017156123041207197,
+      "loss": 1.1467,
+      "step": 3190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017155230144202867,
+      "loss": 1.0003,
+      "step": 3191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017154337247198538,
+      "loss": 1.0288,
+      "step": 3192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017153444350194205,
+      "loss": 1.1151,
+      "step": 3193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017152551453189876,
+      "loss": 0.9865,
+      "step": 3194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017151658556185546,
+      "loss": 1.0018,
+      "step": 3195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017150765659181214,
+      "loss": 0.942,
+      "step": 3196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017149872762176884,
+      "loss": 1.065,
+      "step": 3197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017148979865172555,
+      "loss": 1.0713,
+      "step": 3198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017148086968168222,
+      "loss": 0.9999,
+      "step": 3199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017147194071163893,
+      "loss": 1.0774,
+      "step": 3200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017146301174159563,
+      "loss": 1.2814,
+      "step": 3201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001714540827715523,
+      "loss": 1.0685,
+      "step": 3202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000171445153801509,
+      "loss": 1.1714,
+      "step": 3203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017143622483146572,
+      "loss": 1.1393,
+      "step": 3204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001714272958614224,
+      "loss": 0.9761,
+      "step": 3205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001714183668913791,
+      "loss": 1.0527,
+      "step": 3206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001714094379213358,
+      "loss": 1.0549,
+      "step": 3207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017140050895129248,
+      "loss": 1.09,
+      "step": 3208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017139157998124918,
+      "loss": 1.0165,
+      "step": 3209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001713826510112059,
+      "loss": 1.1335,
+      "step": 3210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017137372204116256,
+      "loss": 1.1362,
+      "step": 3211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017136479307111927,
+      "loss": 1.1061,
+      "step": 3212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017135586410107595,
+      "loss": 1.0686,
+      "step": 3213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017134693513103265,
+      "loss": 1.1253,
+      "step": 3214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017133800616098933,
+      "loss": 0.9223,
+      "step": 3215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017132907719094603,
+      "loss": 1.0234,
+      "step": 3216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001713201482209027,
+      "loss": 1.085,
+      "step": 3217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001713112192508594,
+      "loss": 0.9517,
+      "step": 3218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017130229028081612,
+      "loss": 0.9992,
+      "step": 3219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001712933613107728,
+      "loss": 0.7986,
+      "step": 3220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001712844323407295,
+      "loss": 1.0906,
+      "step": 3221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001712755033706862,
+      "loss": 1.0019,
+      "step": 3222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017126657440064288,
+      "loss": 1.0744,
+      "step": 3223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017125764543059958,
+      "loss": 0.9807,
+      "step": 3224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017124871646055629,
+      "loss": 0.9544,
+      "step": 3225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017123978749051296,
+      "loss": 0.9679,
+      "step": 3226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017123085852046967,
+      "loss": 1.1417,
+      "step": 3227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017122192955042637,
+      "loss": 1.043,
+      "step": 3228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017121300058038305,
+      "loss": 0.9385,
+      "step": 3229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017120407161033975,
+      "loss": 0.971,
+      "step": 3230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017119514264029646,
+      "loss": 1.072,
+      "step": 3231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017118621367025313,
+      "loss": 0.957,
+      "step": 3232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017117728470020984,
+      "loss": 1.1134,
+      "step": 3233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017116835573016654,
+      "loss": 1.0263,
+      "step": 3234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017115942676012322,
+      "loss": 1.0233,
+      "step": 3235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017115049779007992,
+      "loss": 1.0338,
+      "step": 3236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017114156882003663,
+      "loss": 0.9405,
+      "step": 3237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001711326398499933,
+      "loss": 1.0009,
+      "step": 3238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017112371087995,
+      "loss": 0.9621,
+      "step": 3239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001711147819099067,
+      "loss": 1.0389,
+      "step": 3240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001711058529398634,
+      "loss": 1.0099,
+      "step": 3241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001710969239698201,
+      "loss": 1.003,
+      "step": 3242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001710879949997768,
+      "loss": 1.0813,
+      "step": 3243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017107906602973347,
+      "loss": 1.0828,
+      "step": 3244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017107013705969018,
+      "loss": 1.1483,
+      "step": 3245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017106120808964688,
+      "loss": 1.1131,
+      "step": 3246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017105227911960356,
+      "loss": 1.209,
+      "step": 3247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00017104335014956026,
+      "loss": 1.0246,
+      "step": 3248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017103442117951697,
+      "loss": 1.0847,
+      "step": 3249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017102549220947364,
+      "loss": 1.0005,
+      "step": 3250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017101656323943035,
+      "loss": 1.2079,
+      "step": 3251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017100763426938705,
+      "loss": 1.0107,
+      "step": 3252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017099870529934373,
+      "loss": 0.9799,
+      "step": 3253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017098977632930043,
+      "loss": 1.1027,
+      "step": 3254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017098084735925714,
+      "loss": 0.9541,
+      "step": 3255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001709719183892138,
+      "loss": 1.0979,
+      "step": 3256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017096298941917052,
+      "loss": 1.1145,
+      "step": 3257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017095406044912722,
+      "loss": 1.0232,
+      "step": 3258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001709451314790839,
+      "loss": 1.0159,
+      "step": 3259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001709362025090406,
+      "loss": 1.1085,
+      "step": 3260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017092727353899728,
+      "loss": 1.096,
+      "step": 3261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017091834456895398,
+      "loss": 1.0651,
+      "step": 3262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017090941559891066,
+      "loss": 1.0115,
+      "step": 3263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017090048662886736,
+      "loss": 1.0112,
+      "step": 3264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017089155765882404,
+      "loss": 1.0638,
+      "step": 3265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017088262868878074,
+      "loss": 1.0997,
+      "step": 3266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017087369971873745,
+      "loss": 1.0479,
+      "step": 3267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017086477074869413,
+      "loss": 1.1735,
+      "step": 3268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017085584177865083,
+      "loss": 0.9181,
+      "step": 3269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017084691280860753,
+      "loss": 1.0964,
+      "step": 3270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001708379838385642,
+      "loss": 1.124,
+      "step": 3271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017082905486852091,
+      "loss": 0.9754,
+      "step": 3272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017082012589847762,
+      "loss": 0.96,
+      "step": 3273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001708111969284343,
+      "loss": 1.1478,
+      "step": 3274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000170802267958391,
+      "loss": 0.9907,
+      "step": 3275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001707933389883477,
+      "loss": 0.966,
+      "step": 3276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017078441001830438,
+      "loss": 1.0868,
+      "step": 3277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017077548104826108,
+      "loss": 0.9333,
+      "step": 3278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001707665520782178,
+      "loss": 1.1614,
+      "step": 3279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017075762310817447,
+      "loss": 1.0225,
+      "step": 3280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017074869413813117,
+      "loss": 0.973,
+      "step": 3281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017073976516808787,
+      "loss": 1.0117,
+      "step": 3282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017073083619804458,
+      "loss": 1.0392,
+      "step": 3283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017072190722800125,
+      "loss": 1.0588,
+      "step": 3284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017071297825795796,
+      "loss": 1.024,
+      "step": 3285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017070404928791466,
+      "loss": 1.1331,
+      "step": 3286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017069512031787134,
+      "loss": 1.0738,
+      "step": 3287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017068619134782804,
+      "loss": 0.9695,
+      "step": 3288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017067726237778475,
+      "loss": 1.0531,
+      "step": 3289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017066833340774142,
+      "loss": 1.0226,
+      "step": 3290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017065940443769813,
+      "loss": 0.9882,
+      "step": 3291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017065047546765483,
+      "loss": 1.1197,
+      "step": 3292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001706415464976115,
+      "loss": 1.1047,
+      "step": 3293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001706326175275682,
+      "loss": 1.1406,
+      "step": 3294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017062368855752492,
+      "loss": 1.0494,
+      "step": 3295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001706147595874816,
+      "loss": 0.9411,
+      "step": 3296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001706058306174383,
+      "loss": 1.0441,
+      "step": 3297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000170596901647395,
+      "loss": 1.1045,
+      "step": 3298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017058797267735168,
+      "loss": 1.1179,
+      "step": 3299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017057904370730838,
+      "loss": 0.9526,
+      "step": 3300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001705701147372651,
+      "loss": 1.0717,
+      "step": 3301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017056118576722176,
+      "loss": 1.1356,
+      "step": 3302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017055225679717847,
+      "loss": 1.1234,
+      "step": 3303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017054332782713515,
+      "loss": 1.0779,
+      "step": 3304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017053439885709185,
+      "loss": 1.0133,
+      "step": 3305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017052546988704853,
+      "loss": 1.0414,
+      "step": 3306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017051654091700523,
+      "loss": 0.9558,
+      "step": 3307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001705076119469619,
+      "loss": 1.0007,
+      "step": 3308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001704986829769186,
+      "loss": 0.9424,
+      "step": 3309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017048975400687532,
+      "loss": 1.0167,
+      "step": 3310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000170480825036832,
+      "loss": 1.1742,
+      "step": 3311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001704718960667887,
+      "loss": 1.0065,
+      "step": 3312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001704629670967454,
+      "loss": 1.0462,
+      "step": 3313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017045403812670208,
+      "loss": 0.9875,
+      "step": 3314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017044510915665878,
+      "loss": 1.0265,
+      "step": 3315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017043618018661549,
+      "loss": 0.9478,
+      "step": 3316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017042725121657216,
+      "loss": 1.0891,
+      "step": 3317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017041832224652887,
+      "loss": 1.0992,
+      "step": 3318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017040939327648557,
+      "loss": 1.1086,
+      "step": 3319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017040046430644225,
+      "loss": 0.9993,
+      "step": 3320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017039153533639895,
+      "loss": 0.9594,
+      "step": 3321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017038260636635566,
+      "loss": 0.8817,
+      "step": 3322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017037367739631233,
+      "loss": 1.0239,
+      "step": 3323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017036474842626904,
+      "loss": 0.9964,
+      "step": 3324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017035581945622574,
+      "loss": 1.2029,
+      "step": 3325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017034689048618242,
+      "loss": 1.0854,
+      "step": 3326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017033796151613912,
+      "loss": 0.9836,
+      "step": 3327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017032903254609583,
+      "loss": 1.0077,
+      "step": 3328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001703201035760525,
+      "loss": 1.0633,
+      "step": 3329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001703111746060092,
+      "loss": 1.0884,
+      "step": 3330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001703022456359659,
+      "loss": 1.0175,
+      "step": 3331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001702933166659226,
+      "loss": 1.1028,
+      "step": 3332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001702843876958793,
+      "loss": 1.1261,
+      "step": 3333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000170275458725836,
+      "loss": 1.011,
+      "step": 3334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017026652975579267,
+      "loss": 0.9708,
+      "step": 3335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017025760078574938,
+      "loss": 1.0254,
+      "step": 3336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017024867181570608,
+      "loss": 0.9737,
+      "step": 3337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017023974284566276,
+      "loss": 1.0981,
+      "step": 3338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017023081387561946,
+      "loss": 1.1349,
+      "step": 3339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017022188490557617,
+      "loss": 1.0011,
+      "step": 3340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017021295593553284,
+      "loss": 0.9783,
+      "step": 3341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017020402696548955,
+      "loss": 1.1751,
+      "step": 3342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017019509799544625,
+      "loss": 1.1385,
+      "step": 3343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017018616902540293,
+      "loss": 0.9808,
+      "step": 3344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017017724005535963,
+      "loss": 1.0561,
+      "step": 3345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017016831108531633,
+      "loss": 1.0702,
+      "step": 3346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000170159382115273,
+      "loss": 1.0875,
+      "step": 3347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017015045314522972,
+      "loss": 1.3279,
+      "step": 3348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001701415241751864,
+      "loss": 0.9611,
+      "step": 3349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001701325952051431,
+      "loss": 1.0635,
+      "step": 3350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001701236662350998,
+      "loss": 1.0811,
+      "step": 3351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017011473726505648,
+      "loss": 1.0845,
+      "step": 3352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017010580829501318,
+      "loss": 1.015,
+      "step": 3353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017009687932496986,
+      "loss": 1.019,
+      "step": 3354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017008795035492656,
+      "loss": 1.0864,
+      "step": 3355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017007902138488324,
+      "loss": 1.1541,
+      "step": 3356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017007009241483994,
+      "loss": 1.19,
+      "step": 3357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017006116344479665,
+      "loss": 1.0355,
+      "step": 3358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017005223447475332,
+      "loss": 1.0621,
+      "step": 3359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017004330550471003,
+      "loss": 1.0349,
+      "step": 3360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017003437653466673,
+      "loss": 0.9799,
+      "step": 3361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001700254475646234,
+      "loss": 1.15,
+      "step": 3362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017001651859458011,
+      "loss": 1.0754,
+      "step": 3363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017000758962453682,
+      "loss": 1.0974,
+      "step": 3364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001699986606544935,
+      "loss": 0.9313,
+      "step": 3365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001699897316844502,
+      "loss": 1.0748,
+      "step": 3366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001699808027144069,
+      "loss": 0.9416,
+      "step": 3367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016997187374436358,
+      "loss": 1.0346,
+      "step": 3368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016996294477432028,
+      "loss": 0.996,
+      "step": 3369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000169954015804277,
+      "loss": 1.0955,
+      "step": 3370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016994508683423366,
+      "loss": 1.0731,
+      "step": 3371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016993615786419037,
+      "loss": 1.168,
+      "step": 3372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016992722889414707,
+      "loss": 1.0119,
+      "step": 3373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016991829992410375,
+      "loss": 1.0251,
+      "step": 3374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016990937095406045,
+      "loss": 1.0265,
+      "step": 3375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016990044198401716,
+      "loss": 1.0016,
+      "step": 3376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016989151301397383,
+      "loss": 1.0269,
+      "step": 3377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016988258404393054,
+      "loss": 1.0028,
+      "step": 3378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016987365507388724,
+      "loss": 1.0109,
+      "step": 3379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016986472610384395,
+      "loss": 1.0365,
+      "step": 3380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016985579713380062,
+      "loss": 1.0037,
+      "step": 3381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016984686816375733,
+      "loss": 1.021,
+      "step": 3382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016983793919371403,
+      "loss": 1.0972,
+      "step": 3383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001698290102236707,
+      "loss": 0.9631,
+      "step": 3384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001698200812536274,
+      "loss": 0.9994,
+      "step": 3385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016981115228358412,
+      "loss": 0.9814,
+      "step": 3386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001698022233135408,
+      "loss": 0.9961,
+      "step": 3387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001697932943434975,
+      "loss": 1.0609,
+      "step": 3388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001697843653734542,
+      "loss": 1.1687,
+      "step": 3389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016977543640341088,
+      "loss": 0.9725,
+      "step": 3390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016976650743336758,
+      "loss": 1.0456,
+      "step": 3391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001697575784633243,
+      "loss": 1.1083,
+      "step": 3392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016974864949328096,
+      "loss": 1.052,
+      "step": 3393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016973972052323767,
+      "loss": 1.0959,
+      "step": 3394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016973079155319434,
+      "loss": 1.1161,
+      "step": 3395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016972186258315105,
+      "loss": 1.1065,
+      "step": 3396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016971293361310773,
+      "loss": 1.092,
+      "step": 3397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00016970400464306443,
+      "loss": 1.0431,
+      "step": 3398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001696950756730211,
+      "loss": 1.0068,
+      "step": 3399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001696861467029778,
+      "loss": 1.0736,
+      "step": 3400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016967721773293451,
+      "loss": 1.1343,
+      "step": 3401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001696682887628912,
+      "loss": 1.025,
+      "step": 3402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001696593597928479,
+      "loss": 1.0364,
+      "step": 3403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001696504308228046,
+      "loss": 0.9591,
+      "step": 3404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016964150185276128,
+      "loss": 1.0304,
+      "step": 3405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016963257288271798,
+      "loss": 1.2443,
+      "step": 3406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016962364391267468,
+      "loss": 1.1521,
+      "step": 3407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016961471494263136,
+      "loss": 0.9932,
+      "step": 3408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016960578597258807,
+      "loss": 1.091,
+      "step": 3409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016959685700254477,
+      "loss": 1.0639,
+      "step": 3410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016958792803250145,
+      "loss": 1.0972,
+      "step": 3411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016957899906245815,
+      "loss": 0.9665,
+      "step": 3412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016957007009241485,
+      "loss": 1.1446,
+      "step": 3413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016956114112237153,
+      "loss": 1.0363,
+      "step": 3414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016955221215232824,
+      "loss": 0.9717,
+      "step": 3415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016954328318228494,
+      "loss": 1.1838,
+      "step": 3416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016953435421224162,
+      "loss": 1.096,
+      "step": 3417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016952542524219832,
+      "loss": 0.9835,
+      "step": 3418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016951649627215502,
+      "loss": 1.023,
+      "step": 3419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001695075673021117,
+      "loss": 1.0874,
+      "step": 3420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001694986383320684,
+      "loss": 1.0926,
+      "step": 3421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001694897093620251,
+      "loss": 1.0531,
+      "step": 3422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001694807803919818,
+      "loss": 0.9432,
+      "step": 3423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001694718514219385,
+      "loss": 1.0176,
+      "step": 3424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001694629224518952,
+      "loss": 1.0069,
+      "step": 3425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016945399348185187,
+      "loss": 0.8907,
+      "step": 3426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016944506451180858,
+      "loss": 1.063,
+      "step": 3427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016943613554176528,
+      "loss": 1.0646,
+      "step": 3428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016942720657172196,
+      "loss": 1.0087,
+      "step": 3429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016941827760167866,
+      "loss": 1.1247,
+      "step": 3430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016940934863163536,
+      "loss": 1.0462,
+      "step": 3431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016940041966159204,
+      "loss": 1.0722,
+      "step": 3432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016939149069154875,
+      "loss": 0.9713,
+      "step": 3433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016938256172150545,
+      "loss": 1.0904,
+      "step": 3434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016937363275146213,
+      "loss": 1.0302,
+      "step": 3435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016936470378141883,
+      "loss": 1.0529,
+      "step": 3436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016935577481137553,
+      "loss": 1.026,
+      "step": 3437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001693468458413322,
+      "loss": 1.0198,
+      "step": 3438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016933791687128892,
+      "loss": 0.9757,
+      "step": 3439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001693289879012456,
+      "loss": 1.1581,
+      "step": 3440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001693200589312023,
+      "loss": 1.0593,
+      "step": 3441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000169311129961159,
+      "loss": 0.9682,
+      "step": 3442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016930220099111568,
+      "loss": 1.0901,
+      "step": 3443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016929327202107238,
+      "loss": 1.1812,
+      "step": 3444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016928434305102906,
+      "loss": 0.9882,
+      "step": 3445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016927541408098576,
+      "loss": 0.9875,
+      "step": 3446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016926648511094244,
+      "loss": 0.8161,
+      "step": 3447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016925755614089914,
+      "loss": 1.1294,
+      "step": 3448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016924862717085585,
+      "loss": 1.0196,
+      "step": 3449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016923969820081252,
+      "loss": 1.0743,
+      "step": 3450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 1.1899,
+      "step": 3451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016922184026072593,
+      "loss": 1.1499,
+      "step": 3452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001692129112906826,
+      "loss": 1.0981,
+      "step": 3453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001692039823206393,
+      "loss": 0.9522,
+      "step": 3454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016919505335059602,
+      "loss": 0.9608,
+      "step": 3455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001691861243805527,
+      "loss": 1.2046,
+      "step": 3456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001691771954105094,
+      "loss": 1.0201,
+      "step": 3457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001691682664404661,
+      "loss": 0.9765,
+      "step": 3458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016915933747042278,
+      "loss": 1.0443,
+      "step": 3459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016915040850037948,
+      "loss": 0.9862,
+      "step": 3460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001691414795303362,
+      "loss": 1.0205,
+      "step": 3461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016913255056029286,
+      "loss": 1.1152,
+      "step": 3462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016912362159024957,
+      "loss": 1.017,
+      "step": 3463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016911469262020627,
+      "loss": 0.9771,
+      "step": 3464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016910576365016295,
+      "loss": 1.0973,
+      "step": 3465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016909683468011965,
+      "loss": 1.0615,
+      "step": 3466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016908790571007636,
+      "loss": 1.0734,
+      "step": 3467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016907897674003303,
+      "loss": 0.9741,
+      "step": 3468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016907004776998974,
+      "loss": 0.9865,
+      "step": 3469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016906111879994644,
+      "loss": 1.0737,
+      "step": 3470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016905218982990312,
+      "loss": 1.2495,
+      "step": 3471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00016904326085985982,
+      "loss": 1.0569,
+      "step": 3472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016903433188981653,
+      "loss": 1.0004,
+      "step": 3473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001690254029197732,
+      "loss": 1.1315,
+      "step": 3474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001690164739497299,
+      "loss": 0.9572,
+      "step": 3475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001690075449796866,
+      "loss": 0.8684,
+      "step": 3476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001689986160096433,
+      "loss": 1.1878,
+      "step": 3477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001689896870396,
+      "loss": 1.1957,
+      "step": 3478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001689807580695567,
+      "loss": 0.9144,
+      "step": 3479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001689718290995134,
+      "loss": 1.1072,
+      "step": 3480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016896290012947008,
+      "loss": 1.1141,
+      "step": 3481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016895397115942678,
+      "loss": 1.0174,
+      "step": 3482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001689450421893835,
+      "loss": 1.0513,
+      "step": 3483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016893611321934016,
+      "loss": 1.0681,
+      "step": 3484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016892718424929687,
+      "loss": 1.1064,
+      "step": 3485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016891825527925354,
+      "loss": 1.1195,
+      "step": 3486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016890932630921025,
+      "loss": 1.0469,
+      "step": 3487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016890039733916693,
+      "loss": 1.1033,
+      "step": 3488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016889146836912363,
+      "loss": 1.03,
+      "step": 3489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001688825393990803,
+      "loss": 0.9466,
+      "step": 3490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000168873610429037,
+      "loss": 1.0556,
+      "step": 3491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016886468145899371,
+      "loss": 1.0667,
+      "step": 3492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001688557524889504,
+      "loss": 0.9712,
+      "step": 3493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001688468235189071,
+      "loss": 1.0136,
+      "step": 3494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001688378945488638,
+      "loss": 0.9829,
+      "step": 3495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016882896557882048,
+      "loss": 0.9975,
+      "step": 3496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016882003660877718,
+      "loss": 1.1319,
+      "step": 3497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016881110763873388,
+      "loss": 0.9672,
+      "step": 3498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016880217866869056,
+      "loss": 1.0769,
+      "step": 3499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016879324969864727,
+      "loss": 1.0664,
+      "step": 3500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016878432072860397,
+      "loss": 0.9866,
+      "step": 3501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016877539175856065,
+      "loss": 1.0073,
+      "step": 3502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016876646278851735,
+      "loss": 0.9503,
+      "step": 3503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016875753381847405,
+      "loss": 0.8411,
+      "step": 3504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016874860484843073,
+      "loss": 1.0565,
+      "step": 3505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016873967587838744,
+      "loss": 1.1172,
+      "step": 3506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016873074690834414,
+      "loss": 1.0727,
+      "step": 3507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016872181793830082,
+      "loss": 1.0941,
+      "step": 3508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016871288896825752,
+      "loss": 1.0811,
+      "step": 3509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016870395999821422,
+      "loss": 0.9511,
+      "step": 3510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001686950310281709,
+      "loss": 1.0291,
+      "step": 3511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001686861020581276,
+      "loss": 1.0357,
+      "step": 3512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001686771730880843,
+      "loss": 1.0909,
+      "step": 3513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016866824411804099,
+      "loss": 0.9878,
+      "step": 3514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001686593151479977,
+      "loss": 0.9956,
+      "step": 3515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001686503861779544,
+      "loss": 1.0964,
+      "step": 3516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016864145720791107,
+      "loss": 1.0908,
+      "step": 3517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016863252823786778,
+      "loss": 1.0601,
+      "step": 3518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016862359926782448,
+      "loss": 1.0037,
+      "step": 3519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016861467029778116,
+      "loss": 1.226,
+      "step": 3520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016860574132773786,
+      "loss": 1.0045,
+      "step": 3521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016859681235769456,
+      "loss": 1.0595,
+      "step": 3522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016858788338765124,
+      "loss": 1.1035,
+      "step": 3523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016857895441760795,
+      "loss": 0.9831,
+      "step": 3524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016857002544756465,
+      "loss": 0.9765,
+      "step": 3525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016856109647752133,
+      "loss": 1.0845,
+      "step": 3526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016855216750747803,
+      "loss": 1.1154,
+      "step": 3527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016854323853743473,
+      "loss": 1.0756,
+      "step": 3528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001685343095673914,
+      "loss": 0.9323,
+      "step": 3529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016852538059734812,
+      "loss": 1.0176,
+      "step": 3530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001685164516273048,
+      "loss": 1.1573,
+      "step": 3531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001685075226572615,
+      "loss": 0.963,
+      "step": 3532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001684985936872182,
+      "loss": 1.0826,
+      "step": 3533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016848966471717488,
+      "loss": 1.0927,
+      "step": 3534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016848073574713158,
+      "loss": 1.19,
+      "step": 3535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016847180677708826,
+      "loss": 0.9495,
+      "step": 3536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016846287780704496,
+      "loss": 0.9282,
+      "step": 3537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016845394883700164,
+      "loss": 1.0348,
+      "step": 3538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016844501986695834,
+      "loss": 1.2112,
+      "step": 3539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016843609089691505,
+      "loss": 1.2512,
+      "step": 3540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016842716192687172,
+      "loss": 1.1738,
+      "step": 3541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016841823295682843,
+      "loss": 0.9896,
+      "step": 3542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016840930398678513,
+      "loss": 1.058,
+      "step": 3543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001684003750167418,
+      "loss": 1.1671,
+      "step": 3544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001683914460466985,
+      "loss": 1.0282,
+      "step": 3545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00016838251707665522,
+      "loss": 1.0799,
+      "step": 3546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001683735881066119,
+      "loss": 0.9615,
+      "step": 3547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001683646591365686,
+      "loss": 1.1112,
+      "step": 3548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001683557301665253,
+      "loss": 0.9699,
+      "step": 3549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016834680119648198,
+      "loss": 1.0624,
+      "step": 3550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016833787222643868,
+      "loss": 0.9534,
+      "step": 3551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001683289432563954,
+      "loss": 0.8899,
+      "step": 3552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016832001428635206,
+      "loss": 0.9419,
+      "step": 3553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016831108531630877,
+      "loss": 1.0251,
+      "step": 3554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016830215634626547,
+      "loss": 1.05,
+      "step": 3555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016829322737622215,
+      "loss": 1.0161,
+      "step": 3556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016828429840617885,
+      "loss": 1.0251,
+      "step": 3557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016827536943613556,
+      "loss": 1.0617,
+      "step": 3558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016826644046609223,
+      "loss": 1.1365,
+      "step": 3559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016825751149604894,
+      "loss": 1.1234,
+      "step": 3560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016824858252600564,
+      "loss": 0.9042,
+      "step": 3561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016823965355596232,
+      "loss": 1.0923,
+      "step": 3562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016823072458591902,
+      "loss": 1.0072,
+      "step": 3563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016822179561587573,
+      "loss": 1.0142,
+      "step": 3564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001682128666458324,
+      "loss": 1.0131,
+      "step": 3565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001682039376757891,
+      "loss": 1.024,
+      "step": 3566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001681950087057458,
+      "loss": 1.0209,
+      "step": 3567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001681860797357025,
+      "loss": 1.173,
+      "step": 3568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001681771507656592,
+      "loss": 1.0775,
+      "step": 3569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001681682217956159,
+      "loss": 0.9975,
+      "step": 3570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016815929282557257,
+      "loss": 1.2388,
+      "step": 3571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016815036385552928,
+      "loss": 1.0291,
+      "step": 3572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016814143488548598,
+      "loss": 1.1302,
+      "step": 3573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016813250591544266,
+      "loss": 0.9354,
+      "step": 3574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016812357694539936,
+      "loss": 1.084,
+      "step": 3575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016811464797535607,
+      "loss": 1.0522,
+      "step": 3576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016810571900531274,
+      "loss": 1.1774,
+      "step": 3577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016809679003526945,
+      "loss": 1.0385,
+      "step": 3578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016808786106522612,
+      "loss": 0.9808,
+      "step": 3579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016807893209518283,
+      "loss": 1.0368,
+      "step": 3580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001680700031251395,
+      "loss": 0.9937,
+      "step": 3581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001680610741550962,
+      "loss": 1.112,
+      "step": 3582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016805214518505291,
+      "loss": 1.2799,
+      "step": 3583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001680432162150096,
+      "loss": 1.0182,
+      "step": 3584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001680342872449663,
+      "loss": 1.0031,
+      "step": 3585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000168025358274923,
+      "loss": 1.0862,
+      "step": 3586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016801642930487968,
+      "loss": 0.933,
+      "step": 3587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016800750033483638,
+      "loss": 1.0647,
+      "step": 3588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016799857136479308,
+      "loss": 1.0226,
+      "step": 3589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016798964239474976,
+      "loss": 1.0644,
+      "step": 3590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016798071342470646,
+      "loss": 1.0114,
+      "step": 3591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016797178445466317,
+      "loss": 0.9763,
+      "step": 3592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016796285548461985,
+      "loss": 0.9978,
+      "step": 3593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016795392651457655,
+      "loss": 1.1213,
+      "step": 3594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016794499754453325,
+      "loss": 1.0077,
+      "step": 3595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016793606857448993,
+      "loss": 1.0422,
+      "step": 3596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016792713960444663,
+      "loss": 1.0168,
+      "step": 3597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016791821063440334,
+      "loss": 1.1992,
+      "step": 3598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016790928166436002,
+      "loss": 1.0435,
+      "step": 3599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016790035269431672,
+      "loss": 1.0765,
+      "step": 3600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016789142372427342,
+      "loss": 1.0439,
+      "step": 3601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001678824947542301,
+      "loss": 1.144,
+      "step": 3602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001678735657841868,
+      "loss": 1.0565,
+      "step": 3603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001678646368141435,
+      "loss": 0.9926,
+      "step": 3604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016785570784410019,
+      "loss": 1.0815,
+      "step": 3605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001678467788740569,
+      "loss": 1.0546,
+      "step": 3606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001678378499040136,
+      "loss": 0.9497,
+      "step": 3607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016782892093397027,
+      "loss": 0.9902,
+      "step": 3608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016781999196392697,
+      "loss": 1.0367,
+      "step": 3609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016781106299388368,
+      "loss": 0.9372,
+      "step": 3610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016780213402384036,
+      "loss": 1.0103,
+      "step": 3611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016779320505379706,
+      "loss": 1.1154,
+      "step": 3612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016778427608375376,
+      "loss": 1.0386,
+      "step": 3613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016777534711371044,
+      "loss": 0.9915,
+      "step": 3614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016776641814366714,
+      "loss": 1.0166,
+      "step": 3615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016775748917362385,
+      "loss": 1.1336,
+      "step": 3616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016774856020358053,
+      "loss": 0.9545,
+      "step": 3617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016773963123353723,
+      "loss": 1.0665,
+      "step": 3618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016773070226349393,
+      "loss": 1.0178,
+      "step": 3619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001677217732934506,
+      "loss": 0.8941,
+      "step": 3620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016771284432340731,
+      "loss": 0.9181,
+      "step": 3621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000167703915353364,
+      "loss": 1.1636,
+      "step": 3622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001676949863833207,
+      "loss": 1.0657,
+      "step": 3623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001676860574132774,
+      "loss": 1.1009,
+      "step": 3624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016767712844323408,
+      "loss": 1.1302,
+      "step": 3625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016766819947319078,
+      "loss": 0.9992,
+      "step": 3626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016765927050314746,
+      "loss": 1.1087,
+      "step": 3627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016765034153310416,
+      "loss": 1.071,
+      "step": 3628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016764141256306084,
+      "loss": 0.9258,
+      "step": 3629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016763248359301754,
+      "loss": 1.0323,
+      "step": 3630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016762355462297425,
+      "loss": 1.1182,
+      "step": 3631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016761462565293092,
+      "loss": 1.0603,
+      "step": 3632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016760569668288763,
+      "loss": 1.0259,
+      "step": 3633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016759676771284433,
+      "loss": 0.9529,
+      "step": 3634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000167587838742801,
+      "loss": 0.9944,
+      "step": 3635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001675789097727577,
+      "loss": 1.0199,
+      "step": 3636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016756998080271442,
+      "loss": 1.0235,
+      "step": 3637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001675610518326711,
+      "loss": 0.9186,
+      "step": 3638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001675521228626278,
+      "loss": 0.9723,
+      "step": 3639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001675431938925845,
+      "loss": 0.9667,
+      "step": 3640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016753426492254118,
+      "loss": 0.9805,
+      "step": 3641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016752533595249788,
+      "loss": 1.0964,
+      "step": 3642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001675164069824546,
+      "loss": 1.0722,
+      "step": 3643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016750747801241126,
+      "loss": 1.0864,
+      "step": 3644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016749854904236797,
+      "loss": 0.9748,
+      "step": 3645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016748962007232467,
+      "loss": 1.1237,
+      "step": 3646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016748069110228135,
+      "loss": 0.989,
+      "step": 3647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016747176213223805,
+      "loss": 1.1927,
+      "step": 3648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016746283316219476,
+      "loss": 0.9285,
+      "step": 3649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016745390419215143,
+      "loss": 1.0148,
+      "step": 3650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016744497522210814,
+      "loss": 1.0315,
+      "step": 3651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016743604625206484,
+      "loss": 0.9639,
+      "step": 3652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016742711728202152,
+      "loss": 0.9944,
+      "step": 3653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016741818831197822,
+      "loss": 1.122,
+      "step": 3654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016740925934193493,
+      "loss": 1.0335,
+      "step": 3655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001674003303718916,
+      "loss": 1.0674,
+      "step": 3656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001673914014018483,
+      "loss": 1.0089,
+      "step": 3657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000167382472431805,
+      "loss": 1.0523,
+      "step": 3658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001673735434617617,
+      "loss": 1.0743,
+      "step": 3659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001673646144917184,
+      "loss": 1.1194,
+      "step": 3660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001673556855216751,
+      "loss": 1.172,
+      "step": 3661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016734675655163177,
+      "loss": 1.2608,
+      "step": 3662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016733782758158848,
+      "loss": 1.084,
+      "step": 3663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016732889861154518,
+      "loss": 1.1092,
+      "step": 3664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016731996964150186,
+      "loss": 0.9622,
+      "step": 3665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016731104067145856,
+      "loss": 1.0646,
+      "step": 3666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016730211170141527,
+      "loss": 1.0886,
+      "step": 3667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016729318273137194,
+      "loss": 0.9617,
+      "step": 3668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016728425376132865,
+      "loss": 1.0963,
+      "step": 3669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016727532479128532,
+      "loss": 1.0932,
+      "step": 3670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016726639582124203,
+      "loss": 0.9175,
+      "step": 3671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001672574668511987,
+      "loss": 1.0485,
+      "step": 3672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001672485378811554,
+      "loss": 1.183,
+      "step": 3673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016723960891111211,
+      "loss": 1.1099,
+      "step": 3674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001672306799410688,
+      "loss": 0.9567,
+      "step": 3675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001672217509710255,
+      "loss": 1.0401,
+      "step": 3676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001672128220009822,
+      "loss": 1.0655,
+      "step": 3677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016720389303093888,
+      "loss": 1.0807,
+      "step": 3678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016719496406089558,
+      "loss": 1.0165,
+      "step": 3679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016718603509085228,
+      "loss": 1.2764,
+      "step": 3680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016717710612080896,
+      "loss": 1.0759,
+      "step": 3681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016716817715076566,
+      "loss": 1.0752,
+      "step": 3682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016715924818072237,
+      "loss": 1.1525,
+      "step": 3683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016715031921067905,
+      "loss": 1.0475,
+      "step": 3684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016714139024063575,
+      "loss": 1.1258,
+      "step": 3685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016713246127059245,
+      "loss": 1.0731,
+      "step": 3686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016712353230054913,
+      "loss": 1.0193,
+      "step": 3687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016711460333050583,
+      "loss": 1.1808,
+      "step": 3688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016710567436046254,
+      "loss": 1.0085,
+      "step": 3689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016709674539041922,
+      "loss": 1.168,
+      "step": 3690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016708781642037592,
+      "loss": 1.0917,
+      "step": 3691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016707888745033262,
+      "loss": 0.9409,
+      "step": 3692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001670699584802893,
+      "loss": 1.3446,
+      "step": 3693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000167061029510246,
+      "loss": 1.0615,
+      "step": 3694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001670521005402027,
+      "loss": 1.0588,
+      "step": 3695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016704317157015939,
+      "loss": 0.9709,
+      "step": 3696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001670342426001161,
+      "loss": 0.9643,
+      "step": 3697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001670253136300728,
+      "loss": 1.2189,
+      "step": 3698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016701638466002947,
+      "loss": 0.9259,
+      "step": 3699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016700745568998617,
+      "loss": 1.1099,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016699852671994288,
+      "loss": 1.0017,
+      "step": 3701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016698959774989956,
+      "loss": 1.1041,
+      "step": 3702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016698066877985626,
+      "loss": 1.0663,
+      "step": 3703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016697173980981296,
+      "loss": 0.9121,
+      "step": 3704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016696281083976964,
+      "loss": 1.0125,
+      "step": 3705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016695388186972634,
+      "loss": 1.0475,
+      "step": 3706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016694495289968305,
+      "loss": 1.0618,
+      "step": 3707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016693602392963973,
+      "loss": 1.0521,
+      "step": 3708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016692709495959643,
+      "loss": 1.0859,
+      "step": 3709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016691816598955313,
+      "loss": 1.0155,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001669092370195098,
+      "loss": 1.0249,
+      "step": 3711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016690030804946651,
+      "loss": 1.2193,
+      "step": 3712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001668913790794232,
+      "loss": 0.9827,
+      "step": 3713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001668824501093799,
+      "loss": 0.9979,
+      "step": 3714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001668735211393366,
+      "loss": 1.0993,
+      "step": 3715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016686459216929328,
+      "loss": 1.1223,
+      "step": 3716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016685566319924998,
+      "loss": 0.9586,
+      "step": 3717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016684673422920666,
+      "loss": 1.0065,
+      "step": 3718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016683780525916336,
+      "loss": 0.9546,
+      "step": 3719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016682887628912004,
+      "loss": 0.9784,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016681994731907674,
+      "loss": 1.1083,
+      "step": 3721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016681101834903345,
+      "loss": 1.0045,
+      "step": 3722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016680208937899012,
+      "loss": 1.0898,
+      "step": 3723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016679316040894683,
+      "loss": 1.1014,
+      "step": 3724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016678423143890353,
+      "loss": 0.9853,
+      "step": 3725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001667753024688602,
+      "loss": 1.1029,
+      "step": 3726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001667663734988169,
+      "loss": 1.0253,
+      "step": 3727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016675744452877362,
+      "loss": 0.9751,
+      "step": 3728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001667485155587303,
+      "loss": 1.1001,
+      "step": 3729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000166739586588687,
+      "loss": 0.9943,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001667306576186437,
+      "loss": 1.1363,
+      "step": 3731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016672172864860038,
+      "loss": 1.0011,
+      "step": 3732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016671279967855708,
+      "loss": 1.0125,
+      "step": 3733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016670387070851379,
+      "loss": 1.0223,
+      "step": 3734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016669494173847046,
+      "loss": 1.1025,
+      "step": 3735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016668601276842717,
+      "loss": 1.1874,
+      "step": 3736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016667708379838387,
+      "loss": 0.9813,
+      "step": 3737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016666815482834055,
+      "loss": 1.0527,
+      "step": 3738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016665922585829725,
+      "loss": 1.1129,
+      "step": 3739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016665029688825396,
+      "loss": 1.0193,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016664136791821063,
+      "loss": 1.0723,
+      "step": 3741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016663243894816734,
+      "loss": 0.8935,
+      "step": 3742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016662350997812404,
+      "loss": 1.2478,
+      "step": 3743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016661458100808072,
+      "loss": 0.9071,
+      "step": 3744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016660565203803742,
+      "loss": 0.9973,
+      "step": 3745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016659672306799413,
+      "loss": 1.0107,
+      "step": 3746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001665877940979508,
+      "loss": 1.0506,
+      "step": 3747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001665788651279075,
+      "loss": 0.9824,
+      "step": 3748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001665699361578642,
+      "loss": 1.1208,
+      "step": 3749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001665610071878209,
+      "loss": 1.0166,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001665520782177776,
+      "loss": 1.0365,
+      "step": 3751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001665431492477343,
+      "loss": 1.1448,
+      "step": 3752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016653422027769097,
+      "loss": 1.0315,
+      "step": 3753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016652529130764768,
+      "loss": 1.0337,
+      "step": 3754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016651636233760438,
+      "loss": 0.9589,
+      "step": 3755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016650743336756106,
+      "loss": 1.0061,
+      "step": 3756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016649850439751776,
+      "loss": 1.0518,
+      "step": 3757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016648957542747447,
+      "loss": 0.9965,
+      "step": 3758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016648064645743114,
+      "loss": 0.9544,
+      "step": 3759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016647171748738785,
+      "loss": 1.0309,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016646278851734452,
+      "loss": 0.9803,
+      "step": 3761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016645385954730123,
+      "loss": 1.1473,
+      "step": 3762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001664449305772579,
+      "loss": 1.1232,
+      "step": 3763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001664360016072146,
+      "loss": 1.0895,
+      "step": 3764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016642707263717129,
+      "loss": 1.1015,
+      "step": 3765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000166418143667128,
+      "loss": 0.9641,
+      "step": 3766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001664092146970847,
+      "loss": 0.9863,
+      "step": 3767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016640028572704137,
+      "loss": 1.0084,
+      "step": 3768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016639135675699808,
+      "loss": 1.0176,
+      "step": 3769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016638242778695478,
+      "loss": 1.0336,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016637349881691148,
+      "loss": 1.0113,
+      "step": 3771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016636456984686816,
+      "loss": 1.0603,
+      "step": 3772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016635564087682486,
+      "loss": 1.0814,
+      "step": 3773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016634671190678157,
+      "loss": 1.2822,
+      "step": 3774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016633778293673825,
+      "loss": 1.0776,
+      "step": 3775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016632885396669495,
+      "loss": 0.9316,
+      "step": 3776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016631992499665165,
+      "loss": 1.0244,
+      "step": 3777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016631099602660833,
+      "loss": 1.0002,
+      "step": 3778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016630206705656503,
+      "loss": 0.9767,
+      "step": 3779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016629313808652174,
+      "loss": 1.0009,
+      "step": 3780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016628420911647842,
+      "loss": 0.9206,
+      "step": 3781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016627528014643512,
+      "loss": 1.1206,
+      "step": 3782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016626635117639182,
+      "loss": 1.0494,
+      "step": 3783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001662574222063485,
+      "loss": 1.1224,
+      "step": 3784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001662484932363052,
+      "loss": 1.0673,
+      "step": 3785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001662395642662619,
+      "loss": 1.068,
+      "step": 3786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016623063529621859,
+      "loss": 1.0833,
+      "step": 3787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001662217063261753,
+      "loss": 1.0499,
+      "step": 3788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000166212777356132,
+      "loss": 0.9137,
+      "step": 3789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016620384838608867,
+      "loss": 1.021,
+      "step": 3790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016619491941604537,
+      "loss": 1.0399,
+      "step": 3791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016618599044600208,
+      "loss": 1.0868,
+      "step": 3792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016617706147595876,
+      "loss": 1.1471,
+      "step": 3793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016616813250591546,
+      "loss": 1.0422,
+      "step": 3794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016615920353587216,
+      "loss": 1.1011,
+      "step": 3795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016615027456582884,
+      "loss": 1.0326,
+      "step": 3796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016614134559578554,
+      "loss": 0.9538,
+      "step": 3797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016613241662574225,
+      "loss": 1.2007,
+      "step": 3798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016612348765569893,
+      "loss": 1.0132,
+      "step": 3799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016611455868565563,
+      "loss": 1.1344,
+      "step": 3800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016610562971561233,
+      "loss": 1.0702,
+      "step": 3801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000166096700745569,
+      "loss": 1.0711,
+      "step": 3802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016608777177552571,
+      "loss": 1.0571,
+      "step": 3803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001660788428054824,
+      "loss": 1.0078,
+      "step": 3804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001660699138354391,
+      "loss": 1.0062,
+      "step": 3805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001660609848653958,
+      "loss": 0.991,
+      "step": 3806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016605205589535248,
+      "loss": 1.1291,
+      "step": 3807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016604312692530918,
+      "loss": 1.0574,
+      "step": 3808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016603419795526586,
+      "loss": 1.1282,
+      "step": 3809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016602526898522256,
+      "loss": 1.0626,
+      "step": 3810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016601634001517924,
+      "loss": 0.9605,
+      "step": 3811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016600741104513594,
+      "loss": 1.0059,
+      "step": 3812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016599848207509265,
+      "loss": 0.9853,
+      "step": 3813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016598955310504932,
+      "loss": 1.0248,
+      "step": 3814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016598062413500603,
+      "loss": 0.8559,
+      "step": 3815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016597169516496273,
+      "loss": 1.2313,
+      "step": 3816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001659627661949194,
+      "loss": 1.0468,
+      "step": 3817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001659538372248761,
+      "loss": 0.9895,
+      "step": 3818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016594490825483282,
+      "loss": 1.0039,
+      "step": 3819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001659359792847895,
+      "loss": 1.1664,
+      "step": 3820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001659270503147462,
+      "loss": 1.0883,
+      "step": 3821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001659181213447029,
+      "loss": 1.0534,
+      "step": 3822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016590919237465958,
+      "loss": 1.1368,
+      "step": 3823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016590026340461628,
+      "loss": 1.0392,
+      "step": 3824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016589133443457299,
+      "loss": 1.0972,
+      "step": 3825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016588240546452966,
+      "loss": 0.8755,
+      "step": 3826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016587347649448637,
+      "loss": 1.1099,
+      "step": 3827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016586454752444307,
+      "loss": 1.1188,
+      "step": 3828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016585561855439975,
+      "loss": 0.9238,
+      "step": 3829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016584668958435645,
+      "loss": 0.9627,
+      "step": 3830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016583776061431316,
+      "loss": 1.0257,
+      "step": 3831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016582883164426983,
+      "loss": 1.0938,
+      "step": 3832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016581990267422654,
+      "loss": 1.0291,
+      "step": 3833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016581097370418324,
+      "loss": 1.0897,
+      "step": 3834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016580204473413992,
+      "loss": 1.0509,
+      "step": 3835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016579311576409662,
+      "loss": 1.1732,
+      "step": 3836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016578418679405333,
+      "loss": 1.1168,
+      "step": 3837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016577525782401,
+      "loss": 1.0123,
+      "step": 3838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001657663288539667,
+      "loss": 0.9511,
+      "step": 3839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001657573998839234,
+      "loss": 1.0114,
+      "step": 3840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001657484709138801,
+      "loss": 1.0121,
+      "step": 3841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001657395419438368,
+      "loss": 1.0925,
+      "step": 3842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001657306129737935,
+      "loss": 1.0992,
+      "step": 3843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016572168400375017,
+      "loss": 1.0271,
+      "step": 3844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016571275503370688,
+      "loss": 1.0075,
+      "step": 3845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016570382606366358,
+      "loss": 0.8941,
+      "step": 3846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016569489709362026,
+      "loss": 0.9768,
+      "step": 3847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016568596812357696,
+      "loss": 0.9853,
+      "step": 3848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016567703915353367,
+      "loss": 1.1884,
+      "step": 3849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016566811018349034,
+      "loss": 1.1076,
+      "step": 3850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016565918121344705,
+      "loss": 1.0591,
+      "step": 3851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016565025224340372,
+      "loss": 1.0432,
+      "step": 3852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016564132327336043,
+      "loss": 1.0636,
+      "step": 3853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001656323943033171,
+      "loss": 1.209,
+      "step": 3854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001656234653332738,
+      "loss": 1.0323,
+      "step": 3855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016561453636323049,
+      "loss": 1.1564,
+      "step": 3856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001656056073931872,
+      "loss": 1.1687,
+      "step": 3857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001655966784231439,
+      "loss": 0.9273,
+      "step": 3858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016558774945310057,
+      "loss": 0.9252,
+      "step": 3859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016557882048305727,
+      "loss": 1.0668,
+      "step": 3860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016556989151301398,
+      "loss": 1.1136,
+      "step": 3861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016556096254297066,
+      "loss": 0.985,
+      "step": 3862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016555203357292736,
+      "loss": 1.0926,
+      "step": 3863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016554310460288406,
+      "loss": 1.3159,
+      "step": 3864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016553417563284074,
+      "loss": 1.2621,
+      "step": 3865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016552524666279744,
+      "loss": 1.0167,
+      "step": 3866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016551631769275415,
+      "loss": 0.9688,
+      "step": 3867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016550738872271083,
+      "loss": 1.1788,
+      "step": 3868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016549845975266753,
+      "loss": 1.133,
+      "step": 3869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016548953078262423,
+      "loss": 1.0225,
+      "step": 3870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016548060181258094,
+      "loss": 0.929,
+      "step": 3871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016547167284253761,
+      "loss": 0.9908,
+      "step": 3872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016546274387249432,
+      "loss": 1.1393,
+      "step": 3873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016545381490245102,
+      "loss": 1.1018,
+      "step": 3874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001654448859324077,
+      "loss": 0.9391,
+      "step": 3875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001654359569623644,
+      "loss": 0.9418,
+      "step": 3876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001654270279923211,
+      "loss": 1.1308,
+      "step": 3877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016541809902227778,
+      "loss": 1.0527,
+      "step": 3878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001654091700522345,
+      "loss": 1.0095,
+      "step": 3879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001654002410821912,
+      "loss": 1.0382,
+      "step": 3880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016539131211214787,
+      "loss": 1.0021,
+      "step": 3881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016538238314210457,
+      "loss": 0.9986,
+      "step": 3882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016537345417206128,
+      "loss": 1.0775,
+      "step": 3883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016536452520201795,
+      "loss": 1.065,
+      "step": 3884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016535559623197466,
+      "loss": 1.0311,
+      "step": 3885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016534666726193136,
+      "loss": 1.0576,
+      "step": 3886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016533773829188804,
+      "loss": 1.0564,
+      "step": 3887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016532880932184474,
+      "loss": 1.0432,
+      "step": 3888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016531988035180145,
+      "loss": 0.982,
+      "step": 3889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016531095138175812,
+      "loss": 1.0314,
+      "step": 3890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016530202241171483,
+      "loss": 0.965,
+      "step": 3891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016529309344167153,
+      "loss": 0.9282,
+      "step": 3892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001652841644716282,
+      "loss": 1.0011,
+      "step": 3893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016527523550158491,
+      "loss": 1.026,
+      "step": 3894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001652663065315416,
+      "loss": 1.0198,
+      "step": 3895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001652573775614983,
+      "loss": 1.1803,
+      "step": 3896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016524844859145497,
+      "loss": 0.9909,
+      "step": 3897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016523951962141168,
+      "loss": 1.034,
+      "step": 3898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016523059065136838,
+      "loss": 0.9989,
+      "step": 3899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016522166168132506,
+      "loss": 0.8737,
+      "step": 3900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016521273271128176,
+      "loss": 0.9095,
+      "step": 3901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016520380374123844,
+      "loss": 1.1587,
+      "step": 3902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016519487477119514,
+      "loss": 1.0409,
+      "step": 3903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016518594580115185,
+      "loss": 0.9249,
+      "step": 3904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016517701683110852,
+      "loss": 1.0405,
+      "step": 3905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016516808786106523,
+      "loss": 1.2639,
+      "step": 3906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016515915889102193,
+      "loss": 1.0574,
+      "step": 3907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001651502299209786,
+      "loss": 1.103,
+      "step": 3908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001651413009509353,
+      "loss": 1.0926,
+      "step": 3909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016513237198089202,
+      "loss": 1.0557,
+      "step": 3910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001651234430108487,
+      "loss": 1.0472,
+      "step": 3911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001651145140408054,
+      "loss": 1.0363,
+      "step": 3912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001651055850707621,
+      "loss": 1.0203,
+      "step": 3913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016509665610071878,
+      "loss": 1.0379,
+      "step": 3914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016508772713067548,
+      "loss": 1.1018,
+      "step": 3915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016507879816063219,
+      "loss": 0.947,
+      "step": 3916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016506986919058886,
+      "loss": 1.1549,
+      "step": 3917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016506094022054557,
+      "loss": 1.2126,
+      "step": 3918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016505201125050227,
+      "loss": 0.9612,
+      "step": 3919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00016504308228045895,
+      "loss": 1.1505,
+      "step": 3920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016503415331041565,
+      "loss": 1.1118,
+      "step": 3921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016502522434037236,
+      "loss": 0.8981,
+      "step": 3922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016501629537032903,
+      "loss": 0.905,
+      "step": 3923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016500736640028574,
+      "loss": 1.1008,
+      "step": 3924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016499843743024244,
+      "loss": 1.008,
+      "step": 3925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016498950846019912,
+      "loss": 0.9917,
+      "step": 3926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016498057949015582,
+      "loss": 1.0443,
+      "step": 3927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016497165052011253,
+      "loss": 1.0257,
+      "step": 3928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001649627215500692,
+      "loss": 1.0684,
+      "step": 3929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001649537925800259,
+      "loss": 1.1462,
+      "step": 3930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001649448636099826,
+      "loss": 1.0434,
+      "step": 3931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001649359346399393,
+      "loss": 1.0445,
+      "step": 3932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000164927005669896,
+      "loss": 1.2524,
+      "step": 3933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001649180766998527,
+      "loss": 0.9883,
+      "step": 3934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016490914772980937,
+      "loss": 1.0489,
+      "step": 3935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016490021875976608,
+      "loss": 0.9668,
+      "step": 3936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016489128978972278,
+      "loss": 1.1093,
+      "step": 3937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016488236081967946,
+      "loss": 1.0503,
+      "step": 3938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016487343184963616,
+      "loss": 0.9581,
+      "step": 3939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016486450287959287,
+      "loss": 0.9846,
+      "step": 3940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016485557390954954,
+      "loss": 1.0186,
+      "step": 3941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016484664493950625,
+      "loss": 1.0416,
+      "step": 3942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016483771596946292,
+      "loss": 0.9122,
+      "step": 3943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016482878699941963,
+      "loss": 0.889,
+      "step": 3944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001648198580293763,
+      "loss": 1.1141,
+      "step": 3945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000164810929059333,
+      "loss": 1.0868,
+      "step": 3946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016480200008928969,
+      "loss": 1.0188,
+      "step": 3947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001647930711192464,
+      "loss": 1.1295,
+      "step": 3948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001647841421492031,
+      "loss": 1.1593,
+      "step": 3949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016477521317915977,
+      "loss": 1.1656,
+      "step": 3950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016476628420911647,
+      "loss": 1.243,
+      "step": 3951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016475735523907318,
+      "loss": 1.1339,
+      "step": 3952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016474842626902986,
+      "loss": 1.0226,
+      "step": 3953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016473949729898656,
+      "loss": 1.0571,
+      "step": 3954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016473056832894326,
+      "loss": 1.1232,
+      "step": 3955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016472163935889994,
+      "loss": 0.9355,
+      "step": 3956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016471271038885664,
+      "loss": 1.0769,
+      "step": 3957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016470378141881335,
+      "loss": 0.9836,
+      "step": 3958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016469485244877003,
+      "loss": 0.9923,
+      "step": 3959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016468592347872673,
+      "loss": 1.0645,
+      "step": 3960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016467699450868343,
+      "loss": 1.0349,
+      "step": 3961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001646680655386401,
+      "loss": 1.1387,
+      "step": 3962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016465913656859681,
+      "loss": 0.9402,
+      "step": 3963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016465020759855352,
+      "loss": 0.9602,
+      "step": 3964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001646412786285102,
+      "loss": 0.915,
+      "step": 3965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001646323496584669,
+      "loss": 1.1442,
+      "step": 3966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001646234206884236,
+      "loss": 1.0759,
+      "step": 3967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001646144917183803,
+      "loss": 1.0416,
+      "step": 3968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016460556274833698,
+      "loss": 1.031,
+      "step": 3969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001645966337782937,
+      "loss": 1.0783,
+      "step": 3970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001645877048082504,
+      "loss": 1.1355,
+      "step": 3971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016457877583820707,
+      "loss": 0.9846,
+      "step": 3972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016456984686816377,
+      "loss": 0.9436,
+      "step": 3973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016456091789812048,
+      "loss": 0.9886,
+      "step": 3974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016455198892807715,
+      "loss": 0.9554,
+      "step": 3975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016454305995803386,
+      "loss": 1.2156,
+      "step": 3976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016453413098799056,
+      "loss": 0.9539,
+      "step": 3977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016452520201794724,
+      "loss": 1.069,
+      "step": 3978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016451627304790394,
+      "loss": 1.0296,
+      "step": 3979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016450734407786065,
+      "loss": 1.2806,
+      "step": 3980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016449841510781732,
+      "loss": 1.1534,
+      "step": 3981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016448948613777403,
+      "loss": 1.0769,
+      "step": 3982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016448055716773073,
+      "loss": 1.0324,
+      "step": 3983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001644716281976874,
+      "loss": 0.9838,
+      "step": 3984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001644626992276441,
+      "loss": 1.0485,
+      "step": 3985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001644537702576008,
+      "loss": 1.232,
+      "step": 3986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001644448412875575,
+      "loss": 1.0196,
+      "step": 3987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016443591231751417,
+      "loss": 1.0624,
+      "step": 3988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016442698334747088,
+      "loss": 1.0779,
+      "step": 3989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016441805437742758,
+      "loss": 1.093,
+      "step": 3990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016440912540738426,
+      "loss": 1.0892,
+      "step": 3991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016440019643734096,
+      "loss": 0.9191,
+      "step": 3992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016439126746729764,
+      "loss": 0.9435,
+      "step": 3993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016438233849725434,
+      "loss": 1.0195,
+      "step": 3994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00016437340952721105,
+      "loss": 1.7602,
+      "step": 3995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016436448055716772,
+      "loss": 1.2186,
+      "step": 3996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016435555158712443,
+      "loss": 1.0607,
+      "step": 3997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016434662261708113,
+      "loss": 0.965,
+      "step": 3998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001643376936470378,
+      "loss": 1.0327,
+      "step": 3999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001643287646769945,
+      "loss": 1.0864,
+      "step": 4000
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 1.0427271127700806,
+      "eval_runtime": 3047.7449,
+      "eval_samples_per_second": 6.17,
+      "eval_steps_per_second": 0.771,
+      "step": 4000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016431983570695122,
+      "loss": 0.9872,
+      "step": 4001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001643109067369079,
+      "loss": 1.0022,
+      "step": 4002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001643019777668646,
+      "loss": 1.0217,
+      "step": 4003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001642930487968213,
+      "loss": 1.0054,
+      "step": 4004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016428411982677798,
+      "loss": 1.1261,
+      "step": 4005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016427519085673468,
+      "loss": 0.9305,
+      "step": 4006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016426626188669139,
+      "loss": 0.9274,
+      "step": 4007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016425733291664806,
+      "loss": 0.9355,
+      "step": 4008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016424840394660477,
+      "loss": 1.0099,
+      "step": 4009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016423947497656147,
+      "loss": 0.8991,
+      "step": 4010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016423054600651815,
+      "loss": 0.951,
+      "step": 4011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016422161703647485,
+      "loss": 1.1089,
+      "step": 4012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016421268806643156,
+      "loss": 1.0618,
+      "step": 4013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016420375909638823,
+      "loss": 1.1025,
+      "step": 4014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016419483012634494,
+      "loss": 1.0644,
+      "step": 4015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016418590115630164,
+      "loss": 1.0754,
+      "step": 4016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016417697218625832,
+      "loss": 1.033,
+      "step": 4017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016416804321621502,
+      "loss": 1.0022,
+      "step": 4018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016415911424617173,
+      "loss": 1.0063,
+      "step": 4019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001641501852761284,
+      "loss": 0.9705,
+      "step": 4020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001641412563060851,
+      "loss": 0.9611,
+      "step": 4021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001641323273360418,
+      "loss": 1.2909,
+      "step": 4022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001641233983659985,
+      "loss": 1.0822,
+      "step": 4023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001641144693959552,
+      "loss": 1.0921,
+      "step": 4024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001641055404259119,
+      "loss": 1.0295,
+      "step": 4025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016409661145586857,
+      "loss": 0.9717,
+      "step": 4026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016408768248582528,
+      "loss": 1.1647,
+      "step": 4027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016407875351578198,
+      "loss": 1.0952,
+      "step": 4028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016406982454573866,
+      "loss": 1.0966,
+      "step": 4029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016406089557569536,
+      "loss": 1.0454,
+      "step": 4030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016405196660565207,
+      "loss": 1.0174,
+      "step": 4031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016404303763560874,
+      "loss": 1.1288,
+      "step": 4032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016403410866556545,
+      "loss": 1.068,
+      "step": 4033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016402517969552212,
+      "loss": 1.1676,
+      "step": 4034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016401625072547883,
+      "loss": 0.9327,
+      "step": 4035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001640073217554355,
+      "loss": 1.0712,
+      "step": 4036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001639983927853922,
+      "loss": 0.9636,
+      "step": 4037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016398946381534889,
+      "loss": 1.0966,
+      "step": 4038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001639805348453056,
+      "loss": 1.1153,
+      "step": 4039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001639716058752623,
+      "loss": 1.1841,
+      "step": 4040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016396267690521897,
+      "loss": 1.0139,
+      "step": 4041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016395374793517567,
+      "loss": 1.0111,
+      "step": 4042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016394481896513238,
+      "loss": 1.1146,
+      "step": 4043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016393588999508905,
+      "loss": 1.0945,
+      "step": 4044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016392696102504576,
+      "loss": 0.9748,
+      "step": 4045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016391803205500246,
+      "loss": 1.1121,
+      "step": 4046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016390910308495914,
+      "loss": 1.127,
+      "step": 4047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016390017411491584,
+      "loss": 1.0249,
+      "step": 4048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016389124514487255,
+      "loss": 0.9952,
+      "step": 4049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016388231617482922,
+      "loss": 0.9561,
+      "step": 4050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016387338720478593,
+      "loss": 0.901,
+      "step": 4051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016386445823474263,
+      "loss": 1.1343,
+      "step": 4052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001638555292646993,
+      "loss": 1.0883,
+      "step": 4053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016384660029465601,
+      "loss": 1.0395,
+      "step": 4054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016383767132461272,
+      "loss": 1.2717,
+      "step": 4055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001638287423545694,
+      "loss": 0.9927,
+      "step": 4056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001638198133845261,
+      "loss": 1.1957,
+      "step": 4057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001638108844144828,
+      "loss": 1.042,
+      "step": 4058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016380195544443948,
+      "loss": 0.932,
+      "step": 4059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016379302647439618,
+      "loss": 1.0171,
+      "step": 4060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001637840975043529,
+      "loss": 1.0429,
+      "step": 4061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016377516853430956,
+      "loss": 1.1442,
+      "step": 4062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016376623956426627,
+      "loss": 1.0869,
+      "step": 4063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016375731059422297,
+      "loss": 1.0872,
+      "step": 4064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016374838162417968,
+      "loss": 1.1468,
+      "step": 4065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016373945265413635,
+      "loss": 1.07,
+      "step": 4066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016373052368409306,
+      "loss": 1.0417,
+      "step": 4067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016372159471404976,
+      "loss": 1.0584,
+      "step": 4068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016371266574400644,
+      "loss": 1.0411,
+      "step": 4069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00016370373677396314,
+      "loss": 1.0107,
+      "step": 4070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016369480780391985,
+      "loss": 1.0115,
+      "step": 4071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016368587883387652,
+      "loss": 1.0843,
+      "step": 4072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016367694986383323,
+      "loss": 1.0463,
+      "step": 4073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016366802089378993,
+      "loss": 1.0734,
+      "step": 4074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001636590919237466,
+      "loss": 0.9427,
+      "step": 4075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001636501629537033,
+      "loss": 0.9709,
+      "step": 4076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016364123398366,
+      "loss": 1.0333,
+      "step": 4077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001636323050136167,
+      "loss": 1.1015,
+      "step": 4078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016362337604357337,
+      "loss": 0.9306,
+      "step": 4079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016361444707353007,
+      "loss": 1.1489,
+      "step": 4080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016360551810348678,
+      "loss": 1.0582,
+      "step": 4081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016359658913344346,
+      "loss": 0.9865,
+      "step": 4082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016358766016340016,
+      "loss": 1.0082,
+      "step": 4083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016357873119335684,
+      "loss": 1.1307,
+      "step": 4084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016356980222331354,
+      "loss": 0.9725,
+      "step": 4085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016356087325327024,
+      "loss": 1.0032,
+      "step": 4086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016355194428322692,
+      "loss": 1.0239,
+      "step": 4087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016354301531318363,
+      "loss": 0.9872,
+      "step": 4088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016353408634314033,
+      "loss": 1.008,
+      "step": 4089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000163525157373097,
+      "loss": 1.0109,
+      "step": 4090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001635162284030537,
+      "loss": 0.8814,
+      "step": 4091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016350729943301041,
+      "loss": 0.9902,
+      "step": 4092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001634983704629671,
+      "loss": 1.0348,
+      "step": 4093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001634894414929238,
+      "loss": 1.0321,
+      "step": 4094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001634805125228805,
+      "loss": 1.0656,
+      "step": 4095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016347158355283718,
+      "loss": 1.0495,
+      "step": 4096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016346265458279388,
+      "loss": 0.9763,
+      "step": 4097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016345372561275058,
+      "loss": 1.0332,
+      "step": 4098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016344479664270726,
+      "loss": 1.0591,
+      "step": 4099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016343586767266397,
+      "loss": 1.0919,
+      "step": 4100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016342693870262067,
+      "loss": 1.076,
+      "step": 4101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016341800973257735,
+      "loss": 0.9776,
+      "step": 4102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016340908076253405,
+      "loss": 1.0261,
+      "step": 4103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016340015179249075,
+      "loss": 0.9882,
+      "step": 4104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016339122282244743,
+      "loss": 0.9955,
+      "step": 4105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016338229385240414,
+      "loss": 0.8654,
+      "step": 4106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016337336488236084,
+      "loss": 1.0882,
+      "step": 4107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016336443591231752,
+      "loss": 0.9961,
+      "step": 4108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016335550694227422,
+      "loss": 1.0288,
+      "step": 4109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016334657797223092,
+      "loss": 1.2146,
+      "step": 4110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001633376490021876,
+      "loss": 1.116,
+      "step": 4111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001633287200321443,
+      "loss": 0.9316,
+      "step": 4112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000163319791062101,
+      "loss": 1.0319,
+      "step": 4113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001633108620920577,
+      "loss": 1.1556,
+      "step": 4114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001633019331220144,
+      "loss": 0.9749,
+      "step": 4115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001632930041519711,
+      "loss": 1.1241,
+      "step": 4116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016328407518192777,
+      "loss": 1.108,
+      "step": 4117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016327514621188448,
+      "loss": 0.9766,
+      "step": 4118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016326621724184118,
+      "loss": 1.1132,
+      "step": 4119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016325728827179786,
+      "loss": 0.917,
+      "step": 4120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016324835930175456,
+      "loss": 1.0058,
+      "step": 4121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016323943033171126,
+      "loss": 0.8619,
+      "step": 4122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016323050136166794,
+      "loss": 0.9585,
+      "step": 4123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016322157239162465,
+      "loss": 1.0926,
+      "step": 4124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016321264342158132,
+      "loss": 0.9337,
+      "step": 4125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016320371445153803,
+      "loss": 1.0519,
+      "step": 4126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001631947854814947,
+      "loss": 1.2462,
+      "step": 4127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001631858565114514,
+      "loss": 1.068,
+      "step": 4128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016317692754140808,
+      "loss": 1.1499,
+      "step": 4129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001631679985713648,
+      "loss": 1.0381,
+      "step": 4130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001631590696013215,
+      "loss": 0.9203,
+      "step": 4131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016315014063127817,
+      "loss": 1.0598,
+      "step": 4132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016314121166123487,
+      "loss": 0.9435,
+      "step": 4133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016313228269119158,
+      "loss": 1.1685,
+      "step": 4134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016312335372114825,
+      "loss": 0.9945,
+      "step": 4135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016311442475110496,
+      "loss": 1.0635,
+      "step": 4136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016310549578106166,
+      "loss": 1.0735,
+      "step": 4137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016309656681101834,
+      "loss": 1.0662,
+      "step": 4138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016308763784097504,
+      "loss": 0.9799,
+      "step": 4139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016307870887093175,
+      "loss": 1.0749,
+      "step": 4140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016306977990088842,
+      "loss": 1.0558,
+      "step": 4141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016306085093084513,
+      "loss": 0.9735,
+      "step": 4142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016305192196080183,
+      "loss": 1.0421,
+      "step": 4143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001630429929907585,
+      "loss": 0.9431,
+      "step": 4144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00016303406402071521,
+      "loss": 1.0661,
+      "step": 4145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016302513505067192,
+      "loss": 1.051,
+      "step": 4146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001630162060806286,
+      "loss": 1.119,
+      "step": 4147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001630072771105853,
+      "loss": 0.9607,
+      "step": 4148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000162998348140542,
+      "loss": 1.0218,
+      "step": 4149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016298941917049868,
+      "loss": 1.0848,
+      "step": 4150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016298049020045538,
+      "loss": 0.9952,
+      "step": 4151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001629715612304121,
+      "loss": 0.8935,
+      "step": 4152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016296263226036876,
+      "loss": 1.0789,
+      "step": 4153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016295370329032547,
+      "loss": 1.1734,
+      "step": 4154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016294477432028217,
+      "loss": 0.9119,
+      "step": 4155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016293584535023885,
+      "loss": 1.0457,
+      "step": 4156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016292691638019555,
+      "loss": 1.1764,
+      "step": 4157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016291798741015226,
+      "loss": 1.0359,
+      "step": 4158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016290905844010893,
+      "loss": 1.1677,
+      "step": 4159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016290012947006564,
+      "loss": 1.0588,
+      "step": 4160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016289120050002234,
+      "loss": 0.9864,
+      "step": 4161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016288227152997902,
+      "loss": 1.041,
+      "step": 4162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016287334255993572,
+      "loss": 1.1522,
+      "step": 4163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016286441358989243,
+      "loss": 1.022,
+      "step": 4164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016285548461984913,
+      "loss": 1.1184,
+      "step": 4165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001628465556498058,
+      "loss": 1.1061,
+      "step": 4166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001628376266797625,
+      "loss": 1.002,
+      "step": 4167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001628286977097192,
+      "loss": 1.0232,
+      "step": 4168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001628197687396759,
+      "loss": 1.016,
+      "step": 4169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016281083976963257,
+      "loss": 1.0065,
+      "step": 4170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016280191079958927,
+      "loss": 1.0358,
+      "step": 4171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016279298182954598,
+      "loss": 1.1179,
+      "step": 4172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016278405285950266,
+      "loss": 1.0099,
+      "step": 4173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016277512388945936,
+      "loss": 0.9323,
+      "step": 4174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016276619491941604,
+      "loss": 0.956,
+      "step": 4175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016275726594937274,
+      "loss": 1.0004,
+      "step": 4176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016274833697932944,
+      "loss": 1.0978,
+      "step": 4177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016273940800928612,
+      "loss": 1.0821,
+      "step": 4178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016273047903924283,
+      "loss": 1.0082,
+      "step": 4179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016272155006919953,
+      "loss": 0.8808,
+      "step": 4180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001627126210991562,
+      "loss": 1.0506,
+      "step": 4181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001627036921291129,
+      "loss": 1.0512,
+      "step": 4182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016269476315906961,
+      "loss": 1.0489,
+      "step": 4183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001626858341890263,
+      "loss": 0.9623,
+      "step": 4184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000162676905218983,
+      "loss": 1.0268,
+      "step": 4185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001626679762489397,
+      "loss": 0.9889,
+      "step": 4186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016265904727889638,
+      "loss": 1.114,
+      "step": 4187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016265011830885308,
+      "loss": 1.0751,
+      "step": 4188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016264118933880978,
+      "loss": 1.1178,
+      "step": 4189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016263226036876646,
+      "loss": 1.0636,
+      "step": 4190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016262333139872317,
+      "loss": 1.0496,
+      "step": 4191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016261440242867987,
+      "loss": 0.9761,
+      "step": 4192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016260547345863655,
+      "loss": 1.0802,
+      "step": 4193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016259654448859325,
+      "loss": 1.0483,
+      "step": 4194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016258761551854995,
+      "loss": 0.9913,
+      "step": 4195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016257868654850663,
+      "loss": 1.0431,
+      "step": 4196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016256975757846334,
+      "loss": 1.0572,
+      "step": 4197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016256082860842004,
+      "loss": 0.9656,
+      "step": 4198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016255189963837672,
+      "loss": 0.7921,
+      "step": 4199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016254297066833342,
+      "loss": 1.1071,
+      "step": 4200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016253404169829012,
+      "loss": 1.0854,
+      "step": 4201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001625251127282468,
+      "loss": 1.1062,
+      "step": 4202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001625161837582035,
+      "loss": 1.1121,
+      "step": 4203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001625072547881602,
+      "loss": 1.1416,
+      "step": 4204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016249832581811689,
+      "loss": 1.0368,
+      "step": 4205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001624893968480736,
+      "loss": 1.0969,
+      "step": 4206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001624804678780303,
+      "loss": 1.0884,
+      "step": 4207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016247153890798697,
+      "loss": 1.0809,
+      "step": 4208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016246260993794368,
+      "loss": 1.0789,
+      "step": 4209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016245368096790038,
+      "loss": 0.9171,
+      "step": 4210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016244475199785706,
+      "loss": 0.969,
+      "step": 4211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016243582302781376,
+      "loss": 1.0947,
+      "step": 4212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016242689405777046,
+      "loss": 1.0816,
+      "step": 4213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016241796508772714,
+      "loss": 1.1094,
+      "step": 4214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016240903611768385,
+      "loss": 0.9856,
+      "step": 4215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016240010714764052,
+      "loss": 1.0453,
+      "step": 4216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016239117817759723,
+      "loss": 1.1035,
+      "step": 4217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001623822492075539,
+      "loss": 1.0249,
+      "step": 4218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001623733202375106,
+      "loss": 1.0241,
+      "step": 4219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016236439126746728,
+      "loss": 1.1495,
+      "step": 4220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000162355462297424,
+      "loss": 1.0952,
+      "step": 4221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001623465333273807,
+      "loss": 1.0137,
+      "step": 4222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016233760435733737,
+      "loss": 0.953,
+      "step": 4223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016232867538729407,
+      "loss": 1.0833,
+      "step": 4224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016231974641725078,
+      "loss": 1.067,
+      "step": 4225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016231081744720745,
+      "loss": 1.0637,
+      "step": 4226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016230188847716416,
+      "loss": 1.0891,
+      "step": 4227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016229295950712086,
+      "loss": 0.9565,
+      "step": 4228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016228403053707754,
+      "loss": 1.0497,
+      "step": 4229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016227510156703424,
+      "loss": 1.005,
+      "step": 4230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016226617259699095,
+      "loss": 1.1353,
+      "step": 4231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016225724362694762,
+      "loss": 1.1614,
+      "step": 4232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016224831465690433,
+      "loss": 1.0902,
+      "step": 4233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016223938568686103,
+      "loss": 1.1342,
+      "step": 4234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001622304567168177,
+      "loss": 1.0993,
+      "step": 4235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001622215277467744,
+      "loss": 1.1192,
+      "step": 4236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016221259877673112,
+      "loss": 0.8861,
+      "step": 4237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001622036698066878,
+      "loss": 1.1199,
+      "step": 4238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001621947408366445,
+      "loss": 1.0782,
+      "step": 4239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001621858118666012,
+      "loss": 1.0394,
+      "step": 4240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016217688289655788,
+      "loss": 0.9861,
+      "step": 4241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016216795392651458,
+      "loss": 1.0136,
+      "step": 4242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001621590249564713,
+      "loss": 1.109,
+      "step": 4243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016215009598642796,
+      "loss": 1.1032,
+      "step": 4244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016214116701638467,
+      "loss": 1.0833,
+      "step": 4245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016213223804634137,
+      "loss": 1.0178,
+      "step": 4246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016212330907629805,
+      "loss": 1.0073,
+      "step": 4247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016211438010625475,
+      "loss": 1.1875,
+      "step": 4248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016210545113621146,
+      "loss": 1.0452,
+      "step": 4249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016209652216616813,
+      "loss": 0.966,
+      "step": 4250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016208759319612484,
+      "loss": 1.0111,
+      "step": 4251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016207866422608154,
+      "loss": 1.0148,
+      "step": 4252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016206973525603822,
+      "loss": 1.1097,
+      "step": 4253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016206080628599492,
+      "loss": 1.004,
+      "step": 4254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016205187731595163,
+      "loss": 1.1266,
+      "step": 4255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001620429483459083,
+      "loss": 1.0387,
+      "step": 4256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000162034019375865,
+      "loss": 0.9947,
+      "step": 4257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001620250904058217,
+      "loss": 0.9411,
+      "step": 4258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001620161614357784,
+      "loss": 0.9965,
+      "step": 4259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001620072324657351,
+      "loss": 1.0299,
+      "step": 4260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016199830349569177,
+      "loss": 0.9772,
+      "step": 4261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016198937452564847,
+      "loss": 1.0545,
+      "step": 4262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016198044555560518,
+      "loss": 1.2919,
+      "step": 4263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016197151658556186,
+      "loss": 1.0861,
+      "step": 4264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016196258761551856,
+      "loss": 1.0265,
+      "step": 4265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016195365864547524,
+      "loss": 0.9719,
+      "step": 4266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016194472967543194,
+      "loss": 0.9909,
+      "step": 4267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016193580070538864,
+      "loss": 1.2025,
+      "step": 4268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016192687173534532,
+      "loss": 1.0717,
+      "step": 4269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016191794276530203,
+      "loss": 1.0762,
+      "step": 4270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016190901379525873,
+      "loss": 1.0153,
+      "step": 4271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001619000848252154,
+      "loss": 0.9484,
+      "step": 4272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001618911558551721,
+      "loss": 1.0102,
+      "step": 4273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016188222688512881,
+      "loss": 1.0272,
+      "step": 4274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001618732979150855,
+      "loss": 0.9834,
+      "step": 4275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001618643689450422,
+      "loss": 1.0223,
+      "step": 4276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001618554399749989,
+      "loss": 1.0838,
+      "step": 4277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016184651100495558,
+      "loss": 1.0461,
+      "step": 4278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016183758203491228,
+      "loss": 1.0471,
+      "step": 4279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016182865306486898,
+      "loss": 1.1086,
+      "step": 4280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016181972409482566,
+      "loss": 1.0353,
+      "step": 4281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016181079512478237,
+      "loss": 0.9446,
+      "step": 4282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016180186615473907,
+      "loss": 1.1389,
+      "step": 4283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016179293718469575,
+      "loss": 1.054,
+      "step": 4284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016178400821465245,
+      "loss": 1.0981,
+      "step": 4285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016177507924460915,
+      "loss": 1.0896,
+      "step": 4286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016176615027456583,
+      "loss": 1.0846,
+      "step": 4287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016175722130452254,
+      "loss": 1.0435,
+      "step": 4288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016174829233447924,
+      "loss": 1.119,
+      "step": 4289
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016173936336443592,
+      "loss": 0.9962,
+      "step": 4290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016173043439439262,
+      "loss": 1.0336,
+      "step": 4291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016172150542434932,
+      "loss": 1.0526,
+      "step": 4292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000161712576454306,
+      "loss": 1.0854,
+      "step": 4293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001617036474842627,
+      "loss": 0.9952,
+      "step": 4294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001616947185142194,
+      "loss": 1.1535,
+      "step": 4295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016168578954417609,
+      "loss": 0.9761,
+      "step": 4296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001616768605741328,
+      "loss": 1.0696,
+      "step": 4297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001616679316040895,
+      "loss": 1.1529,
+      "step": 4298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016165900263404617,
+      "loss": 1.1251,
+      "step": 4299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016165007366400288,
+      "loss": 1.1308,
+      "step": 4300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016164114469395958,
+      "loss": 1.1421,
+      "step": 4301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016163221572391626,
+      "loss": 0.9602,
+      "step": 4302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016162328675387296,
+      "loss": 1.0348,
+      "step": 4303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016161435778382966,
+      "loss": 0.9558,
+      "step": 4304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016160542881378634,
+      "loss": 0.9381,
+      "step": 4305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016159649984374305,
+      "loss": 1.003,
+      "step": 4306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016158757087369972,
+      "loss": 0.9955,
+      "step": 4307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016157864190365643,
+      "loss": 1.019,
+      "step": 4308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001615697129336131,
+      "loss": 1.0551,
+      "step": 4309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001615607839635698,
+      "loss": 1.0275,
+      "step": 4310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016155185499352648,
+      "loss": 0.9703,
+      "step": 4311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001615429260234832,
+      "loss": 1.1667,
+      "step": 4312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001615339970534399,
+      "loss": 0.963,
+      "step": 4313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016152506808339657,
+      "loss": 0.8985,
+      "step": 4314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016151613911335327,
+      "loss": 1.0245,
+      "step": 4315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016150721014330998,
+      "loss": 1.0437,
+      "step": 4316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016149828117326665,
+      "loss": 1.0863,
+      "step": 4317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016148935220322336,
+      "loss": 1.0341,
+      "step": 4318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016148042323318006,
+      "loss": 0.9989,
+      "step": 4319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016147149426313674,
+      "loss": 1.0276,
+      "step": 4320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016146256529309344,
+      "loss": 1.0716,
+      "step": 4321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016145363632305015,
+      "loss": 1.008,
+      "step": 4322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016144470735300682,
+      "loss": 1.0616,
+      "step": 4323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016143577838296353,
+      "loss": 0.9559,
+      "step": 4324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016142684941292023,
+      "loss": 1.0023,
+      "step": 4325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001614179204428769,
+      "loss": 1.0788,
+      "step": 4326
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001614089914728336,
+      "loss": 0.9931,
+      "step": 4327
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016140006250279032,
+      "loss": 1.1115,
+      "step": 4328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000161391133532747,
+      "loss": 1.0176,
+      "step": 4329
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001613822045627037,
+      "loss": 0.9651,
+      "step": 4330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001613732755926604,
+      "loss": 0.9334,
+      "step": 4331
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016136434662261708,
+      "loss": 1.1106,
+      "step": 4332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016135541765257378,
+      "loss": 0.8824,
+      "step": 4333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001613464886825305,
+      "loss": 1.1502,
+      "step": 4334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016133755971248716,
+      "loss": 1.0154,
+      "step": 4335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016132863074244387,
+      "loss": 1.1722,
+      "step": 4336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016131970177240057,
+      "loss": 0.9749,
+      "step": 4337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016131077280235725,
+      "loss": 1.0803,
+      "step": 4338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016130184383231395,
+      "loss": 1.1168,
+      "step": 4339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016129291486227066,
+      "loss": 1.0442,
+      "step": 4340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016128398589222733,
+      "loss": 0.9268,
+      "step": 4341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016127505692218404,
+      "loss": 1.082,
+      "step": 4342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016126612795214074,
+      "loss": 0.9949,
+      "step": 4343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016125719898209742,
+      "loss": 0.9766,
+      "step": 4344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016124827001205412,
+      "loss": 1.3074,
+      "step": 4345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016123934104201083,
+      "loss": 1.0179,
+      "step": 4346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001612304120719675,
+      "loss": 1.0837,
+      "step": 4347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001612214831019242,
+      "loss": 1.1991,
+      "step": 4348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001612125541318809,
+      "loss": 1.1579,
+      "step": 4349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001612036251618376,
+      "loss": 1.0403,
+      "step": 4350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001611946961917943,
+      "loss": 1.0033,
+      "step": 4351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016118576722175097,
+      "loss": 1.0001,
+      "step": 4352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016117683825170767,
+      "loss": 0.9944,
+      "step": 4353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016116790928166438,
+      "loss": 1.0034,
+      "step": 4354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016115898031162105,
+      "loss": 1.0052,
+      "step": 4355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016115005134157776,
+      "loss": 0.8993,
+      "step": 4356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016114112237153444,
+      "loss": 1.0114,
+      "step": 4357
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016113219340149114,
+      "loss": 0.9987,
+      "step": 4358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016112326443144784,
+      "loss": 1.0697,
+      "step": 4359
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016111433546140452,
+      "loss": 1.0393,
+      "step": 4360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016110540649136122,
+      "loss": 1.0698,
+      "step": 4361
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016109647752131793,
+      "loss": 1.137,
+      "step": 4362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001610875485512746,
+      "loss": 1.0576,
+      "step": 4363
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001610786195812313,
+      "loss": 1.0538,
+      "step": 4364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016106969061118801,
+      "loss": 0.9423,
+      "step": 4365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001610607616411447,
+      "loss": 0.9281,
+      "step": 4366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001610518326711014,
+      "loss": 1.1499,
+      "step": 4367
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001610429037010581,
+      "loss": 1.0796,
+      "step": 4368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016103397473101478,
+      "loss": 0.9805,
+      "step": 4369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016102504576097148,
+      "loss": 1.0518,
+      "step": 4370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016101611679092818,
+      "loss": 0.9718,
+      "step": 4371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016100718782088486,
+      "loss": 1.0073,
+      "step": 4372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016099825885084156,
+      "loss": 1.0451,
+      "step": 4373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016098932988079827,
+      "loss": 1.0702,
+      "step": 4374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016098040091075495,
+      "loss": 0.999,
+      "step": 4375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016097147194071165,
+      "loss": 1.0135,
+      "step": 4376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016096254297066835,
+      "loss": 1.109,
+      "step": 4377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016095361400062503,
+      "loss": 1.1439,
+      "step": 4378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016094468503058173,
+      "loss": 0.9777,
+      "step": 4379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016093575606053844,
+      "loss": 0.985,
+      "step": 4380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016092682709049512,
+      "loss": 0.9662,
+      "step": 4381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016091789812045182,
+      "loss": 1.1121,
+      "step": 4382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016090896915040852,
+      "loss": 1.1497,
+      "step": 4383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001609000401803652,
+      "loss": 1.0671,
+      "step": 4384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001608911112103219,
+      "loss": 1.2058,
+      "step": 4385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001608821822402786,
+      "loss": 0.9957,
+      "step": 4386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016087325327023529,
+      "loss": 1.11,
+      "step": 4387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000160864324300192,
+      "loss": 1.0334,
+      "step": 4388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001608553953301487,
+      "loss": 0.9819,
+      "step": 4389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016084646636010537,
+      "loss": 0.992,
+      "step": 4390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016083753739006207,
+      "loss": 1.0914,
+      "step": 4391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016082860842001878,
+      "loss": 1.0364,
+      "step": 4392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016081967944997546,
+      "loss": 1.0162,
+      "step": 4393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016081075047993216,
+      "loss": 1.0606,
+      "step": 4394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016080182150988886,
+      "loss": 0.9243,
+      "step": 4395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016079289253984554,
+      "loss": 1.0652,
+      "step": 4396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016078396356980224,
+      "loss": 1.0084,
+      "step": 4397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016077503459975892,
+      "loss": 1.0691,
+      "step": 4398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016076610562971563,
+      "loss": 1.0655,
+      "step": 4399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001607571766596723,
+      "loss": 1.0273,
+      "step": 4400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000160748247689629,
+      "loss": 0.9702,
+      "step": 4401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016073931871958568,
+      "loss": 1.003,
+      "step": 4402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001607303897495424,
+      "loss": 1.0397,
+      "step": 4403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001607214607794991,
+      "loss": 0.9239,
+      "step": 4404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016071253180945577,
+      "loss": 0.9876,
+      "step": 4405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016070360283941247,
+      "loss": 1.1481,
+      "step": 4406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016069467386936918,
+      "loss": 0.8795,
+      "step": 4407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016068574489932585,
+      "loss": 0.9507,
+      "step": 4408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016067681592928256,
+      "loss": 1.0437,
+      "step": 4409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016066788695923926,
+      "loss": 0.9731,
+      "step": 4410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016065895798919594,
+      "loss": 0.948,
+      "step": 4411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016065002901915264,
+      "loss": 1.026,
+      "step": 4412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016064110004910935,
+      "loss": 1.1066,
+      "step": 4413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016063217107906602,
+      "loss": 1.0231,
+      "step": 4414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016062324210902273,
+      "loss": 1.0498,
+      "step": 4415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016061431313897943,
+      "loss": 1.0691,
+      "step": 4416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001606053841689361,
+      "loss": 0.9787,
+      "step": 4417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001605964551988928,
+      "loss": 0.9525,
+      "step": 4418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016058752622884952,
+      "loss": 1.1473,
+      "step": 4419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001605785972588062,
+      "loss": 1.0185,
+      "step": 4420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001605696682887629,
+      "loss": 1.0559,
+      "step": 4421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001605607393187196,
+      "loss": 1.2808,
+      "step": 4422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016055181034867628,
+      "loss": 1.0327,
+      "step": 4423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016054288137863298,
+      "loss": 1.0321,
+      "step": 4424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001605339524085897,
+      "loss": 1.089,
+      "step": 4425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016052502343854636,
+      "loss": 1.0669,
+      "step": 4426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016051609446850307,
+      "loss": 1.0148,
+      "step": 4427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016050716549845977,
+      "loss": 0.9106,
+      "step": 4428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016049823652841645,
+      "loss": 0.9636,
+      "step": 4429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016048930755837315,
+      "loss": 1.156,
+      "step": 4430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016048037858832986,
+      "loss": 0.9119,
+      "step": 4431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016047144961828653,
+      "loss": 1.0129,
+      "step": 4432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016046252064824324,
+      "loss": 1.0583,
+      "step": 4433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016045359167819994,
+      "loss": 0.9101,
+      "step": 4434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016044466270815662,
+      "loss": 1.2015,
+      "step": 4435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016043573373811332,
+      "loss": 1.0194,
+      "step": 4436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016042680476807003,
+      "loss": 0.9819,
+      "step": 4437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001604178757980267,
+      "loss": 1.0193,
+      "step": 4438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001604089468279834,
+      "loss": 1.0758,
+      "step": 4439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001604000178579401,
+      "loss": 1.1434,
+      "step": 4440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001603910888878968,
+      "loss": 0.9763,
+      "step": 4441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001603821599178535,
+      "loss": 1.0498,
+      "step": 4442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016037323094781017,
+      "loss": 0.9968,
+      "step": 4443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016036430197776687,
+      "loss": 1.0838,
+      "step": 4444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016035537300772355,
+      "loss": 0.9866,
+      "step": 4445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016034644403768025,
+      "loss": 1.0094,
+      "step": 4446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016033751506763696,
+      "loss": 0.914,
+      "step": 4447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016032858609759364,
+      "loss": 1.003,
+      "step": 4448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016031965712755034,
+      "loss": 1.0417,
+      "step": 4449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016031072815750702,
+      "loss": 1.1089,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016030179918746372,
+      "loss": 1.0781,
+      "step": 4451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016029287021742042,
+      "loss": 0.9724,
+      "step": 4452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001602839412473771,
+      "loss": 0.998,
+      "step": 4453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001602750122773338,
+      "loss": 1.1298,
+      "step": 4454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001602660833072905,
+      "loss": 0.9764,
+      "step": 4455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001602571543372472,
+      "loss": 0.9991,
+      "step": 4456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001602482253672039,
+      "loss": 1.1686,
+      "step": 4457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001602392963971606,
+      "loss": 1.0445,
+      "step": 4458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001602303674271173,
+      "loss": 1.0685,
+      "step": 4459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016022143845707398,
+      "loss": 1.0224,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016021250948703068,
+      "loss": 1.1362,
+      "step": 4461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016020358051698738,
+      "loss": 0.9743,
+      "step": 4462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016019465154694406,
+      "loss": 1.0513,
+      "step": 4463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016018572257690076,
+      "loss": 1.0535,
+      "step": 4464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016017679360685747,
+      "loss": 0.9567,
+      "step": 4465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016016786463681415,
+      "loss": 1.0804,
+      "step": 4466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016015893566677085,
+      "loss": 1.2647,
+      "step": 4467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016015000669672755,
+      "loss": 1.0001,
+      "step": 4468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016014107772668423,
+      "loss": 1.1173,
+      "step": 4469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016013214875664093,
+      "loss": 0.9683,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016012321978659764,
+      "loss": 1.0571,
+      "step": 4471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016011429081655432,
+      "loss": 0.9664,
+      "step": 4472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016010536184651102,
+      "loss": 1.0291,
+      "step": 4473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016009643287646772,
+      "loss": 1.1151,
+      "step": 4474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001600875039064244,
+      "loss": 0.9912,
+      "step": 4475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001600785749363811,
+      "loss": 0.9949,
+      "step": 4476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001600696459663378,
+      "loss": 1.1489,
+      "step": 4477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016006071699629449,
+      "loss": 1.0801,
+      "step": 4478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001600517880262512,
+      "loss": 0.9088,
+      "step": 4479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001600428590562079,
+      "loss": 1.162,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016003393008616457,
+      "loss": 1.0096,
+      "step": 4481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016002500111612127,
+      "loss": 0.9494,
+      "step": 4482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016001607214607798,
+      "loss": 0.989,
+      "step": 4483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016000714317603466,
+      "loss": 1.0237,
+      "step": 4484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015999821420599136,
+      "loss": 1.0092,
+      "step": 4485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015998928523594806,
+      "loss": 0.9864,
+      "step": 4486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015998035626590474,
+      "loss": 1.0081,
+      "step": 4487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015997142729586144,
+      "loss": 1.0043,
+      "step": 4488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015996249832581812,
+      "loss": 0.934,
+      "step": 4489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015995356935577483,
+      "loss": 1.1448,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001599446403857315,
+      "loss": 1.0402,
+      "step": 4491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001599357114156882,
+      "loss": 1.0127,
+      "step": 4492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015992678244564488,
+      "loss": 0.9248,
+      "step": 4493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001599178534756016,
+      "loss": 1.029,
+      "step": 4494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001599089245055583,
+      "loss": 1.0078,
+      "step": 4495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015989999553551497,
+      "loss": 1.0145,
+      "step": 4496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015989106656547167,
+      "loss": 1.1276,
+      "step": 4497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015988213759542838,
+      "loss": 1.198,
+      "step": 4498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015987320862538505,
+      "loss": 1.013,
+      "step": 4499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015986427965534176,
+      "loss": 1.0568,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015985535068529846,
+      "loss": 1.0287,
+      "step": 4501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015984642171525514,
+      "loss": 1.0561,
+      "step": 4502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015983749274521184,
+      "loss": 1.1229,
+      "step": 4503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015982856377516855,
+      "loss": 1.0375,
+      "step": 4504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015981963480512522,
+      "loss": 1.0234,
+      "step": 4505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015981070583508193,
+      "loss": 1.0262,
+      "step": 4506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015980177686503863,
+      "loss": 1.1091,
+      "step": 4507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001597928478949953,
+      "loss": 1.0083,
+      "step": 4508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000159783918924952,
+      "loss": 1.1976,
+      "step": 4509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015977498995490872,
+      "loss": 1.1242,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001597660609848654,
+      "loss": 1.0048,
+      "step": 4511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001597571320148221,
+      "loss": 1.0105,
+      "step": 4512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001597482030447788,
+      "loss": 0.9703,
+      "step": 4513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015973927407473548,
+      "loss": 1.0152,
+      "step": 4514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015973034510469218,
+      "loss": 1.0223,
+      "step": 4515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015972141613464889,
+      "loss": 1.1225,
+      "step": 4516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015971248716460556,
+      "loss": 0.972,
+      "step": 4517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015970355819456227,
+      "loss": 1.0045,
+      "step": 4518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015969462922451897,
+      "loss": 0.9812,
+      "step": 4519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015968570025447565,
+      "loss": 0.9515,
+      "step": 4520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015967677128443235,
+      "loss": 1.0465,
+      "step": 4521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015966784231438906,
+      "loss": 1.0334,
+      "step": 4522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015965891334434573,
+      "loss": 1.1717,
+      "step": 4523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015964998437430244,
+      "loss": 1.0697,
+      "step": 4524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015964105540425914,
+      "loss": 1.0739,
+      "step": 4525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015963212643421582,
+      "loss": 0.9941,
+      "step": 4526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015962319746417252,
+      "loss": 1.0351,
+      "step": 4527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015961426849412923,
+      "loss": 1.0231,
+      "step": 4528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001596053395240859,
+      "loss": 1.0838,
+      "step": 4529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001595964105540426,
+      "loss": 1.0326,
+      "step": 4530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001595874815839993,
+      "loss": 1.0772,
+      "step": 4531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000159578552613956,
+      "loss": 1.123,
+      "step": 4532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001595696236439127,
+      "loss": 0.9451,
+      "step": 4533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015956069467386937,
+      "loss": 1.0776,
+      "step": 4534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015955176570382607,
+      "loss": 1.0503,
+      "step": 4535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015954283673378275,
+      "loss": 1.0565,
+      "step": 4536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015953390776373945,
+      "loss": 0.9575,
+      "step": 4537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015952497879369616,
+      "loss": 1.1757,
+      "step": 4538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015951604982365283,
+      "loss": 1.1156,
+      "step": 4539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015950712085360954,
+      "loss": 1.0383,
+      "step": 4540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015949819188356622,
+      "loss": 1.0316,
+      "step": 4541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015948926291352292,
+      "loss": 1.1118,
+      "step": 4542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015948033394347962,
+      "loss": 1.0261,
+      "step": 4543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001594714049734363,
+      "loss": 1.0386,
+      "step": 4544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000159462476003393,
+      "loss": 1.0658,
+      "step": 4545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001594535470333497,
+      "loss": 0.942,
+      "step": 4546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015944461806330639,
+      "loss": 1.0186,
+      "step": 4547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001594356890932631,
+      "loss": 1.1103,
+      "step": 4548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001594267601232198,
+      "loss": 1.073,
+      "step": 4549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015941783115317647,
+      "loss": 1.1244,
+      "step": 4550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015940890218313317,
+      "loss": 1.0662,
+      "step": 4551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015939997321308988,
+      "loss": 1.0231,
+      "step": 4552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015939104424304656,
+      "loss": 0.9755,
+      "step": 4553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015938211527300326,
+      "loss": 0.9662,
+      "step": 4554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015937318630295996,
+      "loss": 0.9933,
+      "step": 4555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015936425733291667,
+      "loss": 0.9738,
+      "step": 4556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015935532836287334,
+      "loss": 0.9559,
+      "step": 4557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015934639939283005,
+      "loss": 1.0234,
+      "step": 4558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015933747042278675,
+      "loss": 1.0407,
+      "step": 4559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015932854145274343,
+      "loss": 1.0942,
+      "step": 4560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015931961248270013,
+      "loss": 1.2189,
+      "step": 4561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015931068351265684,
+      "loss": 1.1551,
+      "step": 4562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015930175454261351,
+      "loss": 1.0519,
+      "step": 4563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015929282557257022,
+      "loss": 1.0572,
+      "step": 4564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015928389660252692,
+      "loss": 0.9855,
+      "step": 4565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001592749676324836,
+      "loss": 1.0037,
+      "step": 4566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001592660386624403,
+      "loss": 0.9866,
+      "step": 4567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000159257109692397,
+      "loss": 1.0816,
+      "step": 4568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015924818072235368,
+      "loss": 0.9466,
+      "step": 4569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001592392517523104,
+      "loss": 1.0446,
+      "step": 4570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001592303227822671,
+      "loss": 0.9935,
+      "step": 4571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015922139381222377,
+      "loss": 1.1065,
+      "step": 4572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015921246484218047,
+      "loss": 1.0226,
+      "step": 4573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015920353587213718,
+      "loss": 0.9952,
+      "step": 4574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015919460690209385,
+      "loss": 0.9638,
+      "step": 4575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015918567793205056,
+      "loss": 1.12,
+      "step": 4576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015917674896200724,
+      "loss": 1.1279,
+      "step": 4577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015916781999196394,
+      "loss": 1.0478,
+      "step": 4578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015915889102192064,
+      "loss": 1.0473,
+      "step": 4579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015914996205187732,
+      "loss": 1.0512,
+      "step": 4580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015914103308183402,
+      "loss": 0.9659,
+      "step": 4581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001591321041117907,
+      "loss": 1.0474,
+      "step": 4582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001591231751417474,
+      "loss": 1.002,
+      "step": 4583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015911424617170408,
+      "loss": 1.0966,
+      "step": 4584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001591053172016608,
+      "loss": 1.023,
+      "step": 4585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001590963882316175,
+      "loss": 0.9552,
+      "step": 4586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015908745926157417,
+      "loss": 1.0109,
+      "step": 4587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015907853029153087,
+      "loss": 1.0368,
+      "step": 4588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015906960132148758,
+      "loss": 0.9782,
+      "step": 4589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015906067235144425,
+      "loss": 0.978,
+      "step": 4590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015905174338140096,
+      "loss": 1.0873,
+      "step": 4591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015904281441135766,
+      "loss": 1.221,
+      "step": 4592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00015903388544131434,
+      "loss": 1.0648,
+      "step": 4593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015902495647127104,
+      "loss": 1.0887,
+      "step": 4594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015901602750122775,
+      "loss": 1.071,
+      "step": 4595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015900709853118442,
+      "loss": 0.8973,
+      "step": 4596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015899816956114113,
+      "loss": 0.9413,
+      "step": 4597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015898924059109783,
+      "loss": 1.2081,
+      "step": 4598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001589803116210545,
+      "loss": 0.9795,
+      "step": 4599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001589713826510112,
+      "loss": 0.9527,
+      "step": 4600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015896245368096792,
+      "loss": 0.9937,
+      "step": 4601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001589535247109246,
+      "loss": 1.004,
+      "step": 4602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001589445957408813,
+      "loss": 1.1002,
+      "step": 4603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000158935666770838,
+      "loss": 1.0195,
+      "step": 4604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015892673780079468,
+      "loss": 1.1146,
+      "step": 4605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015891780883075138,
+      "loss": 1.096,
+      "step": 4606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015890887986070809,
+      "loss": 1.1351,
+      "step": 4607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015889995089066476,
+      "loss": 1.0485,
+      "step": 4608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015889102192062147,
+      "loss": 0.9795,
+      "step": 4609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015888209295057817,
+      "loss": 0.9497,
+      "step": 4610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015887316398053485,
+      "loss": 1.0995,
+      "step": 4611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015886423501049155,
+      "loss": 1.0337,
+      "step": 4612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015885530604044826,
+      "loss": 1.0753,
+      "step": 4613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015884637707040493,
+      "loss": 0.9694,
+      "step": 4614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015883744810036164,
+      "loss": 1.0025,
+      "step": 4615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015882851913031834,
+      "loss": 1.1759,
+      "step": 4616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015881959016027502,
+      "loss": 1.0361,
+      "step": 4617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015881066119023172,
+      "loss": 1.0404,
+      "step": 4618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015880173222018843,
+      "loss": 0.9469,
+      "step": 4619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587928032501451,
+      "loss": 1.0626,
+      "step": 4620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587838742801018,
+      "loss": 1.0709,
+      "step": 4621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587749453100585,
+      "loss": 1.094,
+      "step": 4622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587660163400152,
+      "loss": 1.0907,
+      "step": 4623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001587570873699719,
+      "loss": 1.0216,
+      "step": 4624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015874815839992857,
+      "loss": 1.1479,
+      "step": 4625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015873922942988527,
+      "loss": 0.963,
+      "step": 4626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015873030045984195,
+      "loss": 0.9971,
+      "step": 4627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015872137148979865,
+      "loss": 1.0886,
+      "step": 4628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015871244251975536,
+      "loss": 0.9491,
+      "step": 4629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015870351354971203,
+      "loss": 0.996,
+      "step": 4630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015869458457966874,
+      "loss": 1.0711,
+      "step": 4631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015868565560962542,
+      "loss": 0.9811,
+      "step": 4632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015867672663958212,
+      "loss": 1.0991,
+      "step": 4633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015866779766953882,
+      "loss": 1.0354,
+      "step": 4634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001586588686994955,
+      "loss": 1.0399,
+      "step": 4635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001586499397294522,
+      "loss": 1.1292,
+      "step": 4636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001586410107594089,
+      "loss": 1.0822,
+      "step": 4637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015863208178936559,
+      "loss": 1.1927,
+      "step": 4638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001586231528193223,
+      "loss": 1.0211,
+      "step": 4639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000158614223849279,
+      "loss": 1.0379,
+      "step": 4640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015860529487923567,
+      "loss": 1.0601,
+      "step": 4641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015859636590919237,
+      "loss": 1.1303,
+      "step": 4642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015858743693914908,
+      "loss": 1.2066,
+      "step": 4643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015857850796910576,
+      "loss": 0.8799,
+      "step": 4644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015856957899906246,
+      "loss": 1.0566,
+      "step": 4645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015856065002901916,
+      "loss": 1.0394,
+      "step": 4646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015855172105897584,
+      "loss": 1.007,
+      "step": 4647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015854279208893254,
+      "loss": 1.1243,
+      "step": 4648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015853386311888925,
+      "loss": 1.146,
+      "step": 4649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015852493414884593,
+      "loss": 0.9406,
+      "step": 4650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015851600517880263,
+      "loss": 1.0605,
+      "step": 4651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015850707620875933,
+      "loss": 1.0196,
+      "step": 4652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015849814723871604,
+      "loss": 0.9403,
+      "step": 4653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015848921826867271,
+      "loss": 1.1413,
+      "step": 4654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015848028929862942,
+      "loss": 1.0433,
+      "step": 4655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015847136032858612,
+      "loss": 1.1812,
+      "step": 4656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001584624313585428,
+      "loss": 1.0582,
+      "step": 4657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001584535023884995,
+      "loss": 1.0382,
+      "step": 4658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001584445734184562,
+      "loss": 1.0171,
+      "step": 4659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015843564444841288,
+      "loss": 0.9924,
+      "step": 4660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001584267154783696,
+      "loss": 1.1052,
+      "step": 4661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001584177865083263,
+      "loss": 1.0965,
+      "step": 4662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015840885753828297,
+      "loss": 1.0515,
+      "step": 4663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015839992856823967,
+      "loss": 1.0806,
+      "step": 4664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015839099959819638,
+      "loss": 1.117,
+      "step": 4665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015838207062815305,
+      "loss": 1.0462,
+      "step": 4666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00015837314165810976,
+      "loss": 1.0187,
+      "step": 4667
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015836421268806644,
+      "loss": 0.9937,
+      "step": 4668
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015835528371802314,
+      "loss": 1.0352,
+      "step": 4669
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015834635474797984,
+      "loss": 1.0168,
+      "step": 4670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015833742577793652,
+      "loss": 0.9046,
+      "step": 4671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015832849680789322,
+      "loss": 1.0403,
+      "step": 4672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001583195678378499,
+      "loss": 1.0311,
+      "step": 4673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001583106388678066,
+      "loss": 1.0508,
+      "step": 4674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015830170989776328,
+      "loss": 0.99,
+      "step": 4675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015829278092772,
+      "loss": 1.0112,
+      "step": 4676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001582838519576767,
+      "loss": 1.016,
+      "step": 4677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015827492298763337,
+      "loss": 0.9859,
+      "step": 4678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015826599401759007,
+      "loss": 0.9468,
+      "step": 4679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015825706504754678,
+      "loss": 1.0077,
+      "step": 4680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015824813607750345,
+      "loss": 1.0897,
+      "step": 4681
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015823920710746016,
+      "loss": 1.1744,
+      "step": 4682
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015823027813741686,
+      "loss": 1.0272,
+      "step": 4683
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015822134916737354,
+      "loss": 1.1167,
+      "step": 4684
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015821242019733024,
+      "loss": 0.9929,
+      "step": 4685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015820349122728695,
+      "loss": 1.0211,
+      "step": 4686
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015819456225724362,
+      "loss": 1.0772,
+      "step": 4687
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015818563328720033,
+      "loss": 1.2468,
+      "step": 4688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015817670431715703,
+      "loss": 1.2345,
+      "step": 4689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001581677753471137,
+      "loss": 1.0922,
+      "step": 4690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001581588463770704,
+      "loss": 1.0049,
+      "step": 4691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015814991740702712,
+      "loss": 1.0754,
+      "step": 4692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001581409884369838,
+      "loss": 0.9219,
+      "step": 4693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001581320594669405,
+      "loss": 1.2148,
+      "step": 4694
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001581231304968972,
+      "loss": 1.0825,
+      "step": 4695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015811420152685388,
+      "loss": 1.0562,
+      "step": 4696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015810527255681058,
+      "loss": 1.0701,
+      "step": 4697
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015809634358676729,
+      "loss": 1.0306,
+      "step": 4698
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015808741461672396,
+      "loss": 1.0552,
+      "step": 4699
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015807848564668067,
+      "loss": 1.0332,
+      "step": 4700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015806955667663737,
+      "loss": 1.0287,
+      "step": 4701
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015806062770659405,
+      "loss": 1.1148,
+      "step": 4702
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015805169873655075,
+      "loss": 0.9398,
+      "step": 4703
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015804276976650746,
+      "loss": 1.0298,
+      "step": 4704
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015803384079646413,
+      "loss": 1.0226,
+      "step": 4705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015802491182642084,
+      "loss": 1.0487,
+      "step": 4706
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015801598285637754,
+      "loss": 0.9479,
+      "step": 4707
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015800705388633422,
+      "loss": 1.097,
+      "step": 4708
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015799812491629092,
+      "loss": 0.9327,
+      "step": 4709
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015798919594624763,
+      "loss": 1.1194,
+      "step": 4710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001579802669762043,
+      "loss": 1.1142,
+      "step": 4711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000157971338006161,
+      "loss": 1.0331,
+      "step": 4712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001579624090361177,
+      "loss": 0.9236,
+      "step": 4713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001579534800660744,
+      "loss": 1.0688,
+      "step": 4714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001579445510960311,
+      "loss": 0.9885,
+      "step": 4715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015793562212598777,
+      "loss": 1.2653,
+      "step": 4716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015792669315594447,
+      "loss": 1.0941,
+      "step": 4717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015791776418590115,
+      "loss": 0.9204,
+      "step": 4718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015790883521585785,
+      "loss": 0.8998,
+      "step": 4719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015789990624581456,
+      "loss": 1.0996,
+      "step": 4720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015789097727577123,
+      "loss": 1.1027,
+      "step": 4721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015788204830572794,
+      "loss": 1.2027,
+      "step": 4722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015787311933568462,
+      "loss": 1.0597,
+      "step": 4723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015786419036564132,
+      "loss": 1.0373,
+      "step": 4724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015785526139559802,
+      "loss": 1.0274,
+      "step": 4725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001578463324255547,
+      "loss": 1.0631,
+      "step": 4726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001578374034555114,
+      "loss": 0.9842,
+      "step": 4727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001578284744854681,
+      "loss": 1.0416,
+      "step": 4728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015781954551542479,
+      "loss": 0.9398,
+      "step": 4729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001578106165453815,
+      "loss": 1.1752,
+      "step": 4730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001578016875753382,
+      "loss": 0.9883,
+      "step": 4731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015779275860529487,
+      "loss": 1.0532,
+      "step": 4732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015778382963525157,
+      "loss": 0.9849,
+      "step": 4733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015777490066520828,
+      "loss": 1.1227,
+      "step": 4734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015776597169516496,
+      "loss": 0.9528,
+      "step": 4735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015775704272512166,
+      "loss": 1.1033,
+      "step": 4736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015774811375507836,
+      "loss": 1.2359,
+      "step": 4737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015773918478503504,
+      "loss": 0.9907,
+      "step": 4738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015773025581499174,
+      "loss": 1.1684,
+      "step": 4739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015772132684494845,
+      "loss": 1.1088,
+      "step": 4740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015771239787490513,
+      "loss": 1.0148,
+      "step": 4741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00015770346890486183,
+      "loss": 1.1425,
+      "step": 4742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015769453993481853,
+      "loss": 1.0882,
+      "step": 4743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001576856109647752,
+      "loss": 0.9715,
+      "step": 4744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015767668199473191,
+      "loss": 1.0254,
+      "step": 4745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015766775302468862,
+      "loss": 1.0484,
+      "step": 4746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001576588240546453,
+      "loss": 1.0699,
+      "step": 4747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000157649895084602,
+      "loss": 1.0282,
+      "step": 4748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001576409661145587,
+      "loss": 1.1019,
+      "step": 4749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001576320371445154,
+      "loss": 1.0161,
+      "step": 4750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015762310817447208,
+      "loss": 0.9607,
+      "step": 4751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001576141792044288,
+      "loss": 1.1469,
+      "step": 4752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001576052502343855,
+      "loss": 0.9971,
+      "step": 4753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015759632126434217,
+      "loss": 1.0665,
+      "step": 4754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015758739229429887,
+      "loss": 1.118,
+      "step": 4755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015757846332425558,
+      "loss": 1.0368,
+      "step": 4756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015756953435421225,
+      "loss": 1.0354,
+      "step": 4757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015756060538416896,
+      "loss": 1.0214,
+      "step": 4758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015755167641412564,
+      "loss": 0.946,
+      "step": 4759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015754274744408234,
+      "loss": 1.0553,
+      "step": 4760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015753381847403904,
+      "loss": 0.9876,
+      "step": 4761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015752488950399572,
+      "loss": 1.0771,
+      "step": 4762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015751596053395242,
+      "loss": 0.8581,
+      "step": 4763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001575070315639091,
+      "loss": 1.0744,
+      "step": 4764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001574981025938658,
+      "loss": 0.9733,
+      "step": 4765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015748917362382248,
+      "loss": 1.1018,
+      "step": 4766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015748024465377919,
+      "loss": 1.0091,
+      "step": 4767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001574713156837359,
+      "loss": 1.0453,
+      "step": 4768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015746238671369257,
+      "loss": 1.0015,
+      "step": 4769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015745345774364927,
+      "loss": 0.9368,
+      "step": 4770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015744452877360598,
+      "loss": 0.9887,
+      "step": 4771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015743559980356265,
+      "loss": 1.0144,
+      "step": 4772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015742667083351936,
+      "loss": 1.076,
+      "step": 4773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015741774186347606,
+      "loss": 1.0283,
+      "step": 4774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015740881289343274,
+      "loss": 1.0887,
+      "step": 4775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015739988392338944,
+      "loss": 1.0782,
+      "step": 4776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015739095495334615,
+      "loss": 1.1048,
+      "step": 4777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015738202598330282,
+      "loss": 0.969,
+      "step": 4778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015737309701325953,
+      "loss": 1.0803,
+      "step": 4779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015736416804321623,
+      "loss": 1.0951,
+      "step": 4780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001573552390731729,
+      "loss": 1.0511,
+      "step": 4781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001573463101031296,
+      "loss": 0.9733,
+      "step": 4782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015733738113308632,
+      "loss": 1.0219,
+      "step": 4783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000157328452163043,
+      "loss": 1.0442,
+      "step": 4784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001573195231929997,
+      "loss": 1.0817,
+      "step": 4785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001573105942229564,
+      "loss": 1.1472,
+      "step": 4786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015730166525291308,
+      "loss": 1.1398,
+      "step": 4787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015729273628286978,
+      "loss": 1.0136,
+      "step": 4788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015728380731282649,
+      "loss": 0.9605,
+      "step": 4789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015727487834278316,
+      "loss": 1.0179,
+      "step": 4790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015726594937273987,
+      "loss": 1.1434,
+      "step": 4791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015725702040269657,
+      "loss": 1.0333,
+      "step": 4792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015724809143265325,
+      "loss": 1.1123,
+      "step": 4793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015723916246260995,
+      "loss": 1.1702,
+      "step": 4794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015723023349256666,
+      "loss": 1.0529,
+      "step": 4795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015722130452252333,
+      "loss": 1.1212,
+      "step": 4796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015721237555248004,
+      "loss": 1.1648,
+      "step": 4797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015720344658243674,
+      "loss": 1.0782,
+      "step": 4798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015719451761239342,
+      "loss": 1.0159,
+      "step": 4799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015718558864235012,
+      "loss": 0.9854,
+      "step": 4800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015717665967230683,
+      "loss": 0.9856,
+      "step": 4801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001571677307022635,
+      "loss": 1.0698,
+      "step": 4802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001571588017322202,
+      "loss": 0.999,
+      "step": 4803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001571498727621769,
+      "loss": 1.0832,
+      "step": 4804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001571409437921336,
+      "loss": 1.1319,
+      "step": 4805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001571320148220903,
+      "loss": 1.0949,
+      "step": 4806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015712308585204697,
+      "loss": 1.3125,
+      "step": 4807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015711415688200367,
+      "loss": 1.0558,
+      "step": 4808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015710522791196035,
+      "loss": 1.0742,
+      "step": 4809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015709629894191705,
+      "loss": 0.9748,
+      "step": 4810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015708736997187376,
+      "loss": 1.4916,
+      "step": 4811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015707844100183043,
+      "loss": 1.0349,
+      "step": 4812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015706951203178714,
+      "loss": 0.9216,
+      "step": 4813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015706058306174381,
+      "loss": 0.9537,
+      "step": 4814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015705165409170052,
+      "loss": 1.0627,
+      "step": 4815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00015704272512165722,
+      "loss": 1.0496,
+      "step": 4816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001570337961516139,
+      "loss": 0.9219,
+      "step": 4817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001570248671815706,
+      "loss": 1.0429,
+      "step": 4818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001570159382115273,
+      "loss": 0.9828,
+      "step": 4819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015700700924148398,
+      "loss": 1.0249,
+      "step": 4820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001569980802714407,
+      "loss": 1.0766,
+      "step": 4821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001569891513013974,
+      "loss": 0.9958,
+      "step": 4822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015698022233135407,
+      "loss": 1.0112,
+      "step": 4823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015697129336131077,
+      "loss": 0.9476,
+      "step": 4824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015696236439126748,
+      "loss": 1.045,
+      "step": 4825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015695343542122415,
+      "loss": 1.1391,
+      "step": 4826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015694450645118086,
+      "loss": 1.0329,
+      "step": 4827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015693557748113756,
+      "loss": 1.0715,
+      "step": 4828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015692664851109424,
+      "loss": 1.0693,
+      "step": 4829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015691771954105094,
+      "loss": 1.0922,
+      "step": 4830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015690879057100765,
+      "loss": 0.9365,
+      "step": 4831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015689986160096432,
+      "loss": 1.1089,
+      "step": 4832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015689093263092103,
+      "loss": 1.0581,
+      "step": 4833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015688200366087773,
+      "loss": 1.0949,
+      "step": 4834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001568730746908344,
+      "loss": 0.9141,
+      "step": 4835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015686414572079111,
+      "loss": 0.9919,
+      "step": 4836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015685521675074782,
+      "loss": 0.9751,
+      "step": 4837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001568462877807045,
+      "loss": 0.9118,
+      "step": 4838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001568373588106612,
+      "loss": 1.0627,
+      "step": 4839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001568284298406179,
+      "loss": 1.0553,
+      "step": 4840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015681950087057458,
+      "loss": 1.071,
+      "step": 4841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015681057190053128,
+      "loss": 1.0006,
+      "step": 4842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000156801642930488,
+      "loss": 1.0548,
+      "step": 4843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015679271396044466,
+      "loss": 1.1769,
+      "step": 4844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015678378499040137,
+      "loss": 0.8939,
+      "step": 4845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015677485602035807,
+      "loss": 1.0506,
+      "step": 4846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015676592705031478,
+      "loss": 1.0296,
+      "step": 4847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015675699808027145,
+      "loss": 1.1693,
+      "step": 4848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015674806911022816,
+      "loss": 1.0548,
+      "step": 4849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015673914014018483,
+      "loss": 1.0413,
+      "step": 4850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015673021117014154,
+      "loss": 1.0263,
+      "step": 4851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015672128220009824,
+      "loss": 0.8993,
+      "step": 4852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015671235323005492,
+      "loss": 1.0754,
+      "step": 4853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015670342426001162,
+      "loss": 1.0018,
+      "step": 4854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001566944952899683,
+      "loss": 1.0717,
+      "step": 4855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000156685566319925,
+      "loss": 0.9865,
+      "step": 4856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015667663734988168,
+      "loss": 0.9174,
+      "step": 4857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015666770837983839,
+      "loss": 1.0008,
+      "step": 4858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001566587794097951,
+      "loss": 1.0951,
+      "step": 4859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015664985043975177,
+      "loss": 1.2507,
+      "step": 4860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015664092146970847,
+      "loss": 1.11,
+      "step": 4861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015663199249966517,
+      "loss": 1.0877,
+      "step": 4862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015662306352962185,
+      "loss": 1.1401,
+      "step": 4863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015661413455957856,
+      "loss": 0.9945,
+      "step": 4864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015660520558953526,
+      "loss": 1.0809,
+      "step": 4865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015659627661949194,
+      "loss": 0.9746,
+      "step": 4866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015658734764944864,
+      "loss": 1.1445,
+      "step": 4867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015657841867940534,
+      "loss": 1.0377,
+      "step": 4868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015656948970936202,
+      "loss": 1.0449,
+      "step": 4869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015656056073931873,
+      "loss": 1.1308,
+      "step": 4870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015655163176927543,
+      "loss": 0.9131,
+      "step": 4871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001565427027992321,
+      "loss": 1.0316,
+      "step": 4872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001565337738291888,
+      "loss": 1.1027,
+      "step": 4873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015652484485914551,
+      "loss": 1.0276,
+      "step": 4874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001565159158891022,
+      "loss": 0.8757,
+      "step": 4875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001565069869190589,
+      "loss": 1.0418,
+      "step": 4876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001564980579490156,
+      "loss": 1.1037,
+      "step": 4877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015648912897897228,
+      "loss": 1.0283,
+      "step": 4878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015648020000892898,
+      "loss": 1.0089,
+      "step": 4879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015647127103888568,
+      "loss": 1.1015,
+      "step": 4880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015646234206884236,
+      "loss": 1.1056,
+      "step": 4881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015645341309879907,
+      "loss": 1.0946,
+      "step": 4882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015644448412875577,
+      "loss": 1.1846,
+      "step": 4883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015643555515871245,
+      "loss": 0.9505,
+      "step": 4884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015642662618866915,
+      "loss": 1.1091,
+      "step": 4885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015641769721862585,
+      "loss": 1.1568,
+      "step": 4886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015640876824858253,
+      "loss": 0.96,
+      "step": 4887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015639983927853924,
+      "loss": 0.9711,
+      "step": 4888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015639091030849594,
+      "loss": 1.04,
+      "step": 4889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015638198133845262,
+      "loss": 1.032,
+      "step": 4890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00015637305236840932,
+      "loss": 1.0717,
+      "step": 4891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015636412339836602,
+      "loss": 0.9679,
+      "step": 4892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001563551944283227,
+      "loss": 1.0796,
+      "step": 4893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001563462654582794,
+      "loss": 1.1367,
+      "step": 4894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001563373364882361,
+      "loss": 0.9514,
+      "step": 4895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001563284075181928,
+      "loss": 1.0258,
+      "step": 4896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001563194785481495,
+      "loss": 1.0132,
+      "step": 4897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015631054957810617,
+      "loss": 1.033,
+      "step": 4898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015630162060806287,
+      "loss": 0.9951,
+      "step": 4899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015629269163801955,
+      "loss": 0.9821,
+      "step": 4900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015628376266797625,
+      "loss": 1.0094,
+      "step": 4901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015627483369793296,
+      "loss": 1.0043,
+      "step": 4902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015626590472788963,
+      "loss": 0.9987,
+      "step": 4903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015625697575784634,
+      "loss": 1.0774,
+      "step": 4904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015624804678780301,
+      "loss": 1.0174,
+      "step": 4905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015623911781775972,
+      "loss": 1.163,
+      "step": 4906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015623018884771642,
+      "loss": 1.1237,
+      "step": 4907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001562212598776731,
+      "loss": 0.9886,
+      "step": 4908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001562123309076298,
+      "loss": 1.1854,
+      "step": 4909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001562034019375865,
+      "loss": 1.1004,
+      "step": 4910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015619447296754318,
+      "loss": 1.0461,
+      "step": 4911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001561855439974999,
+      "loss": 1.0508,
+      "step": 4912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001561766150274566,
+      "loss": 1.0654,
+      "step": 4913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015616768605741327,
+      "loss": 1.1468,
+      "step": 4914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015615875708736997,
+      "loss": 1.0375,
+      "step": 4915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015614982811732668,
+      "loss": 1.0096,
+      "step": 4916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015614089914728335,
+      "loss": 0.9354,
+      "step": 4917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015613197017724006,
+      "loss": 1.122,
+      "step": 4918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015612304120719676,
+      "loss": 1.1349,
+      "step": 4919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015611411223715344,
+      "loss": 0.9706,
+      "step": 4920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015610518326711014,
+      "loss": 1.029,
+      "step": 4921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015609625429706685,
+      "loss": 1.0834,
+      "step": 4922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015608732532702352,
+      "loss": 1.0678,
+      "step": 4923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015607839635698023,
+      "loss": 1.0686,
+      "step": 4924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015606946738693693,
+      "loss": 0.9158,
+      "step": 4925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001560605384168936,
+      "loss": 0.9544,
+      "step": 4926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001560516094468503,
+      "loss": 1.0978,
+      "step": 4927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015604268047680702,
+      "loss": 1.0903,
+      "step": 4928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001560337515067637,
+      "loss": 0.9764,
+      "step": 4929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001560248225367204,
+      "loss": 0.9886,
+      "step": 4930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001560158935666771,
+      "loss": 0.9791,
+      "step": 4931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015600696459663378,
+      "loss": 1.0008,
+      "step": 4932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015599803562659048,
+      "loss": 0.9829,
+      "step": 4933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001559891066565472,
+      "loss": 0.9634,
+      "step": 4934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015598017768650386,
+      "loss": 1.3652,
+      "step": 4935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015597124871646057,
+      "loss": 1.0031,
+      "step": 4936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015596231974641727,
+      "loss": 1.033,
+      "step": 4937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015595339077637395,
+      "loss": 1.0473,
+      "step": 4938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015594446180633065,
+      "loss": 1.0192,
+      "step": 4939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015593553283628736,
+      "loss": 1.0315,
+      "step": 4940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015592660386624403,
+      "loss": 1.0287,
+      "step": 4941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015591767489620074,
+      "loss": 1.039,
+      "step": 4942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015590874592615744,
+      "loss": 1.0424,
+      "step": 4943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015589981695611412,
+      "loss": 1.1141,
+      "step": 4944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015589088798607082,
+      "loss": 1.1092,
+      "step": 4945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001558819590160275,
+      "loss": 1.0768,
+      "step": 4946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001558730300459842,
+      "loss": 1.1045,
+      "step": 4947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015586410107594088,
+      "loss": 0.9459,
+      "step": 4948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015585517210589759,
+      "loss": 0.9516,
+      "step": 4949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001558462431358543,
+      "loss": 0.9834,
+      "step": 4950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015583731416581097,
+      "loss": 1.0366,
+      "step": 4951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015582838519576767,
+      "loss": 1.1437,
+      "step": 4952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015581945622572437,
+      "loss": 0.9647,
+      "step": 4953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015581052725568105,
+      "loss": 1.0864,
+      "step": 4954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015580159828563776,
+      "loss": 1.1527,
+      "step": 4955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015579266931559446,
+      "loss": 1.1955,
+      "step": 4956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015578374034555114,
+      "loss": 0.9888,
+      "step": 4957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015577481137550784,
+      "loss": 1.0622,
+      "step": 4958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015576588240546454,
+      "loss": 1.022,
+      "step": 4959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015575695343542122,
+      "loss": 1.1267,
+      "step": 4960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015574802446537793,
+      "loss": 1.098,
+      "step": 4961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015573909549533463,
+      "loss": 0.9608,
+      "step": 4962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001557301665252913,
+      "loss": 1.0322,
+      "step": 4963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000155721237555248,
+      "loss": 1.0253,
+      "step": 4964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015571230858520471,
+      "loss": 1.0911,
+      "step": 4965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001557033796151614,
+      "loss": 0.9395,
+      "step": 4966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001556944506451181,
+      "loss": 0.9746,
+      "step": 4967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001556855216750748,
+      "loss": 1.1317,
+      "step": 4968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015567659270503148,
+      "loss": 1.0971,
+      "step": 4969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015566766373498818,
+      "loss": 0.9833,
+      "step": 4970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015565873476494488,
+      "loss": 1.0365,
+      "step": 4971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015564980579490156,
+      "loss": 1.1458,
+      "step": 4972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015564087682485827,
+      "loss": 1.2185,
+      "step": 4973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015563194785481497,
+      "loss": 1.0293,
+      "step": 4974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015562301888477165,
+      "loss": 1.0029,
+      "step": 4975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015561408991472835,
+      "loss": 1.0484,
+      "step": 4976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015560516094468505,
+      "loss": 1.059,
+      "step": 4977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015559623197464173,
+      "loss": 1.0035,
+      "step": 4978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015558730300459844,
+      "loss": 0.9852,
+      "step": 4979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015557837403455514,
+      "loss": 0.9432,
+      "step": 4980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015556944506451182,
+      "loss": 1.0092,
+      "step": 4981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015556051609446852,
+      "loss": 1.1229,
+      "step": 4982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015555158712442522,
+      "loss": 0.9457,
+      "step": 4983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001555426581543819,
+      "loss": 0.9952,
+      "step": 4984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001555337291843386,
+      "loss": 0.9027,
+      "step": 4985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001555248002142953,
+      "loss": 0.9692,
+      "step": 4986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015551587124425199,
+      "loss": 1.0781,
+      "step": 4987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001555069422742087,
+      "loss": 1.0478,
+      "step": 4988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015549801330416537,
+      "loss": 1.0732,
+      "step": 4989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015548908433412207,
+      "loss": 1.0636,
+      "step": 4990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015548015536407875,
+      "loss": 0.9468,
+      "step": 4991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015547122639403545,
+      "loss": 1.0104,
+      "step": 4992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015546229742399213,
+      "loss": 1.0542,
+      "step": 4993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015545336845394883,
+      "loss": 1.0199,
+      "step": 4994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015544443948390554,
+      "loss": 0.9984,
+      "step": 4995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015543551051386221,
+      "loss": 1.089,
+      "step": 4996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015542658154381892,
+      "loss": 0.8574,
+      "step": 4997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015541765257377562,
+      "loss": 0.9751,
+      "step": 4998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001554087236037323,
+      "loss": 1.0636,
+      "step": 4999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000155399794633689,
+      "loss": 1.1802,
+      "step": 5000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001553908656636457,
+      "loss": 1.0881,
+      "step": 5001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015538193669360238,
+      "loss": 1.0248,
+      "step": 5002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001553730077235591,
+      "loss": 1.0745,
+      "step": 5003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001553640787535158,
+      "loss": 0.931,
+      "step": 5004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015535514978347247,
+      "loss": 1.0041,
+      "step": 5005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015534622081342917,
+      "loss": 1.0324,
+      "step": 5006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015533729184338588,
+      "loss": 1.1004,
+      "step": 5007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015532836287334255,
+      "loss": 0.9357,
+      "step": 5008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015531943390329926,
+      "loss": 0.9744,
+      "step": 5009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015531050493325596,
+      "loss": 0.9727,
+      "step": 5010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015530157596321264,
+      "loss": 1.1121,
+      "step": 5011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015529264699316934,
+      "loss": 1.0739,
+      "step": 5012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015528371802312605,
+      "loss": 0.9888,
+      "step": 5013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015527478905308272,
+      "loss": 0.9722,
+      "step": 5014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015526586008303943,
+      "loss": 1.074,
+      "step": 5015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015525693111299613,
+      "loss": 1.1168,
+      "step": 5016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001552480021429528,
+      "loss": 1.1752,
+      "step": 5017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001552390731729095,
+      "loss": 1.045,
+      "step": 5018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015523014420286622,
+      "loss": 1.0196,
+      "step": 5019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001552212152328229,
+      "loss": 1.1752,
+      "step": 5020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001552122862627796,
+      "loss": 1.4875,
+      "step": 5021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001552033572927363,
+      "loss": 0.9052,
+      "step": 5022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015519442832269298,
+      "loss": 1.0637,
+      "step": 5023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015518549935264968,
+      "loss": 0.9149,
+      "step": 5024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001551765703826064,
+      "loss": 1.0763,
+      "step": 5025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015516764141256306,
+      "loss": 1.1934,
+      "step": 5026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015515871244251977,
+      "loss": 1.0894,
+      "step": 5027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015514978347247647,
+      "loss": 1.0829,
+      "step": 5028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015514085450243315,
+      "loss": 1.162,
+      "step": 5029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015513192553238985,
+      "loss": 0.9388,
+      "step": 5030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015512299656234656,
+      "loss": 1.0296,
+      "step": 5031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015511406759230323,
+      "loss": 1.0642,
+      "step": 5032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015510513862225994,
+      "loss": 0.92,
+      "step": 5033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015509620965221664,
+      "loss": 1.038,
+      "step": 5034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015508728068217332,
+      "loss": 1.018,
+      "step": 5035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015507835171213002,
+      "loss": 0.9951,
+      "step": 5036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001550694227420867,
+      "loss": 1.0805,
+      "step": 5037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001550604937720434,
+      "loss": 1.1157,
+      "step": 5038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015505156480200008,
+      "loss": 0.9635,
+      "step": 5039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015504263583195678,
+      "loss": 1.12,
+      "step": 5040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015503370686191346,
+      "loss": 1.0755,
+      "step": 5041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015502477789187017,
+      "loss": 0.9975,
+      "step": 5042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015501584892182687,
+      "loss": 1.1059,
+      "step": 5043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015500691995178357,
+      "loss": 1.1817,
+      "step": 5044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015499799098174025,
+      "loss": 1.1305,
+      "step": 5045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015498906201169695,
+      "loss": 0.9606,
+      "step": 5046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015498013304165366,
+      "loss": 1.1885,
+      "step": 5047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015497120407161034,
+      "loss": 0.9163,
+      "step": 5048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015496227510156704,
+      "loss": 1.1074,
+      "step": 5049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015495334613152374,
+      "loss": 0.9563,
+      "step": 5050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015494441716148042,
+      "loss": 1.0251,
+      "step": 5051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015493548819143712,
+      "loss": 0.928,
+      "step": 5052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015492655922139383,
+      "loss": 1.059,
+      "step": 5053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001549176302513505,
+      "loss": 1.1348,
+      "step": 5054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001549087012813072,
+      "loss": 0.9181,
+      "step": 5055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015489977231126391,
+      "loss": 0.9364,
+      "step": 5056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001548908433412206,
+      "loss": 1.0014,
+      "step": 5057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001548819143711773,
+      "loss": 1.063,
+      "step": 5058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000154872985401134,
+      "loss": 1.0353,
+      "step": 5059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015486405643109068,
+      "loss": 1.0502,
+      "step": 5060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015485512746104738,
+      "loss": 1.0252,
+      "step": 5061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015484619849100408,
+      "loss": 0.9283,
+      "step": 5062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015483726952096076,
+      "loss": 0.9852,
+      "step": 5063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015482834055091746,
+      "loss": 1.0159,
+      "step": 5064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015481941158087417,
+      "loss": 0.982,
+      "step": 5065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015481048261083085,
+      "loss": 0.9757,
+      "step": 5066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015480155364078755,
+      "loss": 0.9822,
+      "step": 5067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015479262467074425,
+      "loss": 1.1419,
+      "step": 5068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015478369570070093,
+      "loss": 1.0418,
+      "step": 5069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015477476673065763,
+      "loss": 1.0105,
+      "step": 5070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015476583776061434,
+      "loss": 0.9833,
+      "step": 5071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015475690879057102,
+      "loss": 1.0768,
+      "step": 5072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015474797982052772,
+      "loss": 0.9509,
+      "step": 5073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015473905085048442,
+      "loss": 1.0696,
+      "step": 5074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001547301218804411,
+      "loss": 1.064,
+      "step": 5075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001547211929103978,
+      "loss": 1.1064,
+      "step": 5076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001547122639403545,
+      "loss": 0.9849,
+      "step": 5077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015470333497031119,
+      "loss": 0.9693,
+      "step": 5078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001546944060002679,
+      "loss": 1.1403,
+      "step": 5079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015468547703022457,
+      "loss": 0.9867,
+      "step": 5080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015467654806018127,
+      "loss": 0.8654,
+      "step": 5081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015466761909013795,
+      "loss": 0.9701,
+      "step": 5082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015465869012009465,
+      "loss": 1.1056,
+      "step": 5083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015464976115005133,
+      "loss": 1.1452,
+      "step": 5084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015464083218000803,
+      "loss": 1.1672,
+      "step": 5085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015463190320996474,
+      "loss": 0.9816,
+      "step": 5086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015462297423992141,
+      "loss": 1.0778,
+      "step": 5087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015461404526987812,
+      "loss": 1.0574,
+      "step": 5088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015460511629983482,
+      "loss": 1.151,
+      "step": 5089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001545961873297915,
+      "loss": 0.956,
+      "step": 5090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001545872583597482,
+      "loss": 0.9967,
+      "step": 5091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001545783293897049,
+      "loss": 1.1828,
+      "step": 5092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015456940041966158,
+      "loss": 1.1096,
+      "step": 5093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001545604714496183,
+      "loss": 1.0014,
+      "step": 5094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000154551542479575,
+      "loss": 1.012,
+      "step": 5095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015454261350953167,
+      "loss": 1.0409,
+      "step": 5096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015453368453948837,
+      "loss": 0.9167,
+      "step": 5097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015452475556944508,
+      "loss": 1.1691,
+      "step": 5098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015451582659940175,
+      "loss": 0.9982,
+      "step": 5099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015450689762935846,
+      "loss": 1.0577,
+      "step": 5100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015449796865931516,
+      "loss": 0.9339,
+      "step": 5101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015448903968927184,
+      "loss": 0.966,
+      "step": 5102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015448011071922854,
+      "loss": 1.0923,
+      "step": 5103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015447118174918525,
+      "loss": 1.117,
+      "step": 5104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015446225277914192,
+      "loss": 1.0814,
+      "step": 5105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015445332380909863,
+      "loss": 1.022,
+      "step": 5106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015444439483905533,
+      "loss": 0.9982,
+      "step": 5107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000154435465869012,
+      "loss": 1.183,
+      "step": 5108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001544265368989687,
+      "loss": 1.0943,
+      "step": 5109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015441760792892542,
+      "loss": 0.9406,
+      "step": 5110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001544086789588821,
+      "loss": 1.0254,
+      "step": 5111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001543997499888388,
+      "loss": 1.0074,
+      "step": 5112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001543908210187955,
+      "loss": 1.0038,
+      "step": 5113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015438189204875218,
+      "loss": 1.1617,
+      "step": 5114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015437296307870888,
+      "loss": 1.0936,
+      "step": 5115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001543640341086656,
+      "loss": 1.1058,
+      "step": 5116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015435510513862226,
+      "loss": 0.972,
+      "step": 5117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015434617616857897,
+      "loss": 1.0475,
+      "step": 5118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015433724719853567,
+      "loss": 0.9964,
+      "step": 5119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015432831822849235,
+      "loss": 1.0938,
+      "step": 5120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015431938925844905,
+      "loss": 1.097,
+      "step": 5121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015431046028840576,
+      "loss": 1.0653,
+      "step": 5122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015430153131836243,
+      "loss": 1.0181,
+      "step": 5123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015429260234831914,
+      "loss": 1.0516,
+      "step": 5124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015428367337827581,
+      "loss": 0.9989,
+      "step": 5125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015427474440823252,
+      "loss": 0.9367,
+      "step": 5126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015426581543818922,
+      "loss": 0.9993,
+      "step": 5127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001542568864681459,
+      "loss": 1.0148,
+      "step": 5128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001542479574981026,
+      "loss": 1.0809,
+      "step": 5129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015423902852805928,
+      "loss": 1.0285,
+      "step": 5130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015423009955801598,
+      "loss": 0.9451,
+      "step": 5131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015422117058797266,
+      "loss": 1.0945,
+      "step": 5132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015421224161792937,
+      "loss": 1.0387,
+      "step": 5133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015420331264788607,
+      "loss": 1.0181,
+      "step": 5134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015419438367784275,
+      "loss": 1.146,
+      "step": 5135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015418545470779945,
+      "loss": 1.0032,
+      "step": 5136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015417652573775615,
+      "loss": 0.9863,
+      "step": 5137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015416759676771283,
+      "loss": 0.9969,
+      "step": 5138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015415866779766954,
+      "loss": 1.0098,
+      "step": 5139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015414973882762624,
+      "loss": 1.1583,
+      "step": 5140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015414080985758294,
+      "loss": 1.0427,
+      "step": 5141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015413188088753962,
+      "loss": 1.1673,
+      "step": 5142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015412295191749632,
+      "loss": 1.1509,
+      "step": 5143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015411402294745303,
+      "loss": 1.1468,
+      "step": 5144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001541050939774097,
+      "loss": 0.9997,
+      "step": 5145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001540961650073664,
+      "loss": 1.105,
+      "step": 5146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015408723603732311,
+      "loss": 0.9688,
+      "step": 5147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001540783070672798,
+      "loss": 1.0338,
+      "step": 5148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001540693780972365,
+      "loss": 1.0181,
+      "step": 5149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001540604491271932,
+      "loss": 0.9648,
+      "step": 5150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015405152015714988,
+      "loss": 0.9711,
+      "step": 5151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015404259118710658,
+      "loss": 1.0341,
+      "step": 5152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015403366221706328,
+      "loss": 0.963,
+      "step": 5153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015402473324701996,
+      "loss": 1.1415,
+      "step": 5154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015401580427697666,
+      "loss": 0.9921,
+      "step": 5155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015400687530693337,
+      "loss": 1.0338,
+      "step": 5156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015399794633689005,
+      "loss": 0.9985,
+      "step": 5157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015398901736684675,
+      "loss": 1.1006,
+      "step": 5158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015398008839680345,
+      "loss": 0.9139,
+      "step": 5159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015397115942676013,
+      "loss": 1.0341,
+      "step": 5160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015396223045671683,
+      "loss": 1.0439,
+      "step": 5161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015395330148667354,
+      "loss": 0.9588,
+      "step": 5162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015394437251663022,
+      "loss": 0.9799,
+      "step": 5163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015393544354658692,
+      "loss": 1.1223,
+      "step": 5164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015392651457654362,
+      "loss": 1.0631,
+      "step": 5165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001539175856065003,
+      "loss": 0.9978,
+      "step": 5166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000153908656636457,
+      "loss": 0.9555,
+      "step": 5167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001538997276664137,
+      "loss": 1.1875,
+      "step": 5168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015389079869637039,
+      "loss": 0.972,
+      "step": 5169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001538818697263271,
+      "loss": 1.0985,
+      "step": 5170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015387294075628377,
+      "loss": 1.0501,
+      "step": 5171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015386401178624047,
+      "loss": 1.1471,
+      "step": 5172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015385508281619715,
+      "loss": 1.1262,
+      "step": 5173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.9553,
+      "step": 5174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015383722487611053,
+      "loss": 1.0404,
+      "step": 5175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015382829590606723,
+      "loss": 1.1292,
+      "step": 5176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015381936693602394,
+      "loss": 1.0762,
+      "step": 5177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001538104379659806,
+      "loss": 1.0516,
+      "step": 5178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015380150899593732,
+      "loss": 0.9597,
+      "step": 5179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015379258002589402,
+      "loss": 1.1379,
+      "step": 5180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001537836510558507,
+      "loss": 1.046,
+      "step": 5181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001537747220858074,
+      "loss": 1.1279,
+      "step": 5182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001537657931157641,
+      "loss": 0.9501,
+      "step": 5183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015375686414572078,
+      "loss": 1.061,
+      "step": 5184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001537479351756775,
+      "loss": 1.0345,
+      "step": 5185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001537390062056342,
+      "loss": 1.0512,
+      "step": 5186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015373007723559087,
+      "loss": 0.9889,
+      "step": 5187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015372114826554757,
+      "loss": 1.099,
+      "step": 5188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015371221929550428,
+      "loss": 1.0223,
+      "step": 5189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015370329032546095,
+      "loss": 0.971,
+      "step": 5190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015369436135541766,
+      "loss": 0.9433,
+      "step": 5191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015368543238537436,
+      "loss": 1.1317,
+      "step": 5192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015367650341533104,
+      "loss": 1.0751,
+      "step": 5193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015366757444528774,
+      "loss": 1.0264,
+      "step": 5194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015365864547524445,
+      "loss": 0.957,
+      "step": 5195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015364971650520112,
+      "loss": 0.9214,
+      "step": 5196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015364078753515783,
+      "loss": 1.0415,
+      "step": 5197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015363185856511453,
+      "loss": 1.1396,
+      "step": 5198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001536229295950712,
+      "loss": 1.1345,
+      "step": 5199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001536140006250279,
+      "loss": 0.9866,
+      "step": 5200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015360507165498462,
+      "loss": 1.0201,
+      "step": 5201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001535961426849413,
+      "loss": 1.0451,
+      "step": 5202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000153587213714898,
+      "loss": 1.1327,
+      "step": 5203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001535782847448547,
+      "loss": 0.9927,
+      "step": 5204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015356935577481138,
+      "loss": 1.0407,
+      "step": 5205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015356042680476808,
+      "loss": 1.4246,
+      "step": 5206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015355149783472479,
+      "loss": 0.8536,
+      "step": 5207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015354256886468146,
+      "loss": 0.9898,
+      "step": 5208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015353363989463817,
+      "loss": 1.0066,
+      "step": 5209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015352471092459487,
+      "loss": 0.972,
+      "step": 5210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015351578195455155,
+      "loss": 1.0084,
+      "step": 5211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015350685298450825,
+      "loss": 1.1721,
+      "step": 5212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015349792401446496,
+      "loss": 1.0292,
+      "step": 5213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015348899504442163,
+      "loss": 1.4102,
+      "step": 5214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015348006607437834,
+      "loss": 1.0847,
+      "step": 5215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015347113710433501,
+      "loss": 1.22,
+      "step": 5216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015346220813429172,
+      "loss": 0.9906,
+      "step": 5217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015345327916424842,
+      "loss": 1.0056,
+      "step": 5218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001534443501942051,
+      "loss": 0.9667,
+      "step": 5219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001534354212241618,
+      "loss": 0.9867,
+      "step": 5220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015342649225411848,
+      "loss": 1.1008,
+      "step": 5221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015341756328407518,
+      "loss": 1.2199,
+      "step": 5222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015340863431403186,
+      "loss": 1.0555,
+      "step": 5223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015339970534398857,
+      "loss": 1.0922,
+      "step": 5224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015339077637394527,
+      "loss": 1.1308,
+      "step": 5225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015338184740390195,
+      "loss": 1.0127,
+      "step": 5226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015337291843385865,
+      "loss": 0.9145,
+      "step": 5227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015336398946381535,
+      "loss": 1.0215,
+      "step": 5228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015335506049377203,
+      "loss": 1.1137,
+      "step": 5229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015334613152372874,
+      "loss": 1.0799,
+      "step": 5230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015333720255368544,
+      "loss": 1.0655,
+      "step": 5231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015332827358364212,
+      "loss": 1.1209,
+      "step": 5232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015331934461359882,
+      "loss": 1.0108,
+      "step": 5233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015331041564355552,
+      "loss": 1.136,
+      "step": 5234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001533014866735122,
+      "loss": 0.8652,
+      "step": 5235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001532925577034689,
+      "loss": 0.9696,
+      "step": 5236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001532836287334256,
+      "loss": 0.9967,
+      "step": 5237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001532746997633823,
+      "loss": 1.074,
+      "step": 5238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000153265770793339,
+      "loss": 1.1248,
+      "step": 5239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001532568418232957,
+      "loss": 1.0007,
+      "step": 5240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001532479128532524,
+      "loss": 0.9298,
+      "step": 5241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015323898388320908,
+      "loss": 1.001,
+      "step": 5242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015323005491316578,
+      "loss": 1.015,
+      "step": 5243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015322112594312248,
+      "loss": 1.0145,
+      "step": 5244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015321219697307916,
+      "loss": 0.9884,
+      "step": 5245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015320326800303586,
+      "loss": 1.0291,
+      "step": 5246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015319433903299257,
+      "loss": 1.0106,
+      "step": 5247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015318541006294925,
+      "loss": 1.125,
+      "step": 5248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015317648109290595,
+      "loss": 1.3301,
+      "step": 5249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015316755212286265,
+      "loss": 1.1072,
+      "step": 5250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015315862315281933,
+      "loss": 1.0149,
+      "step": 5251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015314969418277603,
+      "loss": 1.0334,
+      "step": 5252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015314076521273274,
+      "loss": 1.0381,
+      "step": 5253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015313183624268942,
+      "loss": 1.1375,
+      "step": 5254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015312290727264612,
+      "loss": 0.9265,
+      "step": 5255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015311397830260282,
+      "loss": 1.0664,
+      "step": 5256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001531050493325595,
+      "loss": 1.1014,
+      "step": 5257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001530961203625162,
+      "loss": 0.9561,
+      "step": 5258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001530871913924729,
+      "loss": 1.1207,
+      "step": 5259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015307826242242959,
+      "loss": 1.0879,
+      "step": 5260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001530693334523863,
+      "loss": 0.9918,
+      "step": 5261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015306040448234297,
+      "loss": 1.0786,
+      "step": 5262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015305147551229967,
+      "loss": 0.9189,
+      "step": 5263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015304254654225635,
+      "loss": 1.0403,
+      "step": 5264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015303361757221305,
+      "loss": 0.969,
+      "step": 5265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015302468860216973,
+      "loss": 0.9657,
+      "step": 5266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015301575963212643,
+      "loss": 1.0314,
+      "step": 5267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015300683066208314,
+      "loss": 1.0349,
+      "step": 5268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001529979016920398,
+      "loss": 1.2046,
+      "step": 5269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015298897272199652,
+      "loss": 1.1527,
+      "step": 5270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015298004375195322,
+      "loss": 0.9589,
+      "step": 5271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001529711147819099,
+      "loss": 1.0277,
+      "step": 5272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001529621858118666,
+      "loss": 1.1514,
+      "step": 5273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001529532568418233,
+      "loss": 1.0736,
+      "step": 5274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015294432787177998,
+      "loss": 1.0474,
+      "step": 5275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001529353989017367,
+      "loss": 1.1666,
+      "step": 5276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001529264699316934,
+      "loss": 0.984,
+      "step": 5277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015291754096165007,
+      "loss": 1.0955,
+      "step": 5278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015290861199160677,
+      "loss": 1.0363,
+      "step": 5279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015289968302156348,
+      "loss": 1.1483,
+      "step": 5280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015289075405152015,
+      "loss": 0.9754,
+      "step": 5281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015288182508147686,
+      "loss": 1.0526,
+      "step": 5282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015287289611143356,
+      "loss": 1.1442,
+      "step": 5283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015286396714139024,
+      "loss": 1.0767,
+      "step": 5284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015285503817134694,
+      "loss": 1.0838,
+      "step": 5285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015284610920130365,
+      "loss": 1.1109,
+      "step": 5286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015283718023126032,
+      "loss": 1.0802,
+      "step": 5287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015282825126121703,
+      "loss": 1.054,
+      "step": 5288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015281932229117373,
+      "loss": 0.9678,
+      "step": 5289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001528103933211304,
+      "loss": 1.2133,
+      "step": 5290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001528014643510871,
+      "loss": 1.1225,
+      "step": 5291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015279253538104382,
+      "loss": 0.9944,
+      "step": 5292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001527836064110005,
+      "loss": 1.0191,
+      "step": 5293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001527746774409572,
+      "loss": 1.0205,
+      "step": 5294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001527657484709139,
+      "loss": 1.0557,
+      "step": 5295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015275681950087058,
+      "loss": 0.9628,
+      "step": 5296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015274789053082728,
+      "loss": 1.0421,
+      "step": 5297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015273896156078399,
+      "loss": 1.0329,
+      "step": 5298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015273003259074066,
+      "loss": 1.0217,
+      "step": 5299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015272110362069737,
+      "loss": 1.1633,
+      "step": 5300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015271217465065407,
+      "loss": 0.984,
+      "step": 5301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015270324568061075,
+      "loss": 1.0383,
+      "step": 5302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015269431671056745,
+      "loss": 1.0029,
+      "step": 5303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015268538774052416,
+      "loss": 0.9015,
+      "step": 5304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015267645877048083,
+      "loss": 1.0508,
+      "step": 5305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015266752980043754,
+      "loss": 0.9537,
+      "step": 5306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015265860083039421,
+      "loss": 1.069,
+      "step": 5307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015264967186035092,
+      "loss": 0.9814,
+      "step": 5308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015264074289030762,
+      "loss": 1.0303,
+      "step": 5309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001526318139202643,
+      "loss": 1.137,
+      "step": 5310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000152622884950221,
+      "loss": 1.0271,
+      "step": 5311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015261395598017768,
+      "loss": 1.0202,
+      "step": 5312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015260502701013438,
+      "loss": 1.1328,
+      "step": 5313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015259609804009106,
+      "loss": 1.1994,
+      "step": 5314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015258716907004776,
+      "loss": 0.9477,
+      "step": 5315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015257824010000447,
+      "loss": 1.04,
+      "step": 5316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015256931112996115,
+      "loss": 1.0769,
+      "step": 5317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015256038215991785,
+      "loss": 1.0924,
+      "step": 5318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015255145318987455,
+      "loss": 0.9353,
+      "step": 5319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015254252421983123,
+      "loss": 1.0865,
+      "step": 5320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015253359524978793,
+      "loss": 0.9446,
+      "step": 5321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015252466627974464,
+      "loss": 1.0437,
+      "step": 5322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015251573730970132,
+      "loss": 1.0761,
+      "step": 5323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015250680833965802,
+      "loss": 1.1067,
+      "step": 5324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015249787936961472,
+      "loss": 1.0312,
+      "step": 5325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001524889503995714,
+      "loss": 1.0585,
+      "step": 5326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001524800214295281,
+      "loss": 1.243,
+      "step": 5327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001524710924594848,
+      "loss": 1.1547,
+      "step": 5328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015246216348944149,
+      "loss": 1.0514,
+      "step": 5329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001524532345193982,
+      "loss": 1.0012,
+      "step": 5330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001524443055493549,
+      "loss": 0.9977,
+      "step": 5331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015243537657931157,
+      "loss": 1.3972,
+      "step": 5332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015242644760926827,
+      "loss": 1.1783,
+      "step": 5333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015241751863922498,
+      "loss": 0.9929,
+      "step": 5334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015240858966918166,
+      "loss": 1.0424,
+      "step": 5335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015239966069913836,
+      "loss": 0.9403,
+      "step": 5336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015239073172909506,
+      "loss": 1.0588,
+      "step": 5337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015238180275905177,
+      "loss": 1.0447,
+      "step": 5338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015237287378900844,
+      "loss": 1.0579,
+      "step": 5339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015236394481896515,
+      "loss": 0.9985,
+      "step": 5340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015235501584892185,
+      "loss": 0.936,
+      "step": 5341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015234608687887853,
+      "loss": 1.0942,
+      "step": 5342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015233715790883523,
+      "loss": 1.0496,
+      "step": 5343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015232822893879194,
+      "loss": 1.077,
+      "step": 5344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015231929996874861,
+      "loss": 0.9887,
+      "step": 5345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015231037099870532,
+      "loss": 1.0437,
+      "step": 5346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015230144202866202,
+      "loss": 1.0979,
+      "step": 5347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001522925130586187,
+      "loss": 1.0283,
+      "step": 5348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001522835840885754,
+      "loss": 1.0901,
+      "step": 5349
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001522746551185321,
+      "loss": 1.1363,
+      "step": 5350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015226572614848878,
+      "loss": 1.0629,
+      "step": 5351
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001522567971784455,
+      "loss": 0.9352,
+      "step": 5352
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015224786820840217,
+      "loss": 0.9322,
+      "step": 5353
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015223893923835887,
+      "loss": 1.0257,
+      "step": 5354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015223001026831555,
+      "loss": 1.0194,
+      "step": 5355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015222108129827225,
+      "loss": 1.0139,
+      "step": 5356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015221215232822893,
+      "loss": 1.1923,
+      "step": 5357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015220322335818563,
+      "loss": 1.0562,
+      "step": 5358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015219429438814234,
+      "loss": 1.1329,
+      "step": 5359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000152185365418099,
+      "loss": 0.9258,
+      "step": 5360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015217643644805572,
+      "loss": 1.0169,
+      "step": 5361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015216750747801242,
+      "loss": 1.0341,
+      "step": 5362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001521585785079691,
+      "loss": 1.0912,
+      "step": 5363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001521496495379258,
+      "loss": 1.1082,
+      "step": 5364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001521407205678825,
+      "loss": 1.0105,
+      "step": 5365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015213179159783918,
+      "loss": 1.0335,
+      "step": 5366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001521228626277959,
+      "loss": 1.031,
+      "step": 5367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001521139336577526,
+      "loss": 0.9381,
+      "step": 5368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015210500468770927,
+      "loss": 0.9627,
+      "step": 5369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015209607571766597,
+      "loss": 1.0183,
+      "step": 5370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015208714674762268,
+      "loss": 1.072,
+      "step": 5371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015207821777757935,
+      "loss": 1.0693,
+      "step": 5372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015206928880753606,
+      "loss": 1.0842,
+      "step": 5373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015206035983749276,
+      "loss": 1.1534,
+      "step": 5374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015205143086744944,
+      "loss": 0.9952,
+      "step": 5375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015204250189740614,
+      "loss": 1.0926,
+      "step": 5376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015203357292736285,
+      "loss": 1.0345,
+      "step": 5377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015202464395731952,
+      "loss": 1.0058,
+      "step": 5378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015201571498727623,
+      "loss": 1.0474,
+      "step": 5379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015200678601723293,
+      "loss": 1.0666,
+      "step": 5380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001519978570471896,
+      "loss": 1.1365,
+      "step": 5381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001519889280771463,
+      "loss": 0.9789,
+      "step": 5382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015197999910710302,
+      "loss": 0.9757,
+      "step": 5383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001519710701370597,
+      "loss": 1.1939,
+      "step": 5384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001519621411670164,
+      "loss": 0.9794,
+      "step": 5385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001519532121969731,
+      "loss": 1.1177,
+      "step": 5386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015194428322692978,
+      "loss": 1.2166,
+      "step": 5387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015193535425688648,
+      "loss": 1.1321,
+      "step": 5388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015192642528684319,
+      "loss": 1.0444,
+      "step": 5389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015191749631679986,
+      "loss": 1.2213,
+      "step": 5390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015190856734675657,
+      "loss": 1.0314,
+      "step": 5391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015189963837671327,
+      "loss": 0.976,
+      "step": 5392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015189070940666995,
+      "loss": 0.9816,
+      "step": 5393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015188178043662665,
+      "loss": 1.0202,
+      "step": 5394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015187285146658336,
+      "loss": 1.0281,
+      "step": 5395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015186392249654003,
+      "loss": 0.9846,
+      "step": 5396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015185499352649674,
+      "loss": 1.0824,
+      "step": 5397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001518460645564534,
+      "loss": 1.0038,
+      "step": 5398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015183713558641012,
+      "loss": 1.0052,
+      "step": 5399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015182820661636682,
+      "loss": 1.0741,
+      "step": 5400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001518192776463235,
+      "loss": 0.9812,
+      "step": 5401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001518103486762802,
+      "loss": 1.0452,
+      "step": 5402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015180141970623688,
+      "loss": 0.9412,
+      "step": 5403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015179249073619358,
+      "loss": 1.0807,
+      "step": 5404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015178356176615026,
+      "loss": 1.1312,
+      "step": 5405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015177463279610696,
+      "loss": 1.0453,
+      "step": 5406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015176570382606367,
+      "loss": 1.0751,
+      "step": 5407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015175677485602035,
+      "loss": 1.1807,
+      "step": 5408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015174784588597705,
+      "loss": 0.9857,
+      "step": 5409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015173891691593375,
+      "loss": 1.0575,
+      "step": 5410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015172998794589043,
+      "loss": 1.0226,
+      "step": 5411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015172105897584713,
+      "loss": 1.0818,
+      "step": 5412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015171213000580384,
+      "loss": 1.0049,
+      "step": 5413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015170320103576052,
+      "loss": 1.0549,
+      "step": 5414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015169427206571722,
+      "loss": 1.1077,
+      "step": 5415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015168534309567392,
+      "loss": 1.1462,
+      "step": 5416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001516764141256306,
+      "loss": 0.9681,
+      "step": 5417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001516674851555873,
+      "loss": 1.112,
+      "step": 5418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000151658556185544,
+      "loss": 0.989,
+      "step": 5419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015164962721550069,
+      "loss": 0.9825,
+      "step": 5420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001516406982454574,
+      "loss": 1.092,
+      "step": 5421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001516317692754141,
+      "loss": 0.9754,
+      "step": 5422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015162284030537077,
+      "loss": 1.0605,
+      "step": 5423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015161391133532747,
+      "loss": 1.0448,
+      "step": 5424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015160498236528418,
+      "loss": 1.1114,
+      "step": 5425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015159605339524086,
+      "loss": 1.0411,
+      "step": 5426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015158712442519756,
+      "loss": 1.0832,
+      "step": 5427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015157819545515426,
+      "loss": 1.142,
+      "step": 5428
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015156926648511094,
+      "loss": 1.027,
+      "step": 5429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015156033751506764,
+      "loss": 1.0741,
+      "step": 5430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015155140854502435,
+      "loss": 1.0654,
+      "step": 5431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015154247957498103,
+      "loss": 0.9157,
+      "step": 5432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015153355060493773,
+      "loss": 0.9996,
+      "step": 5433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015152462163489443,
+      "loss": 1.1572,
+      "step": 5434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015151569266485114,
+      "loss": 1.0348,
+      "step": 5435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015150676369480781,
+      "loss": 1.0301,
+      "step": 5436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015149783472476452,
+      "loss": 0.982,
+      "step": 5437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015148890575472122,
+      "loss": 1.0241,
+      "step": 5438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001514799767846779,
+      "loss": 0.9511,
+      "step": 5439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001514710478146346,
+      "loss": 1.0339,
+      "step": 5440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001514621188445913,
+      "loss": 1.1244,
+      "step": 5441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015145318987454798,
+      "loss": 1.1674,
+      "step": 5442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001514442609045047,
+      "loss": 1.2103,
+      "step": 5443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015143533193446137,
+      "loss": 1.1445,
+      "step": 5444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015142640296441807,
+      "loss": 1.1157,
+      "step": 5445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015141747399437475,
+      "loss": 0.9841,
+      "step": 5446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015140854502433145,
+      "loss": 0.9799,
+      "step": 5447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015139961605428813,
+      "loss": 0.9524,
+      "step": 5448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015139068708424483,
+      "loss": 1.0227,
+      "step": 5449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015138175811420154,
+      "loss": 1.1517,
+      "step": 5450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001513728291441582,
+      "loss": 1.1396,
+      "step": 5451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015136390017411492,
+      "loss": 1.1557,
+      "step": 5452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015135497120407162,
+      "loss": 0.8972,
+      "step": 5453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001513460422340283,
+      "loss": 1.0301,
+      "step": 5454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000151337113263985,
+      "loss": 0.9507,
+      "step": 5455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001513281842939417,
+      "loss": 1.0293,
+      "step": 5456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015131925532389838,
+      "loss": 1.0667,
+      "step": 5457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015131032635385509,
+      "loss": 0.985,
+      "step": 5458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001513013973838118,
+      "loss": 1.0268,
+      "step": 5459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015129246841376847,
+      "loss": 1.003,
+      "step": 5460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015128353944372517,
+      "loss": 1.0476,
+      "step": 5461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015127461047368188,
+      "loss": 1.0406,
+      "step": 5462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015126568150363855,
+      "loss": 1.0759,
+      "step": 5463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015125675253359526,
+      "loss": 1.1082,
+      "step": 5464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015124782356355196,
+      "loss": 1.0061,
+      "step": 5465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015123889459350864,
+      "loss": 1.0976,
+      "step": 5466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015122996562346534,
+      "loss": 1.0428,
+      "step": 5467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015122103665342205,
+      "loss": 1.0157,
+      "step": 5468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015121210768337872,
+      "loss": 1.2625,
+      "step": 5469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015120317871333543,
+      "loss": 1.0236,
+      "step": 5470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015119424974329213,
+      "loss": 1.0134,
+      "step": 5471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001511853207732488,
+      "loss": 0.9405,
+      "step": 5472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001511763918032055,
+      "loss": 1.0243,
+      "step": 5473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015116746283316222,
+      "loss": 1.0491,
+      "step": 5474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001511585338631189,
+      "loss": 1.1418,
+      "step": 5475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001511496048930756,
+      "loss": 0.9562,
+      "step": 5476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001511406759230323,
+      "loss": 1.0585,
+      "step": 5477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015113174695298898,
+      "loss": 1.0648,
+      "step": 5478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015112281798294568,
+      "loss": 1.1451,
+      "step": 5479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015111388901290239,
+      "loss": 1.1116,
+      "step": 5480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015110496004285906,
+      "loss": 1.0267,
+      "step": 5481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015109603107281577,
+      "loss": 0.9293,
+      "step": 5482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015108710210277247,
+      "loss": 0.9854,
+      "step": 5483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015107817313272915,
+      "loss": 1.1329,
+      "step": 5484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015106924416268585,
+      "loss": 1.0054,
+      "step": 5485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015106031519264256,
+      "loss": 0.9783,
+      "step": 5486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015105138622259923,
+      "loss": 0.9872,
+      "step": 5487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015104245725255594,
+      "loss": 1.0961,
+      "step": 5488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001510335282825126,
+      "loss": 0.9858,
+      "step": 5489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015102459931246932,
+      "loss": 1.0978,
+      "step": 5490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015101567034242602,
+      "loss": 1.051,
+      "step": 5491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001510067413723827,
+      "loss": 0.9159,
+      "step": 5492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001509978124023394,
+      "loss": 1.0718,
+      "step": 5493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015098888343229608,
+      "loss": 0.9779,
+      "step": 5494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015097995446225278,
+      "loss": 1.0686,
+      "step": 5495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015097102549220946,
+      "loss": 0.9735,
+      "step": 5496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015096209652216616,
+      "loss": 1.0582,
+      "step": 5497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015095316755212287,
+      "loss": 1.0167,
+      "step": 5498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015094423858207955,
+      "loss": 1.0074,
+      "step": 5499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015093530961203625,
+      "loss": 1.0762,
+      "step": 5500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015092638064199295,
+      "loss": 0.9668,
+      "step": 5501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015091745167194963,
+      "loss": 1.1445,
+      "step": 5502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015090852270190633,
+      "loss": 1.1547,
+      "step": 5503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015089959373186304,
+      "loss": 1.1137,
+      "step": 5504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015089066476181972,
+      "loss": 1.0363,
+      "step": 5505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015088173579177642,
+      "loss": 1.0632,
+      "step": 5506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015087280682173312,
+      "loss": 1.0614,
+      "step": 5507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001508638778516898,
+      "loss": 1.0104,
+      "step": 5508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001508549488816465,
+      "loss": 0.9545,
+      "step": 5509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001508460199116032,
+      "loss": 1.0756,
+      "step": 5510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015083709094155989,
+      "loss": 1.1153,
+      "step": 5511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001508281619715166,
+      "loss": 0.9366,
+      "step": 5512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001508192330014733,
+      "loss": 1.0037,
+      "step": 5513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015081030403142997,
+      "loss": 0.9623,
+      "step": 5514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015080137506138667,
+      "loss": 1.2761,
+      "step": 5515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015079244609134338,
+      "loss": 1.0999,
+      "step": 5516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015078351712130006,
+      "loss": 0.9354,
+      "step": 5517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015077458815125676,
+      "loss": 1.1308,
+      "step": 5518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015076565918121346,
+      "loss": 0.965,
+      "step": 5519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015075673021117014,
+      "loss": 0.9984,
+      "step": 5520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015074780124112684,
+      "loss": 1.0993,
+      "step": 5521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015073887227108355,
+      "loss": 1.0386,
+      "step": 5522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015072994330104022,
+      "loss": 1.1665,
+      "step": 5523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015072101433099693,
+      "loss": 1.0239,
+      "step": 5524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015071208536095363,
+      "loss": 1.0098,
+      "step": 5525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001507031563909103,
+      "loss": 1.0204,
+      "step": 5526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015069422742086701,
+      "loss": 1.0281,
+      "step": 5527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015068529845082372,
+      "loss": 0.9571,
+      "step": 5528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001506763694807804,
+      "loss": 1.0131,
+      "step": 5529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001506674405107371,
+      "loss": 1.0007,
+      "step": 5530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001506585115406938,
+      "loss": 1.1102,
+      "step": 5531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001506495825706505,
+      "loss": 1.0405,
+      "step": 5532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015064065360060718,
+      "loss": 1.1005,
+      "step": 5533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001506317246305639,
+      "loss": 1.2568,
+      "step": 5534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015062279566052056,
+      "loss": 0.9903,
+      "step": 5535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015061386669047727,
+      "loss": 1.0608,
+      "step": 5536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015060493772043395,
+      "loss": 1.1137,
+      "step": 5537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015059600875039065,
+      "loss": 1.1676,
+      "step": 5538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015058707978034733,
+      "loss": 1.0246,
+      "step": 5539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015057815081030403,
+      "loss": 0.9775,
+      "step": 5540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015056922184026073,
+      "loss": 0.9051,
+      "step": 5541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001505602928702174,
+      "loss": 1.0804,
+      "step": 5542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015055136390017412,
+      "loss": 1.0827,
+      "step": 5543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015054243493013082,
+      "loss": 1.1074,
+      "step": 5544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001505335059600875,
+      "loss": 1.0696,
+      "step": 5545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001505245769900442,
+      "loss": 0.9878,
+      "step": 5546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001505156480200009,
+      "loss": 1.0503,
+      "step": 5547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015050671904995758,
+      "loss": 0.9587,
+      "step": 5548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015049779007991429,
+      "loss": 1.0481,
+      "step": 5549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000150488861109871,
+      "loss": 1.1964,
+      "step": 5550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015047993213982767,
+      "loss": 1.1091,
+      "step": 5551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015047100316978437,
+      "loss": 1.187,
+      "step": 5552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015046207419974107,
+      "loss": 1.0728,
+      "step": 5553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015045314522969775,
+      "loss": 1.0638,
+      "step": 5554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015044421625965446,
+      "loss": 1.1883,
+      "step": 5555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015043528728961116,
+      "loss": 1.076,
+      "step": 5556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015042635831956784,
+      "loss": 1.0502,
+      "step": 5557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015041742934952454,
+      "loss": 1.34,
+      "step": 5558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015040850037948124,
+      "loss": 1.0615,
+      "step": 5559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015039957140943792,
+      "loss": 1.1038,
+      "step": 5560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015039064243939463,
+      "loss": 1.1697,
+      "step": 5561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00015038171346935133,
+      "loss": 0.9384,
+      "step": 5562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000150372784499308,
+      "loss": 1.0855,
+      "step": 5563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001503638555292647,
+      "loss": 0.9821,
+      "step": 5564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015035492655922141,
+      "loss": 1.0545,
+      "step": 5565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001503459975891781,
+      "loss": 0.9663,
+      "step": 5566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001503370686191348,
+      "loss": 1.0968,
+      "step": 5567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001503281396490915,
+      "loss": 0.945,
+      "step": 5568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015031921067904818,
+      "loss": 0.9586,
+      "step": 5569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015031028170900488,
+      "loss": 1.0499,
+      "step": 5570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015030135273896158,
+      "loss": 1.0287,
+      "step": 5571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015029242376891826,
+      "loss": 1.149,
+      "step": 5572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015028349479887497,
+      "loss": 1.0657,
+      "step": 5573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015027456582883167,
+      "loss": 0.9899,
+      "step": 5574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015026563685878835,
+      "loss": 1.1182,
+      "step": 5575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015025670788874505,
+      "loss": 1.0667,
+      "step": 5576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015024777891870175,
+      "loss": 1.0694,
+      "step": 5577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015023884994865843,
+      "loss": 0.9671,
+      "step": 5578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015022992097861514,
+      "loss": 1.012,
+      "step": 5579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001502209920085718,
+      "loss": 1.0049,
+      "step": 5580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015021206303852852,
+      "loss": 1.0411,
+      "step": 5581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015020313406848522,
+      "loss": 0.9842,
+      "step": 5582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001501942050984419,
+      "loss": 1.1219,
+      "step": 5583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001501852761283986,
+      "loss": 1.0224,
+      "step": 5584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015017634715835528,
+      "loss": 1.1852,
+      "step": 5585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015016741818831198,
+      "loss": 1.1245,
+      "step": 5586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015015848921826866,
+      "loss": 0.9465,
+      "step": 5587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015014956024822536,
+      "loss": 1.0589,
+      "step": 5588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015014063127818207,
+      "loss": 1.0626,
+      "step": 5589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015013170230813874,
+      "loss": 0.9822,
+      "step": 5590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015012277333809545,
+      "loss": 0.9246,
+      "step": 5591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015011384436805215,
+      "loss": 1.1855,
+      "step": 5592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015010491539800883,
+      "loss": 1.0742,
+      "step": 5593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015009598642796553,
+      "loss": 1.1858,
+      "step": 5594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015008705745792224,
+      "loss": 1.1541,
+      "step": 5595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015007812848787891,
+      "loss": 1.0655,
+      "step": 5596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015006919951783562,
+      "loss": 0.9158,
+      "step": 5597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015006027054779232,
+      "loss": 0.9206,
+      "step": 5598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000150051341577749,
+      "loss": 0.9374,
+      "step": 5599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001500424126077057,
+      "loss": 1.0118,
+      "step": 5600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001500334836376624,
+      "loss": 1.0594,
+      "step": 5601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015002455466761908,
+      "loss": 1.0291,
+      "step": 5602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001500156256975758,
+      "loss": 1.0362,
+      "step": 5603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001500066967275325,
+      "loss": 1.1353,
+      "step": 5604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014999776775748917,
+      "loss": 0.9766,
+      "step": 5605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014998883878744587,
+      "loss": 0.9278,
+      "step": 5606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014997990981740258,
+      "loss": 1.0031,
+      "step": 5607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014997098084735925,
+      "loss": 1.0614,
+      "step": 5608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014996205187731596,
+      "loss": 0.8669,
+      "step": 5609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014995312290727266,
+      "loss": 1.07,
+      "step": 5610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014994419393722934,
+      "loss": 0.9857,
+      "step": 5611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014993526496718604,
+      "loss": 1.0275,
+      "step": 5612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014992633599714275,
+      "loss": 1.0498,
+      "step": 5613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014991740702709942,
+      "loss": 1.1624,
+      "step": 5614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014990847805705613,
+      "loss": 0.9998,
+      "step": 5615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014989954908701283,
+      "loss": 0.9389,
+      "step": 5616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001498906201169695,
+      "loss": 0.982,
+      "step": 5617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014988169114692621,
+      "loss": 1.224,
+      "step": 5618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014987276217688292,
+      "loss": 1.142,
+      "step": 5619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001498638332068396,
+      "loss": 0.9776,
+      "step": 5620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001498549042367963,
+      "loss": 0.9643,
+      "step": 5621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000149845975266753,
+      "loss": 0.9621,
+      "step": 5622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014983704629670968,
+      "loss": 1.0504,
+      "step": 5623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014982811732666638,
+      "loss": 0.9255,
+      "step": 5624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001498191883566231,
+      "loss": 1.1359,
+      "step": 5625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014981025938657976,
+      "loss": 0.9704,
+      "step": 5626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014980133041653647,
+      "loss": 1.005,
+      "step": 5627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014979240144649315,
+      "loss": 0.8953,
+      "step": 5628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014978347247644985,
+      "loss": 1.0181,
+      "step": 5629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014977454350640653,
+      "loss": 1.0054,
+      "step": 5630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014976561453636323,
+      "loss": 0.9923,
+      "step": 5631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014975668556631993,
+      "loss": 1.0344,
+      "step": 5632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001497477565962766,
+      "loss": 1.167,
+      "step": 5633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014973882762623332,
+      "loss": 0.9998,
+      "step": 5634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014972989865619002,
+      "loss": 0.9955,
+      "step": 5635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001497209696861467,
+      "loss": 1.0943,
+      "step": 5636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001497120407161034,
+      "loss": 1.0838,
+      "step": 5637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001497031117460601,
+      "loss": 1.0776,
+      "step": 5638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014969418277601678,
+      "loss": 1.0618,
+      "step": 5639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014968525380597349,
+      "loss": 0.9786,
+      "step": 5640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001496763248359302,
+      "loss": 1.0988,
+      "step": 5641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014966739586588687,
+      "loss": 1.0908,
+      "step": 5642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014965846689584357,
+      "loss": 0.9969,
+      "step": 5643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014964953792580027,
+      "loss": 1.1245,
+      "step": 5644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014964060895575695,
+      "loss": 0.9974,
+      "step": 5645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014963167998571366,
+      "loss": 1.0327,
+      "step": 5646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014962275101567036,
+      "loss": 1.1925,
+      "step": 5647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014961382204562704,
+      "loss": 1.1184,
+      "step": 5648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014960489307558374,
+      "loss": 1.1019,
+      "step": 5649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014959596410554044,
+      "loss": 1.0772,
+      "step": 5650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014958703513549712,
+      "loss": 1.0667,
+      "step": 5651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014957810616545383,
+      "loss": 0.9916,
+      "step": 5652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014956917719541053,
+      "loss": 1.0315,
+      "step": 5653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001495602482253672,
+      "loss": 1.207,
+      "step": 5654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001495513192553239,
+      "loss": 1.0431,
+      "step": 5655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014954239028528061,
+      "loss": 1.0553,
+      "step": 5656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001495334613152373,
+      "loss": 1.0334,
+      "step": 5657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000149524532345194,
+      "loss": 0.9987,
+      "step": 5658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001495156033751507,
+      "loss": 0.9919,
+      "step": 5659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014950667440510738,
+      "loss": 0.9616,
+      "step": 5660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014949774543506408,
+      "loss": 1.031,
+      "step": 5661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014948881646502078,
+      "loss": 0.9759,
+      "step": 5662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014947988749497746,
+      "loss": 0.9986,
+      "step": 5663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014947095852493417,
+      "loss": 1.0932,
+      "step": 5664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014946202955489087,
+      "loss": 0.9829,
+      "step": 5665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014945310058484755,
+      "loss": 1.0624,
+      "step": 5666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014944417161480425,
+      "loss": 1.0362,
+      "step": 5667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014943524264476095,
+      "loss": 0.9161,
+      "step": 5668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014942631367471763,
+      "loss": 1.1597,
+      "step": 5669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014941738470467434,
+      "loss": 1.0498,
+      "step": 5670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000149408455734631,
+      "loss": 1.0839,
+      "step": 5671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014939952676458772,
+      "loss": 0.9343,
+      "step": 5672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001493905977945444,
+      "loss": 1.0799,
+      "step": 5673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001493816688245011,
+      "loss": 0.9729,
+      "step": 5674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001493727398544578,
+      "loss": 0.9789,
+      "step": 5675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014936381088441448,
+      "loss": 1.2118,
+      "step": 5676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014935488191437118,
+      "loss": 1.0293,
+      "step": 5677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014934595294432786,
+      "loss": 1.0578,
+      "step": 5678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014933702397428456,
+      "loss": 1.2393,
+      "step": 5679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014932809500424127,
+      "loss": 1.0531,
+      "step": 5680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014931916603419794,
+      "loss": 1.0963,
+      "step": 5681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014931023706415465,
+      "loss": 1.0042,
+      "step": 5682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014930130809411135,
+      "loss": 1.0925,
+      "step": 5683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014929237912406803,
+      "loss": 1.0245,
+      "step": 5684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014928345015402473,
+      "loss": 1.076,
+      "step": 5685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014927452118398144,
+      "loss": 1.0118,
+      "step": 5686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014926559221393811,
+      "loss": 1.0478,
+      "step": 5687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014925666324389482,
+      "loss": 1.1061,
+      "step": 5688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014924773427385152,
+      "loss": 1.0394,
+      "step": 5689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001492388053038082,
+      "loss": 0.9276,
+      "step": 5690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001492298763337649,
+      "loss": 1.0691,
+      "step": 5691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001492209473637216,
+      "loss": 0.9841,
+      "step": 5692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014921201839367828,
+      "loss": 1.0203,
+      "step": 5693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000149203089423635,
+      "loss": 0.9406,
+      "step": 5694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001491941604535917,
+      "loss": 0.9854,
+      "step": 5695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014918523148354837,
+      "loss": 1.0261,
+      "step": 5696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014917630251350507,
+      "loss": 0.9534,
+      "step": 5697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014916737354346178,
+      "loss": 1.0739,
+      "step": 5698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014915844457341845,
+      "loss": 1.0784,
+      "step": 5699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014914951560337516,
+      "loss": 1.1819,
+      "step": 5700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014914058663333186,
+      "loss": 1.201,
+      "step": 5701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014913165766328854,
+      "loss": 1.0697,
+      "step": 5702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014912272869324524,
+      "loss": 0.9851,
+      "step": 5703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014911379972320195,
+      "loss": 0.9659,
+      "step": 5704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014910487075315862,
+      "loss": 1.0286,
+      "step": 5705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014909594178311533,
+      "loss": 1.1396,
+      "step": 5706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014908701281307203,
+      "loss": 0.8548,
+      "step": 5707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001490780838430287,
+      "loss": 1.0633,
+      "step": 5708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001490691548729854,
+      "loss": 0.9348,
+      "step": 5709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014906022590294212,
+      "loss": 1.0302,
+      "step": 5710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001490512969328988,
+      "loss": 1.0765,
+      "step": 5711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001490423679628555,
+      "loss": 1.2478,
+      "step": 5712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001490334389928122,
+      "loss": 1.052,
+      "step": 5713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014902451002276888,
+      "loss": 1.0447,
+      "step": 5714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014901558105272558,
+      "loss": 1.0302,
+      "step": 5715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001490066520826823,
+      "loss": 1.0876,
+      "step": 5716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014899772311263896,
+      "loss": 1.0904,
+      "step": 5717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014898879414259567,
+      "loss": 0.996,
+      "step": 5718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014897986517255235,
+      "loss": 1.1194,
+      "step": 5719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014897093620250905,
+      "loss": 1.0488,
+      "step": 5720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014896200723246573,
+      "loss": 1.218,
+      "step": 5721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014895307826242243,
+      "loss": 0.9803,
+      "step": 5722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001489441492923791,
+      "loss": 1.071,
+      "step": 5723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001489352203223358,
+      "loss": 1.0745,
+      "step": 5724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014892629135229252,
+      "loss": 1.0268,
+      "step": 5725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001489173623822492,
+      "loss": 0.9844,
+      "step": 5726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001489084334122059,
+      "loss": 1.081,
+      "step": 5727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001488995044421626,
+      "loss": 1.0011,
+      "step": 5728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001488905754721193,
+      "loss": 1.1373,
+      "step": 5729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014888164650207598,
+      "loss": 1.0389,
+      "step": 5730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014887271753203269,
+      "loss": 1.1771,
+      "step": 5731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001488637885619894,
+      "loss": 1.0973,
+      "step": 5732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014885485959194607,
+      "loss": 1.0251,
+      "step": 5733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014884593062190277,
+      "loss": 1.0095,
+      "step": 5734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014883700165185947,
+      "loss": 1.093,
+      "step": 5735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014882807268181615,
+      "loss": 1.0312,
+      "step": 5736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014881914371177286,
+      "loss": 1.0545,
+      "step": 5737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014881021474172956,
+      "loss": 1.0132,
+      "step": 5738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014880128577168624,
+      "loss": 0.9022,
+      "step": 5739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014879235680164294,
+      "loss": 1.2068,
+      "step": 5740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014878342783159964,
+      "loss": 1.1791,
+      "step": 5741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014877449886155632,
+      "loss": 0.994,
+      "step": 5742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014876556989151303,
+      "loss": 1.0727,
+      "step": 5743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014875664092146973,
+      "loss": 1.0667,
+      "step": 5744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001487477119514264,
+      "loss": 1.0069,
+      "step": 5745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001487387829813831,
+      "loss": 0.9811,
+      "step": 5746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014872985401133981,
+      "loss": 0.9733,
+      "step": 5747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001487209250412965,
+      "loss": 1.0346,
+      "step": 5748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001487119960712532,
+      "loss": 1.03,
+      "step": 5749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001487030671012099,
+      "loss": 1.0126,
+      "step": 5750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014869413813116658,
+      "loss": 1.1041,
+      "step": 5751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014868520916112328,
+      "loss": 1.0514,
+      "step": 5752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014867628019107998,
+      "loss": 1.0449,
+      "step": 5753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014866735122103666,
+      "loss": 1.1179,
+      "step": 5754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014865842225099337,
+      "loss": 1.0099,
+      "step": 5755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014864949328095007,
+      "loss": 1.0493,
+      "step": 5756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014864056431090675,
+      "loss": 1.0193,
+      "step": 5757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014863163534086345,
+      "loss": 1.0869,
+      "step": 5758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014862270637082015,
+      "loss": 1.1064,
+      "step": 5759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014861377740077683,
+      "loss": 1.0528,
+      "step": 5760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014860484843073354,
+      "loss": 1.1605,
+      "step": 5761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001485959194606902,
+      "loss": 1.102,
+      "step": 5762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014858699049064692,
+      "loss": 1.2673,
+      "step": 5763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001485780615206036,
+      "loss": 0.9988,
+      "step": 5764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001485691325505603,
+      "loss": 0.982,
+      "step": 5765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000148560203580517,
+      "loss": 1.1745,
+      "step": 5766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014855127461047368,
+      "loss": 1.1718,
+      "step": 5767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014854234564043038,
+      "loss": 0.9479,
+      "step": 5768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014853341667038706,
+      "loss": 0.9957,
+      "step": 5769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014852448770034376,
+      "loss": 1.0602,
+      "step": 5770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014851555873030047,
+      "loss": 1.0668,
+      "step": 5771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014850662976025714,
+      "loss": 1.1471,
+      "step": 5772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014849770079021385,
+      "loss": 0.9521,
+      "step": 5773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014848877182017055,
+      "loss": 1.0224,
+      "step": 5774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014847984285012723,
+      "loss": 1.0201,
+      "step": 5775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014847091388008393,
+      "loss": 1.3318,
+      "step": 5776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014846198491004064,
+      "loss": 0.9829,
+      "step": 5777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014845305593999731,
+      "loss": 1.1155,
+      "step": 5778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014844412696995402,
+      "loss": 1.0525,
+      "step": 5779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014843519799991072,
+      "loss": 0.9785,
+      "step": 5780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001484262690298674,
+      "loss": 1.0371,
+      "step": 5781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001484173400598241,
+      "loss": 1.1041,
+      "step": 5782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001484084110897808,
+      "loss": 1.0665,
+      "step": 5783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014839948211973748,
+      "loss": 1.0641,
+      "step": 5784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001483905531496942,
+      "loss": 1.1263,
+      "step": 5785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001483816241796509,
+      "loss": 1.0583,
+      "step": 5786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014837269520960757,
+      "loss": 1.0373,
+      "step": 5787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014836376623956427,
+      "loss": 1.1307,
+      "step": 5788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014835483726952098,
+      "loss": 1.0014,
+      "step": 5789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014834590829947765,
+      "loss": 0.9926,
+      "step": 5790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014833697932943436,
+      "loss": 0.9976,
+      "step": 5791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014832805035939106,
+      "loss": 1.0089,
+      "step": 5792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014831912138934774,
+      "loss": 1.2293,
+      "step": 5793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014831019241930444,
+      "loss": 0.9894,
+      "step": 5794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014830126344926115,
+      "loss": 1.0003,
+      "step": 5795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014829233447921782,
+      "loss": 0.9901,
+      "step": 5796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014828340550917453,
+      "loss": 1.0099,
+      "step": 5797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014827447653913123,
+      "loss": 1.1476,
+      "step": 5798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001482655475690879,
+      "loss": 1.1852,
+      "step": 5799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001482566185990446,
+      "loss": 1.0354,
+      "step": 5800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014824768962900132,
+      "loss": 1.0867,
+      "step": 5801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000148238760658958,
+      "loss": 0.9714,
+      "step": 5802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001482298316889147,
+      "loss": 1.063,
+      "step": 5803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001482209027188714,
+      "loss": 1.084,
+      "step": 5804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014821197374882808,
+      "loss": 1.0,
+      "step": 5805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014820304477878478,
+      "loss": 0.9419,
+      "step": 5806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001481941158087415,
+      "loss": 1.147,
+      "step": 5807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014818518683869816,
+      "loss": 1.0469,
+      "step": 5808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014817625786865487,
+      "loss": 1.0498,
+      "step": 5809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014816732889861154,
+      "loss": 0.933,
+      "step": 5810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014815839992856825,
+      "loss": 0.9875,
+      "step": 5811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014814947095852493,
+      "loss": 0.9657,
+      "step": 5812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014814054198848163,
+      "loss": 1.0286,
+      "step": 5813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001481316130184383,
+      "loss": 1.0375,
+      "step": 5814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000148122684048395,
+      "loss": 0.9618,
+      "step": 5815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014811375507835171,
+      "loss": 1.0561,
+      "step": 5816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001481048261083084,
+      "loss": 1.013,
+      "step": 5817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001480958971382651,
+      "loss": 1.0836,
+      "step": 5818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001480869681682218,
+      "loss": 1.0659,
+      "step": 5819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014807803919817848,
+      "loss": 0.9833,
+      "step": 5820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014806911022813518,
+      "loss": 1.0607,
+      "step": 5821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014806018125809188,
+      "loss": 0.9317,
+      "step": 5822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014805125228804856,
+      "loss": 1.0488,
+      "step": 5823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014804232331800527,
+      "loss": 0.9592,
+      "step": 5824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014803339434796197,
+      "loss": 0.9909,
+      "step": 5825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014802446537791867,
+      "loss": 0.9418,
+      "step": 5826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014801553640787535,
+      "loss": 1.0867,
+      "step": 5827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014800660743783205,
+      "loss": 0.9807,
+      "step": 5828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014799767846778876,
+      "loss": 1.0921,
+      "step": 5829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014798874949774544,
+      "loss": 0.9684,
+      "step": 5830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014797982052770214,
+      "loss": 1.0029,
+      "step": 5831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014797089155765884,
+      "loss": 1.0381,
+      "step": 5832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014796196258761552,
+      "loss": 1.0261,
+      "step": 5833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014795303361757222,
+      "loss": 1.0917,
+      "step": 5834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014794410464752893,
+      "loss": 1.0643,
+      "step": 5835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001479351756774856,
+      "loss": 1.0117,
+      "step": 5836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001479262467074423,
+      "loss": 1.1128,
+      "step": 5837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014791731773739901,
+      "loss": 0.9863,
+      "step": 5838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001479083887673557,
+      "loss": 1.171,
+      "step": 5839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001478994597973124,
+      "loss": 0.9937,
+      "step": 5840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001478905308272691,
+      "loss": 0.9675,
+      "step": 5841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014788160185722578,
+      "loss": 1.0728,
+      "step": 5842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014787267288718248,
+      "loss": 1.0412,
+      "step": 5843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014786374391713918,
+      "loss": 1.0244,
+      "step": 5844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014785481494709586,
+      "loss": 0.9941,
+      "step": 5845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014784588597705256,
+      "loss": 1.0268,
+      "step": 5846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014783695700700927,
+      "loss": 1.0424,
+      "step": 5847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014782802803696595,
+      "loss": 1.0329,
+      "step": 5848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014781909906692265,
+      "loss": 0.9794,
+      "step": 5849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014781017009687935,
+      "loss": 0.9922,
+      "step": 5850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014780124112683603,
+      "loss": 0.9387,
+      "step": 5851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014779231215679273,
+      "loss": 1.0736,
+      "step": 5852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001477833831867494,
+      "loss": 1.0341,
+      "step": 5853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014777445421670612,
+      "loss": 1.034,
+      "step": 5854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001477655252466628,
+      "loss": 1.1727,
+      "step": 5855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001477565962766195,
+      "loss": 1.002,
+      "step": 5856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001477476673065762,
+      "loss": 0.9989,
+      "step": 5857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014773873833653288,
+      "loss": 1.117,
+      "step": 5858
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014772980936648958,
+      "loss": 0.9985,
+      "step": 5859
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014772088039644626,
+      "loss": 1.0569,
+      "step": 5860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014771195142640296,
+      "loss": 0.9347,
+      "step": 5861
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014770302245635967,
+      "loss": 0.9678,
+      "step": 5862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014769409348631634,
+      "loss": 1.0854,
+      "step": 5863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014768516451627305,
+      "loss": 1.1155,
+      "step": 5864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014767623554622975,
+      "loss": 1.1705,
+      "step": 5865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014766730657618643,
+      "loss": 1.0579,
+      "step": 5866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014765837760614313,
+      "loss": 0.9732,
+      "step": 5867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014764944863609984,
+      "loss": 1.0231,
+      "step": 5868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014764051966605651,
+      "loss": 0.9742,
+      "step": 5869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014763159069601322,
+      "loss": 1.2396,
+      "step": 5870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014762266172596992,
+      "loss": 0.9429,
+      "step": 5871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001476137327559266,
+      "loss": 1.1588,
+      "step": 5872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001476048037858833,
+      "loss": 0.9191,
+      "step": 5873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014759587481584,
+      "loss": 0.9891,
+      "step": 5874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014758694584579668,
+      "loss": 1.0152,
+      "step": 5875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001475780168757534,
+      "loss": 0.9743,
+      "step": 5876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001475690879057101,
+      "loss": 1.0525,
+      "step": 5877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014756015893566677,
+      "loss": 0.8863,
+      "step": 5878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014755122996562347,
+      "loss": 1.0248,
+      "step": 5879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014754230099558018,
+      "loss": 0.981,
+      "step": 5880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014753337202553685,
+      "loss": 0.9144,
+      "step": 5881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014752444305549356,
+      "loss": 1.0506,
+      "step": 5882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014751551408545026,
+      "loss": 1.0204,
+      "step": 5883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014750658511540694,
+      "loss": 1.2461,
+      "step": 5884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014749765614536364,
+      "loss": 0.9763,
+      "step": 5885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014748872717532035,
+      "loss": 1.1248,
+      "step": 5886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014747979820527702,
+      "loss": 0.986,
+      "step": 5887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014747086923523373,
+      "loss": 1.0186,
+      "step": 5888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014746194026519043,
+      "loss": 1.0117,
+      "step": 5889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001474530112951471,
+      "loss": 1.0958,
+      "step": 5890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001474440823251038,
+      "loss": 1.0344,
+      "step": 5891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014743515335506052,
+      "loss": 0.9848,
+      "step": 5892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001474262243850172,
+      "loss": 1.1805,
+      "step": 5893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001474172954149739,
+      "loss": 1.0308,
+      "step": 5894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001474083664449306,
+      "loss": 1.0356,
+      "step": 5895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014739943747488728,
+      "loss": 1.2982,
+      "step": 5896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014739050850484398,
+      "loss": 1.0899,
+      "step": 5897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001473815795348007,
+      "loss": 1.1001,
+      "step": 5898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014737265056475736,
+      "loss": 1.1534,
+      "step": 5899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014736372159471407,
+      "loss": 1.1159,
+      "step": 5900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014735479262467074,
+      "loss": 0.9816,
+      "step": 5901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014734586365462745,
+      "loss": 1.1693,
+      "step": 5902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014733693468458413,
+      "loss": 1.16,
+      "step": 5903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014732800571454083,
+      "loss": 1.1828,
+      "step": 5904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001473190767444975,
+      "loss": 1.0306,
+      "step": 5905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001473101477744542,
+      "loss": 0.998,
+      "step": 5906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014730121880441091,
+      "loss": 1.2294,
+      "step": 5907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001472922898343676,
+      "loss": 1.077,
+      "step": 5908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001472833608643243,
+      "loss": 0.8637,
+      "step": 5909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000147274431894281,
+      "loss": 0.9799,
+      "step": 5910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014726550292423768,
+      "loss": 0.9282,
+      "step": 5911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014725657395419438,
+      "loss": 1.1327,
+      "step": 5912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014724764498415108,
+      "loss": 1.0699,
+      "step": 5913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014723871601410776,
+      "loss": 1.0769,
+      "step": 5914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014722978704406447,
+      "loss": 1.0389,
+      "step": 5915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014722085807402117,
+      "loss": 1.1346,
+      "step": 5916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014721192910397785,
+      "loss": 0.9872,
+      "step": 5917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014720300013393455,
+      "loss": 1.1415,
+      "step": 5918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014719407116389125,
+      "loss": 1.0415,
+      "step": 5919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014718514219384793,
+      "loss": 0.9544,
+      "step": 5920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014717621322380464,
+      "loss": 1.0703,
+      "step": 5921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014716728425376134,
+      "loss": 1.0644,
+      "step": 5922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014715835528371804,
+      "loss": 1.0051,
+      "step": 5923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014714942631367472,
+      "loss": 1.0618,
+      "step": 5924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014714049734363142,
+      "loss": 1.0346,
+      "step": 5925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014713156837358813,
+      "loss": 0.9264,
+      "step": 5926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001471226394035448,
+      "loss": 1.0314,
+      "step": 5927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001471137104335015,
+      "loss": 1.0196,
+      "step": 5928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001471047814634582,
+      "loss": 1.162,
+      "step": 5929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001470958524934149,
+      "loss": 0.923,
+      "step": 5930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001470869235233716,
+      "loss": 1.0204,
+      "step": 5931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001470779945533283,
+      "loss": 1.2138,
+      "step": 5932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014706906558328498,
+      "loss": 1.0954,
+      "step": 5933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014706013661324168,
+      "loss": 0.9444,
+      "step": 5934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014705120764319838,
+      "loss": 0.9926,
+      "step": 5935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014704227867315506,
+      "loss": 1.0013,
+      "step": 5936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014703334970311176,
+      "loss": 1.0124,
+      "step": 5937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014702442073306847,
+      "loss": 0.9864,
+      "step": 5938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014701549176302515,
+      "loss": 1.0101,
+      "step": 5939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014700656279298185,
+      "loss": 0.9396,
+      "step": 5940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014699763382293855,
+      "loss": 1.0443,
+      "step": 5941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014698870485289523,
+      "loss": 1.0962,
+      "step": 5942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014697977588285193,
+      "loss": 0.9667,
+      "step": 5943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001469708469128086,
+      "loss": 0.8812,
+      "step": 5944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014696191794276532,
+      "loss": 0.9736,
+      "step": 5945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000146952988972722,
+      "loss": 1.0632,
+      "step": 5946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001469440600026787,
+      "loss": 1.1051,
+      "step": 5947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001469351310326354,
+      "loss": 1.0637,
+      "step": 5948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014692620206259208,
+      "loss": 1.0906,
+      "step": 5949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014691727309254878,
+      "loss": 1.0522,
+      "step": 5950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014690834412250546,
+      "loss": 1.0999,
+      "step": 5951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014689941515246216,
+      "loss": 1.0989,
+      "step": 5952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014689048618241887,
+      "loss": 1.0255,
+      "step": 5953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014688155721237554,
+      "loss": 0.9752,
+      "step": 5954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014687262824233225,
+      "loss": 1.0023,
+      "step": 5955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014686369927228895,
+      "loss": 1.2419,
+      "step": 5956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014685477030224563,
+      "loss": 1.0304,
+      "step": 5957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014684584133220233,
+      "loss": 1.0054,
+      "step": 5958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014683691236215904,
+      "loss": 1.0134,
+      "step": 5959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001468279833921157,
+      "loss": 1.0422,
+      "step": 5960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014681905442207242,
+      "loss": 1.2137,
+      "step": 5961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014681012545202912,
+      "loss": 1.0546,
+      "step": 5962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001468011964819858,
+      "loss": 0.9862,
+      "step": 5963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001467922675119425,
+      "loss": 0.9747,
+      "step": 5964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001467833385418992,
+      "loss": 1.0209,
+      "step": 5965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014677440957185588,
+      "loss": 1.0739,
+      "step": 5966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001467654806018126,
+      "loss": 0.9613,
+      "step": 5967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001467565516317693,
+      "loss": 1.0974,
+      "step": 5968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014674762266172597,
+      "loss": 1.0536,
+      "step": 5969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014673869369168267,
+      "loss": 1.0164,
+      "step": 5970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014672976472163938,
+      "loss": 1.0507,
+      "step": 5971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014672083575159605,
+      "loss": 1.0778,
+      "step": 5972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014671190678155276,
+      "loss": 1.0059,
+      "step": 5973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014670297781150946,
+      "loss": 0.993,
+      "step": 5974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014669404884146614,
+      "loss": 1.0322,
+      "step": 5975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014668511987142284,
+      "loss": 0.9301,
+      "step": 5976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014667619090137955,
+      "loss": 0.9841,
+      "step": 5977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014666726193133622,
+      "loss": 0.9948,
+      "step": 5978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014665833296129293,
+      "loss": 1.0214,
+      "step": 5979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014664940399124963,
+      "loss": 1.1039,
+      "step": 5980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001466404750212063,
+      "loss": 1.0109,
+      "step": 5981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000146631546051163,
+      "loss": 0.9072,
+      "step": 5982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014662261708111972,
+      "loss": 1.0489,
+      "step": 5983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001466136881110764,
+      "loss": 1.0057,
+      "step": 5984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001466047591410331,
+      "loss": 0.9252,
+      "step": 5985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001465958301709898,
+      "loss": 0.9734,
+      "step": 5986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014658690120094648,
+      "loss": 1.057,
+      "step": 5987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014657797223090318,
+      "loss": 0.9866,
+      "step": 5988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014656904326085989,
+      "loss": 0.9332,
+      "step": 5989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014656011429081656,
+      "loss": 0.9834,
+      "step": 5990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014655118532077327,
+      "loss": 1.1174,
+      "step": 5991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014654225635072994,
+      "loss": 1.083,
+      "step": 5992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014653332738068665,
+      "loss": 0.9672,
+      "step": 5993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014652439841064333,
+      "loss": 0.9933,
+      "step": 5994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014651546944060003,
+      "loss": 1.0965,
+      "step": 5995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001465065404705567,
+      "loss": 1.0366,
+      "step": 5996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001464976115005134,
+      "loss": 1.0682,
+      "step": 5997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014648868253047011,
+      "loss": 1.0423,
+      "step": 5998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001464797535604268,
+      "loss": 1.0704,
+      "step": 5999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001464708245903835,
+      "loss": 1.0824,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.033518671989441,
+      "eval_runtime": 3050.0486,
+      "eval_samples_per_second": 6.166,
+      "eval_steps_per_second": 0.771,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001464618956203402,
+      "loss": 0.9606,
+      "step": 6001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014645296665029688,
+      "loss": 0.9989,
+      "step": 6002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014644403768025358,
+      "loss": 1.1673,
+      "step": 6003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014643510871021028,
+      "loss": 1.0561,
+      "step": 6004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014642617974016696,
+      "loss": 0.989,
+      "step": 6005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014641725077012367,
+      "loss": 0.9986,
+      "step": 6006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014640832180008037,
+      "loss": 1.1078,
+      "step": 6007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014639939283003705,
+      "loss": 1.0747,
+      "step": 6008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014639046385999375,
+      "loss": 1.0478,
+      "step": 6009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014638153488995045,
+      "loss": 1.0235,
+      "step": 6010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014637260591990713,
+      "loss": 1.095,
+      "step": 6011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014636367694986384,
+      "loss": 0.9362,
+      "step": 6012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014635474797982054,
+      "loss": 0.9913,
+      "step": 6013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014634581900977722,
+      "loss": 1.1693,
+      "step": 6014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014633689003973392,
+      "loss": 0.9883,
+      "step": 6015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014632796106969062,
+      "loss": 1.0456,
+      "step": 6016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001463190320996473,
+      "loss": 0.8934,
+      "step": 6017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000146310103129604,
+      "loss": 1.0581,
+      "step": 6018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001463011741595607,
+      "loss": 1.2401,
+      "step": 6019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014629224518951739,
+      "loss": 1.081,
+      "step": 6020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001462833162194741,
+      "loss": 0.9476,
+      "step": 6021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001462743872494308,
+      "loss": 0.9849,
+      "step": 6022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001462654582793875,
+      "loss": 1.0494,
+      "step": 6023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014625652930934417,
+      "loss": 0.9368,
+      "step": 6024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014624760033930088,
+      "loss": 0.8776,
+      "step": 6025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014623867136925758,
+      "loss": 1.0386,
+      "step": 6026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014622974239921426,
+      "loss": 1.0123,
+      "step": 6027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014622081342917096,
+      "loss": 1.0656,
+      "step": 6028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014621188445912767,
+      "loss": 0.9439,
+      "step": 6029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014620295548908434,
+      "loss": 1.0828,
+      "step": 6030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014619402651904105,
+      "loss": 1.1077,
+      "step": 6031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014618509754899775,
+      "loss": 0.8524,
+      "step": 6032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014617616857895443,
+      "loss": 0.9378,
+      "step": 6033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014616723960891113,
+      "loss": 0.9754,
+      "step": 6034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001461583106388678,
+      "loss": 1.0419,
+      "step": 6035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014614938166882451,
+      "loss": 0.9929,
+      "step": 6036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001461404526987812,
+      "loss": 0.9974,
+      "step": 6037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001461315237287379,
+      "loss": 1.115,
+      "step": 6038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001461225947586946,
+      "loss": 1.0419,
+      "step": 6039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014611366578865128,
+      "loss": 0.9985,
+      "step": 6040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014610473681860798,
+      "loss": 1.038,
+      "step": 6041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014609580784856466,
+      "loss": 0.9434,
+      "step": 6042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014608687887852136,
+      "loss": 1.1236,
+      "step": 6043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014607794990847807,
+      "loss": 1.1495,
+      "step": 6044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014606902093843474,
+      "loss": 0.9721,
+      "step": 6045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014606009196839145,
+      "loss": 0.9784,
+      "step": 6046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014605116299834815,
+      "loss": 0.9858,
+      "step": 6047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014604223402830483,
+      "loss": 1.1128,
+      "step": 6048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014603330505826153,
+      "loss": 1.1587,
+      "step": 6049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014602437608821824,
+      "loss": 0.969,
+      "step": 6050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001460154471181749,
+      "loss": 1.1029,
+      "step": 6051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014600651814813162,
+      "loss": 1.0449,
+      "step": 6052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014599758917808832,
+      "loss": 1.0343,
+      "step": 6053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000145988660208045,
+      "loss": 0.9973,
+      "step": 6054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001459797312380017,
+      "loss": 1.0397,
+      "step": 6055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001459708022679584,
+      "loss": 1.149,
+      "step": 6056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014596187329791508,
+      "loss": 0.9842,
+      "step": 6057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001459529443278718,
+      "loss": 0.9045,
+      "step": 6058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001459440153578285,
+      "loss": 0.9679,
+      "step": 6059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014593508638778517,
+      "loss": 1.0267,
+      "step": 6060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014592615741774187,
+      "loss": 1.0726,
+      "step": 6061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014591722844769858,
+      "loss": 1.0041,
+      "step": 6062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014590829947765525,
+      "loss": 1.0075,
+      "step": 6063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014589937050761196,
+      "loss": 0.9968,
+      "step": 6064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014589044153756866,
+      "loss": 0.997,
+      "step": 6065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014588151256752534,
+      "loss": 1.1846,
+      "step": 6066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014587258359748204,
+      "loss": 0.9328,
+      "step": 6067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014586365462743875,
+      "loss": 0.9171,
+      "step": 6068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014585472565739542,
+      "loss": 1.0595,
+      "step": 6069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014584579668735213,
+      "loss": 0.9549,
+      "step": 6070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014583686771730883,
+      "loss": 1.0511,
+      "step": 6071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001458279387472655,
+      "loss": 0.9952,
+      "step": 6072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001458190097772222,
+      "loss": 0.9585,
+      "step": 6073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014581008080717892,
+      "loss": 1.0504,
+      "step": 6074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001458011518371356,
+      "loss": 0.9752,
+      "step": 6075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001457922228670923,
+      "loss": 1.0284,
+      "step": 6076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000145783293897049,
+      "loss": 0.9971,
+      "step": 6077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014577436492700568,
+      "loss": 1.0875,
+      "step": 6078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014576543595696238,
+      "loss": 1.0454,
+      "step": 6079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014575650698691909,
+      "loss": 1.058,
+      "step": 6080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014574757801687576,
+      "loss": 1.0763,
+      "step": 6081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014573864904683247,
+      "loss": 1.1388,
+      "step": 6082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014572972007678914,
+      "loss": 1.2247,
+      "step": 6083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014572079110674585,
+      "loss": 1.1782,
+      "step": 6084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014571186213670252,
+      "loss": 1.0771,
+      "step": 6085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00014570293316665923,
+      "loss": 0.9504,
+      "step": 6086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001456940041966159,
+      "loss": 1.0025,
+      "step": 6087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001456850752265726,
+      "loss": 0.9236,
+      "step": 6088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014567614625652931,
+      "loss": 1.0322,
+      "step": 6089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000145667217286486,
+      "loss": 0.9961,
+      "step": 6090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001456582883164427,
+      "loss": 0.9854,
+      "step": 6091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001456493593463994,
+      "loss": 1.0306,
+      "step": 6092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014564043037635608,
+      "loss": 1.0069,
+      "step": 6093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014563150140631278,
+      "loss": 0.9444,
+      "step": 6094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014562257243626948,
+      "loss": 1.1597,
+      "step": 6095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014561364346622616,
+      "loss": 1.0949,
+      "step": 6096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014560471449618286,
+      "loss": 1.0714,
+      "step": 6097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014559578552613957,
+      "loss": 0.9182,
+      "step": 6098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014558685655609625,
+      "loss": 0.9797,
+      "step": 6099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014557792758605295,
+      "loss": 1.0358,
+      "step": 6100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014556899861600965,
+      "loss": 1.228,
+      "step": 6101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014556006964596633,
+      "loss": 0.9762,
+      "step": 6102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014555114067592303,
+      "loss": 0.8681,
+      "step": 6103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014554221170587974,
+      "loss": 1.0967,
+      "step": 6104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014553328273583642,
+      "loss": 1.02,
+      "step": 6105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014552435376579312,
+      "loss": 1.0263,
+      "step": 6106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014551542479574982,
+      "loss": 0.9974,
+      "step": 6107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001455064958257065,
+      "loss": 1.04,
+      "step": 6108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001454975668556632,
+      "loss": 1.0885,
+      "step": 6109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001454886378856199,
+      "loss": 0.9928,
+      "step": 6110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014547970891557659,
+      "loss": 0.9446,
+      "step": 6111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001454707799455333,
+      "loss": 1.021,
+      "step": 6112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014546185097549,
+      "loss": 1.0453,
+      "step": 6113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014545292200544667,
+      "loss": 1.1828,
+      "step": 6114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014544399303540337,
+      "loss": 1.3148,
+      "step": 6115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014543506406536008,
+      "loss": 1.0841,
+      "step": 6116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014542613509531676,
+      "loss": 0.9918,
+      "step": 6117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014541720612527346,
+      "loss": 1.0791,
+      "step": 6118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014540827715523016,
+      "loss": 0.9516,
+      "step": 6119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014539934818518687,
+      "loss": 0.8691,
+      "step": 6120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014539041921514354,
+      "loss": 1.0662,
+      "step": 6121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014538149024510025,
+      "loss": 1.2454,
+      "step": 6122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014537256127505695,
+      "loss": 0.9895,
+      "step": 6123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014536363230501363,
+      "loss": 1.041,
+      "step": 6124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014535470333497033,
+      "loss": 1.299,
+      "step": 6125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000145345774364927,
+      "loss": 1.1211,
+      "step": 6126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014533684539488371,
+      "loss": 1.1083,
+      "step": 6127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001453279164248404,
+      "loss": 1.0862,
+      "step": 6128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001453189874547971,
+      "loss": 1.1648,
+      "step": 6129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001453100584847538,
+      "loss": 1.0486,
+      "step": 6130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014530112951471048,
+      "loss": 1.0659,
+      "step": 6131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014529220054466718,
+      "loss": 1.0165,
+      "step": 6132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014528327157462386,
+      "loss": 1.1975,
+      "step": 6133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014527434260458056,
+      "loss": 1.0567,
+      "step": 6134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014526541363453727,
+      "loss": 1.0779,
+      "step": 6135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014525648466449394,
+      "loss": 0.9858,
+      "step": 6136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014524755569445065,
+      "loss": 1.035,
+      "step": 6137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014523862672440735,
+      "loss": 1.2067,
+      "step": 6138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014522969775436403,
+      "loss": 2.2426,
+      "step": 6139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014522076878432073,
+      "loss": 1.0566,
+      "step": 6140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014521183981427744,
+      "loss": 0.925,
+      "step": 6141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001452029108442341,
+      "loss": 1.2762,
+      "step": 6142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014519398187419082,
+      "loss": 1.0081,
+      "step": 6143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014518505290414752,
+      "loss": 0.9076,
+      "step": 6144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001451761239341042,
+      "loss": 1.0576,
+      "step": 6145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001451671949640609,
+      "loss": 1.0171,
+      "step": 6146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001451582659940176,
+      "loss": 1.0446,
+      "step": 6147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014514933702397428,
+      "loss": 1.0925,
+      "step": 6148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000145140408053931,
+      "loss": 1.1492,
+      "step": 6149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001451314790838877,
+      "loss": 1.1432,
+      "step": 6150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014512255011384437,
+      "loss": 1.0914,
+      "step": 6151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014511362114380107,
+      "loss": 1.0273,
+      "step": 6152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014510469217375778,
+      "loss": 1.0997,
+      "step": 6153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014509576320371445,
+      "loss": 1.0556,
+      "step": 6154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014508683423367116,
+      "loss": 1.2342,
+      "step": 6155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014507790526362786,
+      "loss": 1.0367,
+      "step": 6156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014506897629358454,
+      "loss": 1.0253,
+      "step": 6157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014506004732354124,
+      "loss": 0.9311,
+      "step": 6158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014505111835349795,
+      "loss": 1.0748,
+      "step": 6159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014504218938345462,
+      "loss": 1.1249,
+      "step": 6160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00014503326041341133,
+      "loss": 1.1446,
+      "step": 6161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014502433144336803,
+      "loss": 1.1949,
+      "step": 6162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001450154024733247,
+      "loss": 1.0792,
+      "step": 6163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001450064735032814,
+      "loss": 1.1296,
+      "step": 6164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014499754453323812,
+      "loss": 1.0179,
+      "step": 6165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001449886155631948,
+      "loss": 1.0217,
+      "step": 6166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001449796865931515,
+      "loss": 1.146,
+      "step": 6167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001449707576231082,
+      "loss": 1.1529,
+      "step": 6168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014496182865306488,
+      "loss": 1.1193,
+      "step": 6169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014495289968302158,
+      "loss": 1.1195,
+      "step": 6170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014494397071297829,
+      "loss": 0.9494,
+      "step": 6171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014493504174293496,
+      "loss": 0.9983,
+      "step": 6172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014492611277289167,
+      "loss": 1.1149,
+      "step": 6173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014491718380284834,
+      "loss": 1.0058,
+      "step": 6174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014490825483280505,
+      "loss": 0.9075,
+      "step": 6175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014489932586276172,
+      "loss": 1.0015,
+      "step": 6176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014489039689271843,
+      "loss": 1.1892,
+      "step": 6177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001448814679226751,
+      "loss": 1.0813,
+      "step": 6178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001448725389526318,
+      "loss": 1.0514,
+      "step": 6179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001448636099825885,
+      "loss": 1.1824,
+      "step": 6180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001448546810125452,
+      "loss": 0.9396,
+      "step": 6181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001448457520425019,
+      "loss": 1.2003,
+      "step": 6182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001448368230724586,
+      "loss": 1.1839,
+      "step": 6183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014482789410241528,
+      "loss": 1.0267,
+      "step": 6184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014481896513237198,
+      "loss": 0.998,
+      "step": 6185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014481003616232868,
+      "loss": 1.0001,
+      "step": 6186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014480110719228536,
+      "loss": 1.0282,
+      "step": 6187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014479217822224206,
+      "loss": 1.0417,
+      "step": 6188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014478324925219877,
+      "loss": 0.9463,
+      "step": 6189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014477432028215545,
+      "loss": 0.9558,
+      "step": 6190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014476539131211215,
+      "loss": 0.9745,
+      "step": 6191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014475646234206885,
+      "loss": 0.9854,
+      "step": 6192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014474753337202553,
+      "loss": 1.0377,
+      "step": 6193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014473860440198223,
+      "loss": 1.1259,
+      "step": 6194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014472967543193894,
+      "loss": 0.9693,
+      "step": 6195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014472074646189562,
+      "loss": 0.9138,
+      "step": 6196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014471181749185232,
+      "loss": 0.9936,
+      "step": 6197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014470288852180902,
+      "loss": 1.0427,
+      "step": 6198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001446939595517657,
+      "loss": 1.08,
+      "step": 6199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001446850305817224,
+      "loss": 1.1539,
+      "step": 6200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001446761016116791,
+      "loss": 1.0123,
+      "step": 6201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014466717264163579,
+      "loss": 1.1286,
+      "step": 6202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001446582436715925,
+      "loss": 0.9847,
+      "step": 6203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001446493147015492,
+      "loss": 1.0609,
+      "step": 6204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014464038573150587,
+      "loss": 0.9895,
+      "step": 6205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014463145676146257,
+      "loss": 1.2295,
+      "step": 6206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014462252779141928,
+      "loss": 1.0357,
+      "step": 6207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014461359882137596,
+      "loss": 1.0391,
+      "step": 6208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014460466985133266,
+      "loss": 1.0555,
+      "step": 6209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014459574088128936,
+      "loss": 1.1383,
+      "step": 6210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014458681191124604,
+      "loss": 1.014,
+      "step": 6211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014457788294120274,
+      "loss": 1.1167,
+      "step": 6212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014456895397115945,
+      "loss": 1.0849,
+      "step": 6213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014456002500111613,
+      "loss": 1.0202,
+      "step": 6214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014455109603107283,
+      "loss": 1.0106,
+      "step": 6215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014454216706102953,
+      "loss": 1.1213,
+      "step": 6216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001445332380909862,
+      "loss": 1.0668,
+      "step": 6217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014452430912094291,
+      "loss": 1.1441,
+      "step": 6218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001445153801508996,
+      "loss": 1.0209,
+      "step": 6219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001445064511808563,
+      "loss": 1.0502,
+      "step": 6220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014449752221081297,
+      "loss": 0.942,
+      "step": 6221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014448859324076968,
+      "loss": 1.1269,
+      "step": 6222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014447966427072638,
+      "loss": 1.0928,
+      "step": 6223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014447073530068306,
+      "loss": 1.0062,
+      "step": 6224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014446180633063976,
+      "loss": 0.9622,
+      "step": 6225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014445287736059647,
+      "loss": 1.0344,
+      "step": 6226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014444394839055314,
+      "loss": 1.1372,
+      "step": 6227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014443501942050985,
+      "loss": 1.0209,
+      "step": 6228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014442609045046655,
+      "loss": 1.0042,
+      "step": 6229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014441716148042323,
+      "loss": 1.0139,
+      "step": 6230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014440823251037993,
+      "loss": 1.0013,
+      "step": 6231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014439930354033664,
+      "loss": 1.122,
+      "step": 6232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001443903745702933,
+      "loss": 1.1611,
+      "step": 6233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014438144560025002,
+      "loss": 0.9464,
+      "step": 6234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00014437251663020672,
+      "loss": 0.9271,
+      "step": 6235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001443635876601634,
+      "loss": 1.0145,
+      "step": 6236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001443546586901201,
+      "loss": 0.9461,
+      "step": 6237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001443457297200768,
+      "loss": 0.9845,
+      "step": 6238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014433680075003348,
+      "loss": 1.0,
+      "step": 6239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014432787177999019,
+      "loss": 1.097,
+      "step": 6240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001443189428099469,
+      "loss": 1.0793,
+      "step": 6241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014431001383990357,
+      "loss": 1.2316,
+      "step": 6242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014430108486986027,
+      "loss": 1.3078,
+      "step": 6243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014429215589981698,
+      "loss": 1.4064,
+      "step": 6244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014428322692977365,
+      "loss": 1.0286,
+      "step": 6245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014427429795973036,
+      "loss": 1.1221,
+      "step": 6246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014426536898968706,
+      "loss": 1.031,
+      "step": 6247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014425644001964374,
+      "loss": 0.9266,
+      "step": 6248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014424751104960044,
+      "loss": 1.1212,
+      "step": 6249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014423858207955715,
+      "loss": 1.0025,
+      "step": 6250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014422965310951382,
+      "loss": 1.0416,
+      "step": 6251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014422072413947053,
+      "loss": 1.1956,
+      "step": 6252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014421179516942723,
+      "loss": 0.9132,
+      "step": 6253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001442028661993839,
+      "loss": 1.0771,
+      "step": 6254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001441939372293406,
+      "loss": 1.0033,
+      "step": 6255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014418500825929732,
+      "loss": 1.0588,
+      "step": 6256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000144176079289254,
+      "loss": 1.1229,
+      "step": 6257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001441671503192107,
+      "loss": 1.1477,
+      "step": 6258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001441582213491674,
+      "loss": 1.2233,
+      "step": 6259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014414929237912408,
+      "loss": 0.9446,
+      "step": 6260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014414036340908078,
+      "loss": 1.0662,
+      "step": 6261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014413143443903749,
+      "loss": 1.0512,
+      "step": 6262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014412250546899416,
+      "loss": 0.9763,
+      "step": 6263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014411357649895087,
+      "loss": 1.0923,
+      "step": 6264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014410464752890754,
+      "loss": 0.9544,
+      "step": 6265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014409571855886425,
+      "loss": 0.9906,
+      "step": 6266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014408678958882092,
+      "loss": 0.9798,
+      "step": 6267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014407786061877763,
+      "loss": 1.044,
+      "step": 6268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001440689316487343,
+      "loss": 1.0635,
+      "step": 6269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000144060002678691,
+      "loss": 0.9804,
+      "step": 6270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001440510737086477,
+      "loss": 0.9294,
+      "step": 6271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001440421447386044,
+      "loss": 1.101,
+      "step": 6272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001440332157685611,
+      "loss": 1.0228,
+      "step": 6273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001440242867985178,
+      "loss": 1.121,
+      "step": 6274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014401535782847447,
+      "loss": 1.0062,
+      "step": 6275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014400642885843118,
+      "loss": 1.0589,
+      "step": 6276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014399749988838788,
+      "loss": 1.1937,
+      "step": 6277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014398857091834456,
+      "loss": 0.8974,
+      "step": 6278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014397964194830126,
+      "loss": 1.1521,
+      "step": 6279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014397071297825797,
+      "loss": 0.9563,
+      "step": 6280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014396178400821464,
+      "loss": 0.9982,
+      "step": 6281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014395285503817135,
+      "loss": 0.9438,
+      "step": 6282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014394392606812805,
+      "loss": 1.0407,
+      "step": 6283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014393499709808473,
+      "loss": 1.0622,
+      "step": 6284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014392606812804143,
+      "loss": 0.888,
+      "step": 6285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014391713915799814,
+      "loss": 0.9752,
+      "step": 6286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014390821018795481,
+      "loss": 1.103,
+      "step": 6287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014389928121791152,
+      "loss": 1.0149,
+      "step": 6288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014389035224786822,
+      "loss": 1.0833,
+      "step": 6289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001438814232778249,
+      "loss": 1.0808,
+      "step": 6290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001438724943077816,
+      "loss": 1.0599,
+      "step": 6291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001438635653377383,
+      "loss": 1.1424,
+      "step": 6292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014385463636769498,
+      "loss": 0.9088,
+      "step": 6293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001438457073976517,
+      "loss": 1.096,
+      "step": 6294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001438367784276084,
+      "loss": 0.9902,
+      "step": 6295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014382784945756507,
+      "loss": 1.0053,
+      "step": 6296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014381892048752177,
+      "loss": 1.093,
+      "step": 6297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014380999151747848,
+      "loss": 1.0691,
+      "step": 6298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014380106254743515,
+      "loss": 1.1025,
+      "step": 6299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014379213357739186,
+      "loss": 1.0464,
+      "step": 6300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014378320460734856,
+      "loss": 0.9893,
+      "step": 6301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014377427563730524,
+      "loss": 1.0986,
+      "step": 6302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014376534666726194,
+      "loss": 1.1507,
+      "step": 6303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014375641769721865,
+      "loss": 1.0066,
+      "step": 6304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014374748872717532,
+      "loss": 1.0628,
+      "step": 6305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014373855975713203,
+      "loss": 0.9848,
+      "step": 6306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014372963078708873,
+      "loss": 1.0957,
+      "step": 6307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001437207018170454,
+      "loss": 1.0744,
+      "step": 6308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00014371177284700211,
+      "loss": 1.1401,
+      "step": 6309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001437028438769588,
+      "loss": 1.0968,
+      "step": 6310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001436939149069155,
+      "loss": 1.0947,
+      "step": 6311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014368498593687217,
+      "loss": 1.0274,
+      "step": 6312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014367605696682888,
+      "loss": 1.2179,
+      "step": 6313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014366712799678558,
+      "loss": 1.0306,
+      "step": 6314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014365819902674226,
+      "loss": 0.9993,
+      "step": 6315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014364927005669896,
+      "loss": 0.8673,
+      "step": 6316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014364034108665566,
+      "loss": 1.1034,
+      "step": 6317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014363141211661234,
+      "loss": 1.0368,
+      "step": 6318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014362248314656905,
+      "loss": 1.1532,
+      "step": 6319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014361355417652575,
+      "loss": 0.9641,
+      "step": 6320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014360462520648243,
+      "loss": 1.0509,
+      "step": 6321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014359569623643913,
+      "loss": 1.1514,
+      "step": 6322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014358676726639583,
+      "loss": 0.8933,
+      "step": 6323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001435778382963525,
+      "loss": 1.0422,
+      "step": 6324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014356890932630922,
+      "loss": 1.1423,
+      "step": 6325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014355998035626592,
+      "loss": 1.0463,
+      "step": 6326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001435510513862226,
+      "loss": 1.1079,
+      "step": 6327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001435421224161793,
+      "loss": 1.0373,
+      "step": 6328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000143533193446136,
+      "loss": 1.0314,
+      "step": 6329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014352426447609268,
+      "loss": 1.1355,
+      "step": 6330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014351533550604939,
+      "loss": 0.9818,
+      "step": 6331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001435064065360061,
+      "loss": 1.0143,
+      "step": 6332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014349747756596277,
+      "loss": 1.0177,
+      "step": 6333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014348854859591947,
+      "loss": 0.964,
+      "step": 6334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014347961962587617,
+      "loss": 0.9853,
+      "step": 6335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014347069065583285,
+      "loss": 1.0363,
+      "step": 6336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014346176168578956,
+      "loss": 1.111,
+      "step": 6337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014345283271574626,
+      "loss": 1.0393,
+      "step": 6338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014344390374570294,
+      "loss": 1.0548,
+      "step": 6339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014343497477565964,
+      "loss": 0.8962,
+      "step": 6340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014342604580561634,
+      "loss": 1.0969,
+      "step": 6341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014341711683557302,
+      "loss": 1.371,
+      "step": 6342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014340818786552973,
+      "loss": 0.9544,
+      "step": 6343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014339925889548643,
+      "loss": 0.9913,
+      "step": 6344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001433903299254431,
+      "loss": 0.9819,
+      "step": 6345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001433814009553998,
+      "loss": 1.022,
+      "step": 6346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014337247198535651,
+      "loss": 0.9056,
+      "step": 6347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001433635430153132,
+      "loss": 1.1187,
+      "step": 6348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001433546140452699,
+      "loss": 1.1397,
+      "step": 6349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001433456850752266,
+      "loss": 1.0869,
+      "step": 6350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014333675610518328,
+      "loss": 1.1669,
+      "step": 6351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014332782713513998,
+      "loss": 1.0625,
+      "step": 6352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014331889816509666,
+      "loss": 1.054,
+      "step": 6353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014330996919505336,
+      "loss": 1.0816,
+      "step": 6354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014330104022501007,
+      "loss": 1.055,
+      "step": 6355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014329211125496674,
+      "loss": 0.992,
+      "step": 6356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014328318228492345,
+      "loss": 1.0955,
+      "step": 6357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014327425331488012,
+      "loss": 0.9884,
+      "step": 6358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014326532434483683,
+      "loss": 1.021,
+      "step": 6359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001432563953747935,
+      "loss": 0.9719,
+      "step": 6360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001432474664047502,
+      "loss": 1.0583,
+      "step": 6361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001432385374347069,
+      "loss": 1.0056,
+      "step": 6362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001432296084646636,
+      "loss": 1.0007,
+      "step": 6363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001432206794946203,
+      "loss": 1.0149,
+      "step": 6364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000143211750524577,
+      "loss": 1.0125,
+      "step": 6365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014320282155453367,
+      "loss": 1.1554,
+      "step": 6366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014319389258449038,
+      "loss": 0.9572,
+      "step": 6367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014318496361444708,
+      "loss": 1.0518,
+      "step": 6368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014317603464440376,
+      "loss": 1.06,
+      "step": 6369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014316710567436046,
+      "loss": 0.9157,
+      "step": 6370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014315817670431717,
+      "loss": 0.9782,
+      "step": 6371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014314924773427384,
+      "loss": 1.1124,
+      "step": 6372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014314031876423055,
+      "loss": 1.1735,
+      "step": 6373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014313138979418725,
+      "loss": 1.0657,
+      "step": 6374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014312246082414393,
+      "loss": 1.1047,
+      "step": 6375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014311353185410063,
+      "loss": 1.0609,
+      "step": 6376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014310460288405734,
+      "loss": 0.9199,
+      "step": 6377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014309567391401401,
+      "loss": 1.0016,
+      "step": 6378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014308674494397072,
+      "loss": 1.0354,
+      "step": 6379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014307781597392742,
+      "loss": 1.199,
+      "step": 6380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001430688870038841,
+      "loss": 1.053,
+      "step": 6381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001430599580338408,
+      "loss": 1.1099,
+      "step": 6382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001430510290637975,
+      "loss": 0.9547,
+      "step": 6383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00014304210009375418,
+      "loss": 0.9721,
+      "step": 6384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001430331711237109,
+      "loss": 0.9782,
+      "step": 6385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001430242421536676,
+      "loss": 0.9121,
+      "step": 6386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014301531318362427,
+      "loss": 1.066,
+      "step": 6387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014300638421358097,
+      "loss": 0.9664,
+      "step": 6388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014299745524353768,
+      "loss": 1.0382,
+      "step": 6389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014298852627349435,
+      "loss": 1.136,
+      "step": 6390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014297959730345106,
+      "loss": 0.9604,
+      "step": 6391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014297066833340776,
+      "loss": 1.0578,
+      "step": 6392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014296173936336444,
+      "loss": 1.0807,
+      "step": 6393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014295281039332114,
+      "loss": 1.0116,
+      "step": 6394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014294388142327785,
+      "loss": 1.0157,
+      "step": 6395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014293495245323452,
+      "loss": 1.1062,
+      "step": 6396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014292602348319123,
+      "loss": 1.0656,
+      "step": 6397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014291709451314793,
+      "loss": 1.0648,
+      "step": 6398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001429081655431046,
+      "loss": 1.1201,
+      "step": 6399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001428992365730613,
+      "loss": 1.2432,
+      "step": 6400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000142890307603018,
+      "loss": 0.9265,
+      "step": 6401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001428813786329747,
+      "loss": 1.0995,
+      "step": 6402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014287244966293137,
+      "loss": 0.9694,
+      "step": 6403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014286352069288808,
+      "loss": 1.0156,
+      "step": 6404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014285459172284478,
+      "loss": 0.991,
+      "step": 6405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014284566275280146,
+      "loss": 1.0668,
+      "step": 6406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014283673378275816,
+      "loss": 1.2136,
+      "step": 6407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014282780481271484,
+      "loss": 0.9938,
+      "step": 6408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014281887584267154,
+      "loss": 0.9503,
+      "step": 6409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014280994687262825,
+      "loss": 0.9575,
+      "step": 6410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014280101790258492,
+      "loss": 0.9934,
+      "step": 6411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014279208893254163,
+      "loss": 0.8872,
+      "step": 6412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014278315996249833,
+      "loss": 1.0156,
+      "step": 6413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014277423099245503,
+      "loss": 1.0882,
+      "step": 6414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001427653020224117,
+      "loss": 0.9771,
+      "step": 6415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014275637305236842,
+      "loss": 1.1968,
+      "step": 6416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014274744408232512,
+      "loss": 0.9087,
+      "step": 6417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001427385151122818,
+      "loss": 0.9481,
+      "step": 6418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001427295861422385,
+      "loss": 1.1007,
+      "step": 6419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001427206571721952,
+      "loss": 0.9994,
+      "step": 6420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014271172820215188,
+      "loss": 1.0961,
+      "step": 6421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014270279923210859,
+      "loss": 1.1574,
+      "step": 6422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001426938702620653,
+      "loss": 0.9617,
+      "step": 6423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014268494129202197,
+      "loss": 1.0131,
+      "step": 6424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014267601232197867,
+      "loss": 1.0424,
+      "step": 6425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014266708335193537,
+      "loss": 0.9874,
+      "step": 6426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014265815438189205,
+      "loss": 0.9775,
+      "step": 6427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014264922541184876,
+      "loss": 1.1434,
+      "step": 6428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014264029644180546,
+      "loss": 1.0944,
+      "step": 6429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014263136747176214,
+      "loss": 1.116,
+      "step": 6430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014262243850171884,
+      "loss": 0.9822,
+      "step": 6431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014261350953167554,
+      "loss": 1.0211,
+      "step": 6432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014260458056163222,
+      "loss": 1.0072,
+      "step": 6433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014259565159158893,
+      "loss": 1.0275,
+      "step": 6434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014258672262154563,
+      "loss": 0.974,
+      "step": 6435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001425777936515023,
+      "loss": 1.0826,
+      "step": 6436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000142568864681459,
+      "loss": 0.9458,
+      "step": 6437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014255993571141571,
+      "loss": 1.0773,
+      "step": 6438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001425510067413724,
+      "loss": 1.0386,
+      "step": 6439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001425420777713291,
+      "loss": 1.1428,
+      "step": 6440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001425331488012858,
+      "loss": 1.0027,
+      "step": 6441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014252421983124248,
+      "loss": 0.9649,
+      "step": 6442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014251529086119918,
+      "loss": 1.1358,
+      "step": 6443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014250636189115586,
+      "loss": 0.8175,
+      "step": 6444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014249743292111256,
+      "loss": 1.0078,
+      "step": 6445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014248850395106927,
+      "loss": 0.9915,
+      "step": 6446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014247957498102594,
+      "loss": 0.8875,
+      "step": 6447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014247064601098265,
+      "loss": 0.9843,
+      "step": 6448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014246171704093932,
+      "loss": 0.9139,
+      "step": 6449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014245278807089603,
+      "loss": 1.0612,
+      "step": 6450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001424438591008527,
+      "loss": 0.9413,
+      "step": 6451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001424349301308094,
+      "loss": 1.0145,
+      "step": 6452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001424260011607661,
+      "loss": 0.9715,
+      "step": 6453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001424170721907228,
+      "loss": 1.0988,
+      "step": 6454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001424081432206795,
+      "loss": 1.0103,
+      "step": 6455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001423992142506362,
+      "loss": 1.2216,
+      "step": 6456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014239028528059287,
+      "loss": 1.0795,
+      "step": 6457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014238135631054958,
+      "loss": 1.0853,
+      "step": 6458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014237242734050628,
+      "loss": 0.9466,
+      "step": 6459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00014236349837046296,
+      "loss": 0.9353,
+      "step": 6460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014235456940041966,
+      "loss": 1.2071,
+      "step": 6461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014234564043037637,
+      "loss": 0.9554,
+      "step": 6462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014233671146033304,
+      "loss": 1.0829,
+      "step": 6463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014232778249028975,
+      "loss": 1.0114,
+      "step": 6464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014231885352024645,
+      "loss": 0.9011,
+      "step": 6465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014230992455020313,
+      "loss": 1.1004,
+      "step": 6466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014230099558015983,
+      "loss": 1.0383,
+      "step": 6467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014229206661011654,
+      "loss": 1.2444,
+      "step": 6468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014228313764007321,
+      "loss": 1.1732,
+      "step": 6469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014227420867002992,
+      "loss": 1.0075,
+      "step": 6470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014226527969998662,
+      "loss": 1.1492,
+      "step": 6471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001422563507299433,
+      "loss": 0.9345,
+      "step": 6472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001422474217599,
+      "loss": 1.0597,
+      "step": 6473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001422384927898567,
+      "loss": 1.111,
+      "step": 6474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014222956381981338,
+      "loss": 1.0213,
+      "step": 6475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001422206348497701,
+      "loss": 1.1597,
+      "step": 6476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001422117058797268,
+      "loss": 1.0436,
+      "step": 6477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014220277690968347,
+      "loss": 1.1207,
+      "step": 6478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014219384793964017,
+      "loss": 1.0254,
+      "step": 6479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014218491896959688,
+      "loss": 1.0785,
+      "step": 6480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014217598999955355,
+      "loss": 1.0784,
+      "step": 6481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014216706102951026,
+      "loss": 1.0792,
+      "step": 6482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014215813205946696,
+      "loss": 0.9934,
+      "step": 6483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014214920308942364,
+      "loss": 0.9753,
+      "step": 6484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014214027411938034,
+      "loss": 1.0654,
+      "step": 6485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014213134514933705,
+      "loss": 1.1589,
+      "step": 6486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014212241617929372,
+      "loss": 1.0282,
+      "step": 6487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014211348720925043,
+      "loss": 1.0969,
+      "step": 6488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014210455823920713,
+      "loss": 1.0397,
+      "step": 6489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001420956292691638,
+      "loss": 1.0073,
+      "step": 6490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001420867002991205,
+      "loss": 1.1495,
+      "step": 6491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001420777713290772,
+      "loss": 1.1583,
+      "step": 6492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001420688423590339,
+      "loss": 1.1776,
+      "step": 6493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014205991338899057,
+      "loss": 0.9023,
+      "step": 6494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014205098441894728,
+      "loss": 0.896,
+      "step": 6495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014204205544890398,
+      "loss": 1.0434,
+      "step": 6496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014203312647886066,
+      "loss": 1.1527,
+      "step": 6497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014202419750881736,
+      "loss": 1.0387,
+      "step": 6498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014201526853877404,
+      "loss": 1.0703,
+      "step": 6499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014200633956873074,
+      "loss": 0.9946,
+      "step": 6500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014199741059868745,
+      "loss": 1.1204,
+      "step": 6501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014198848162864412,
+      "loss": 1.0201,
+      "step": 6502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014197955265860083,
+      "loss": 1.0963,
+      "step": 6503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014197062368855753,
+      "loss": 1.0667,
+      "step": 6504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001419616947185142,
+      "loss": 1.0962,
+      "step": 6505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001419527657484709,
+      "loss": 1.0656,
+      "step": 6506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014194383677842761,
+      "loss": 1.0603,
+      "step": 6507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001419349078083843,
+      "loss": 1.2691,
+      "step": 6508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000141925978838341,
+      "loss": 1.1725,
+      "step": 6509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001419170498682977,
+      "loss": 1.202,
+      "step": 6510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001419081208982544,
+      "loss": 1.0049,
+      "step": 6511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014189919192821108,
+      "loss": 1.1331,
+      "step": 6512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014189026295816778,
+      "loss": 1.0894,
+      "step": 6513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001418813339881245,
+      "loss": 1.1352,
+      "step": 6514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014187240501808117,
+      "loss": 1.0216,
+      "step": 6515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014186347604803787,
+      "loss": 0.9602,
+      "step": 6516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014185454707799457,
+      "loss": 1.1005,
+      "step": 6517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014184561810795125,
+      "loss": 1.0406,
+      "step": 6518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014183668913790795,
+      "loss": 1.0134,
+      "step": 6519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014182776016786466,
+      "loss": 0.9731,
+      "step": 6520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014181883119782134,
+      "loss": 1.0344,
+      "step": 6521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014180990222777804,
+      "loss": 0.9675,
+      "step": 6522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014180097325773474,
+      "loss": 0.9993,
+      "step": 6523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014179204428769142,
+      "loss": 1.0446,
+      "step": 6524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014178311531764812,
+      "loss": 1.0076,
+      "step": 6525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014177418634760483,
+      "loss": 1.0258,
+      "step": 6526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001417652573775615,
+      "loss": 1.0721,
+      "step": 6527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001417563284075182,
+      "loss": 1.0614,
+      "step": 6528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014174739943747491,
+      "loss": 1.04,
+      "step": 6529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001417384704674316,
+      "loss": 1.1661,
+      "step": 6530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001417295414973883,
+      "loss": 0.9654,
+      "step": 6531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000141720612527345,
+      "loss": 0.9819,
+      "step": 6532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014171168355730168,
+      "loss": 1.0687,
+      "step": 6533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00014170275458725838,
+      "loss": 1.0021,
+      "step": 6534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014169382561721506,
+      "loss": 1.0321,
+      "step": 6535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014168489664717176,
+      "loss": 1.1414,
+      "step": 6536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014167596767712846,
+      "loss": 0.9951,
+      "step": 6537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014166703870708514,
+      "loss": 1.045,
+      "step": 6538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014165810973704185,
+      "loss": 1.179,
+      "step": 6539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014164918076699852,
+      "loss": 1.0183,
+      "step": 6540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014164025179695523,
+      "loss": 0.9236,
+      "step": 6541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001416313228269119,
+      "loss": 1.0163,
+      "step": 6542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001416223938568686,
+      "loss": 1.0942,
+      "step": 6543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001416134648868253,
+      "loss": 1.0657,
+      "step": 6544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000141604535916782,
+      "loss": 1.0733,
+      "step": 6545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001415956069467387,
+      "loss": 1.1688,
+      "step": 6546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001415866779766954,
+      "loss": 0.9535,
+      "step": 6547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014157774900665207,
+      "loss": 1.0135,
+      "step": 6548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014156882003660878,
+      "loss": 0.9104,
+      "step": 6549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014155989106656548,
+      "loss": 1.1472,
+      "step": 6550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014155096209652216,
+      "loss": 1.0131,
+      "step": 6551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014154203312647886,
+      "loss": 1.0669,
+      "step": 6552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014153310415643557,
+      "loss": 0.9732,
+      "step": 6553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014152417518639224,
+      "loss": 1.0223,
+      "step": 6554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014151524621634895,
+      "loss": 0.9402,
+      "step": 6555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014150631724630565,
+      "loss": 1.2196,
+      "step": 6556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014149738827626233,
+      "loss": 0.9352,
+      "step": 6557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014148845930621903,
+      "loss": 1.0172,
+      "step": 6558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014147953033617574,
+      "loss": 1.2457,
+      "step": 6559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014147060136613241,
+      "loss": 1.0724,
+      "step": 6560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014146167239608912,
+      "loss": 0.9957,
+      "step": 6561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014145274342604582,
+      "loss": 0.9975,
+      "step": 6562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001414438144560025,
+      "loss": 1.109,
+      "step": 6563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001414348854859592,
+      "loss": 0.9979,
+      "step": 6564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001414259565159159,
+      "loss": 1.0295,
+      "step": 6565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014141702754587258,
+      "loss": 1.0103,
+      "step": 6566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001414080985758293,
+      "loss": 1.0895,
+      "step": 6567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000141399169605786,
+      "loss": 0.9591,
+      "step": 6568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014139024063574267,
+      "loss": 1.0198,
+      "step": 6569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014138131166569937,
+      "loss": 1.0113,
+      "step": 6570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014137238269565608,
+      "loss": 1.1148,
+      "step": 6571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014136345372561275,
+      "loss": 1.1065,
+      "step": 6572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014135452475556946,
+      "loss": 0.9962,
+      "step": 6573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014134559578552616,
+      "loss": 1.1381,
+      "step": 6574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014133666681548284,
+      "loss": 0.9496,
+      "step": 6575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014132773784543954,
+      "loss": 1.0021,
+      "step": 6576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014131880887539625,
+      "loss": 0.6926,
+      "step": 6577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014130987990535292,
+      "loss": 1.093,
+      "step": 6578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014130095093530963,
+      "loss": 1.03,
+      "step": 6579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014129202196526633,
+      "loss": 0.9952,
+      "step": 6580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000141283092995223,
+      "loss": 1.0327,
+      "step": 6581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001412741640251797,
+      "loss": 1.0584,
+      "step": 6582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001412652350551364,
+      "loss": 0.9286,
+      "step": 6583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001412563060850931,
+      "loss": 1.0866,
+      "step": 6584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014124737711504977,
+      "loss": 1.0749,
+      "step": 6585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014123844814500647,
+      "loss": 1.0319,
+      "step": 6586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014122951917496318,
+      "loss": 0.9924,
+      "step": 6587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014122059020491986,
+      "loss": 1.1167,
+      "step": 6588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014121166123487656,
+      "loss": 1.0106,
+      "step": 6589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014120273226483324,
+      "loss": 1.0327,
+      "step": 6590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014119380329478994,
+      "loss": 1.0742,
+      "step": 6591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014118487432474664,
+      "loss": 1.031,
+      "step": 6592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014117594535470332,
+      "loss": 1.1086,
+      "step": 6593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014116701638466003,
+      "loss": 0.976,
+      "step": 6594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014115808741461673,
+      "loss": 1.0403,
+      "step": 6595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001411491584445734,
+      "loss": 1.0078,
+      "step": 6596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001411402294745301,
+      "loss": 1.0578,
+      "step": 6597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014113130050448681,
+      "loss": 0.9911,
+      "step": 6598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001411223715344435,
+      "loss": 1.039,
+      "step": 6599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001411134425644002,
+      "loss": 0.9832,
+      "step": 6600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001411045135943569,
+      "loss": 1.1948,
+      "step": 6601
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014109558462431358,
+      "loss": 1.087,
+      "step": 6602
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014108665565427028,
+      "loss": 0.8876,
+      "step": 6603
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014107772668422698,
+      "loss": 1.0754,
+      "step": 6604
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014106879771418366,
+      "loss": 1.0011,
+      "step": 6605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014105986874414037,
+      "loss": 1.1869,
+      "step": 6606
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014105093977409707,
+      "loss": 1.1844,
+      "step": 6607
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014104201080405377,
+      "loss": 1.163,
+      "step": 6608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00014103308183401045,
+      "loss": 1.0506,
+      "step": 6609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014102415286396715,
+      "loss": 1.1441,
+      "step": 6610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014101522389392386,
+      "loss": 1.1051,
+      "step": 6611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014100629492388054,
+      "loss": 1.0739,
+      "step": 6612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014099736595383724,
+      "loss": 1.0101,
+      "step": 6613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014098843698379394,
+      "loss": 1.1022,
+      "step": 6614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014097950801375062,
+      "loss": 0.9829,
+      "step": 6615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014097057904370732,
+      "loss": 1.0255,
+      "step": 6616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014096165007366403,
+      "loss": 1.12,
+      "step": 6617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001409527211036207,
+      "loss": 1.0635,
+      "step": 6618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001409437921335774,
+      "loss": 1.0922,
+      "step": 6619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014093486316353411,
+      "loss": 1.022,
+      "step": 6620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001409259341934908,
+      "loss": 1.0268,
+      "step": 6621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001409170052234475,
+      "loss": 1.0868,
+      "step": 6622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001409080762534042,
+      "loss": 0.9517,
+      "step": 6623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014089914728336088,
+      "loss": 1.0441,
+      "step": 6624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014089021831331758,
+      "loss": 0.9124,
+      "step": 6625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014088128934327426,
+      "loss": 1.0137,
+      "step": 6626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014087236037323096,
+      "loss": 1.0579,
+      "step": 6627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014086343140318766,
+      "loss": 1.249,
+      "step": 6628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014085450243314434,
+      "loss": 1.0784,
+      "step": 6629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014084557346310105,
+      "loss": 1.0559,
+      "step": 6630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014083664449305772,
+      "loss": 0.9732,
+      "step": 6631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014082771552301443,
+      "loss": 0.92,
+      "step": 6632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001408187865529711,
+      "loss": 0.982,
+      "step": 6633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001408098575829278,
+      "loss": 1.0381,
+      "step": 6634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001408009286128845,
+      "loss": 1.0363,
+      "step": 6635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001407919996428412,
+      "loss": 0.9807,
+      "step": 6636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001407830706727979,
+      "loss": 1.0986,
+      "step": 6637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001407741417027546,
+      "loss": 1.0877,
+      "step": 6638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014076521273271127,
+      "loss": 0.9673,
+      "step": 6639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014075628376266798,
+      "loss": 1.098,
+      "step": 6640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014074735479262468,
+      "loss": 0.9658,
+      "step": 6641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014073842582258136,
+      "loss": 0.9852,
+      "step": 6642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014072949685253806,
+      "loss": 1.032,
+      "step": 6643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014072056788249477,
+      "loss": 1.1974,
+      "step": 6644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014071163891245144,
+      "loss": 1.0721,
+      "step": 6645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014070270994240815,
+      "loss": 1.0093,
+      "step": 6646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014069378097236485,
+      "loss": 1.032,
+      "step": 6647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014068485200232153,
+      "loss": 0.9918,
+      "step": 6648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014067592303227823,
+      "loss": 1.0855,
+      "step": 6649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014066699406223494,
+      "loss": 1.1874,
+      "step": 6650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001406580650921916,
+      "loss": 1.1147,
+      "step": 6651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014064913612214832,
+      "loss": 1.0986,
+      "step": 6652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014064020715210502,
+      "loss": 1.0684,
+      "step": 6653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001406312781820617,
+      "loss": 1.0377,
+      "step": 6654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001406223492120184,
+      "loss": 1.0082,
+      "step": 6655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001406134202419751,
+      "loss": 1.0255,
+      "step": 6656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014060449127193178,
+      "loss": 1.0153,
+      "step": 6657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001405955623018885,
+      "loss": 1.0631,
+      "step": 6658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001405866333318452,
+      "loss": 0.9947,
+      "step": 6659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014057770436180187,
+      "loss": 1.0638,
+      "step": 6660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014056877539175857,
+      "loss": 0.99,
+      "step": 6661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014055984642171528,
+      "loss": 1.0338,
+      "step": 6662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014055091745167195,
+      "loss": 0.9613,
+      "step": 6663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014054198848162866,
+      "loss": 0.9685,
+      "step": 6664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014053305951158536,
+      "loss": 1.1242,
+      "step": 6665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014052413054154204,
+      "loss": 0.875,
+      "step": 6666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014051520157149874,
+      "loss": 1.0127,
+      "step": 6667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014050627260145545,
+      "loss": 1.1213,
+      "step": 6668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014049734363141212,
+      "loss": 1.0292,
+      "step": 6669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014048841466136883,
+      "loss": 0.9762,
+      "step": 6670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014047948569132553,
+      "loss": 1.1763,
+      "step": 6671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001404705567212822,
+      "loss": 1.1914,
+      "step": 6672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001404616277512389,
+      "loss": 1.1249,
+      "step": 6673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001404526987811956,
+      "loss": 1.1746,
+      "step": 6674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001404437698111523,
+      "loss": 0.9676,
+      "step": 6675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014043484084110897,
+      "loss": 1.1092,
+      "step": 6676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014042591187106567,
+      "loss": 1.1384,
+      "step": 6677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014041698290102235,
+      "loss": 1.0694,
+      "step": 6678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014040805393097906,
+      "loss": 1.0372,
+      "step": 6679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014039912496093576,
+      "loss": 1.0064,
+      "step": 6680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014039019599089244,
+      "loss": 1.0519,
+      "step": 6681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014038126702084914,
+      "loss": 1.063,
+      "step": 6682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014037233805080584,
+      "loss": 1.0383,
+      "step": 6683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00014036340908076252,
+      "loss": 1.103,
+      "step": 6684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014035448011071923,
+      "loss": 0.9711,
+      "step": 6685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014034555114067593,
+      "loss": 0.9297,
+      "step": 6686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001403366221706326,
+      "loss": 1.0802,
+      "step": 6687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001403276932005893,
+      "loss": 1.0337,
+      "step": 6688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014031876423054601,
+      "loss": 1.1464,
+      "step": 6689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001403098352605027,
+      "loss": 1.0073,
+      "step": 6690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001403009062904594,
+      "loss": 1.0702,
+      "step": 6691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001402919773204161,
+      "loss": 0.9247,
+      "step": 6692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014028304835037278,
+      "loss": 0.9428,
+      "step": 6693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014027411938032948,
+      "loss": 1.0353,
+      "step": 6694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014026519041028618,
+      "loss": 0.9259,
+      "step": 6695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014025626144024286,
+      "loss": 1.061,
+      "step": 6696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014024733247019957,
+      "loss": 0.966,
+      "step": 6697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014023840350015627,
+      "loss": 1.1105,
+      "step": 6698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014022947453011295,
+      "loss": 1.1809,
+      "step": 6699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014022054556006965,
+      "loss": 1.029,
+      "step": 6700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014021161659002635,
+      "loss": 0.9648,
+      "step": 6701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014020268761998303,
+      "loss": 1.1382,
+      "step": 6702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014019375864993974,
+      "loss": 1.0424,
+      "step": 6703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014018482967989644,
+      "loss": 0.95,
+      "step": 6704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014017590070985312,
+      "loss": 1.1763,
+      "step": 6705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014016697173980982,
+      "loss": 0.9885,
+      "step": 6706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014015804276976652,
+      "loss": 1.1036,
+      "step": 6707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014014911379972323,
+      "loss": 1.1365,
+      "step": 6708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001401401848296799,
+      "loss": 0.9839,
+      "step": 6709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001401312558596366,
+      "loss": 0.8931,
+      "step": 6710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001401223268895933,
+      "loss": 1.0426,
+      "step": 6711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014011339791955,
+      "loss": 1.2688,
+      "step": 6712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001401044689495067,
+      "loss": 0.9973,
+      "step": 6713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001400955399794634,
+      "loss": 1.0484,
+      "step": 6714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014008661100942008,
+      "loss": 0.9894,
+      "step": 6715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014007768203937678,
+      "loss": 1.1565,
+      "step": 6716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014006875306933346,
+      "loss": 1.1145,
+      "step": 6717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014005982409929016,
+      "loss": 0.9924,
+      "step": 6718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014005089512924686,
+      "loss": 1.0128,
+      "step": 6719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014004196615920354,
+      "loss": 1.0435,
+      "step": 6720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014003303718916025,
+      "loss": 1.0612,
+      "step": 6721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014002410821911692,
+      "loss": 1.0712,
+      "step": 6722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00014001517924907363,
+      "loss": 1.0201,
+      "step": 6723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001400062502790303,
+      "loss": 0.8826,
+      "step": 6724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000139997321308987,
+      "loss": 0.9574,
+      "step": 6725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001399883923389437,
+      "loss": 0.9782,
+      "step": 6726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001399794633689004,
+      "loss": 0.9294,
+      "step": 6727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001399705343988571,
+      "loss": 1.0254,
+      "step": 6728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001399616054288138,
+      "loss": 0.9808,
+      "step": 6729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013995267645877047,
+      "loss": 1.0514,
+      "step": 6730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013994374748872718,
+      "loss": 0.9541,
+      "step": 6731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013993481851868388,
+      "loss": 1.1081,
+      "step": 6732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013992588954864056,
+      "loss": 1.0634,
+      "step": 6733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013991696057859726,
+      "loss": 1.0548,
+      "step": 6734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013990803160855397,
+      "loss": 0.968,
+      "step": 6735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013989910263851064,
+      "loss": 0.9696,
+      "step": 6736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013989017366846735,
+      "loss": 1.1363,
+      "step": 6737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013988124469842405,
+      "loss": 1.1312,
+      "step": 6738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013987231572838073,
+      "loss": 0.9973,
+      "step": 6739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013986338675833743,
+      "loss": 1.0344,
+      "step": 6740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013985445778829414,
+      "loss": 1.0184,
+      "step": 6741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001398455288182508,
+      "loss": 0.9681,
+      "step": 6742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013983659984820752,
+      "loss": 1.1692,
+      "step": 6743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013982767087816422,
+      "loss": 1.0602,
+      "step": 6744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001398187419081209,
+      "loss": 1.0348,
+      "step": 6745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001398098129380776,
+      "loss": 1.1294,
+      "step": 6746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001398008839680343,
+      "loss": 1.0077,
+      "step": 6747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013979195499799098,
+      "loss": 1.1519,
+      "step": 6748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001397830260279477,
+      "loss": 1.2106,
+      "step": 6749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001397740970579044,
+      "loss": 0.9632,
+      "step": 6750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013976516808786107,
+      "loss": 1.0809,
+      "step": 6751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013975623911781777,
+      "loss": 0.931,
+      "step": 6752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013974731014777448,
+      "loss": 1.218,
+      "step": 6753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013973838117773115,
+      "loss": 0.9644,
+      "step": 6754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013972945220768786,
+      "loss": 1.031,
+      "step": 6755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013972052323764456,
+      "loss": 1.0704,
+      "step": 6756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013971159426760124,
+      "loss": 0.9825,
+      "step": 6757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00013970266529755794,
+      "loss": 0.9874,
+      "step": 6758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013969373632751465,
+      "loss": 0.9641,
+      "step": 6759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013968480735747132,
+      "loss": 1.1587,
+      "step": 6760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013967587838742803,
+      "loss": 1.0358,
+      "step": 6761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013966694941738473,
+      "loss": 1.0106,
+      "step": 6762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001396580204473414,
+      "loss": 1.1212,
+      "step": 6763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001396490914772981,
+      "loss": 1.033,
+      "step": 6764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001396401625072548,
+      "loss": 1.0515,
+      "step": 6765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001396312335372115,
+      "loss": 1.0896,
+      "step": 6766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013962230456716817,
+      "loss": 0.9969,
+      "step": 6767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013961337559712487,
+      "loss": 1.0456,
+      "step": 6768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013960444662708155,
+      "loss": 0.9719,
+      "step": 6769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013959551765703825,
+      "loss": 1.1036,
+      "step": 6770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013958658868699496,
+      "loss": 1.0031,
+      "step": 6771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013957765971695164,
+      "loss": 1.0118,
+      "step": 6772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013956873074690834,
+      "loss": 1.0479,
+      "step": 6773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013955980177686504,
+      "loss": 0.992,
+      "step": 6774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013955087280682172,
+      "loss": 1.0733,
+      "step": 6775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013954194383677842,
+      "loss": 1.0474,
+      "step": 6776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013953301486673513,
+      "loss": 1.0589,
+      "step": 6777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001395240858966918,
+      "loss": 1.048,
+      "step": 6778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001395151569266485,
+      "loss": 1.0195,
+      "step": 6779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013950622795660521,
+      "loss": 1.0469,
+      "step": 6780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001394972989865619,
+      "loss": 1.0693,
+      "step": 6781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001394883700165186,
+      "loss": 1.0876,
+      "step": 6782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001394794410464753,
+      "loss": 1.04,
+      "step": 6783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013947051207643198,
+      "loss": 1.0739,
+      "step": 6784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013946158310638868,
+      "loss": 1.2016,
+      "step": 6785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013945265413634538,
+      "loss": 1.1033,
+      "step": 6786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013944372516630206,
+      "loss": 1.0848,
+      "step": 6787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013943479619625876,
+      "loss": 1.0677,
+      "step": 6788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013942586722621547,
+      "loss": 1.0641,
+      "step": 6789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013941693825617215,
+      "loss": 0.9491,
+      "step": 6790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013940800928612885,
+      "loss": 0.983,
+      "step": 6791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013939908031608555,
+      "loss": 1.0254,
+      "step": 6792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013939015134604223,
+      "loss": 0.9741,
+      "step": 6793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013938122237599893,
+      "loss": 0.8881,
+      "step": 6794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013937229340595564,
+      "loss": 1.1549,
+      "step": 6795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013936336443591232,
+      "loss": 1.1065,
+      "step": 6796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013935443546586902,
+      "loss": 1.0259,
+      "step": 6797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013934550649582572,
+      "loss": 0.9382,
+      "step": 6798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001393365775257824,
+      "loss": 1.0474,
+      "step": 6799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001393276485557391,
+      "loss": 1.1255,
+      "step": 6800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001393187195856958,
+      "loss": 0.9751,
+      "step": 6801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013930979061565249,
+      "loss": 1.0536,
+      "step": 6802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001393008616456092,
+      "loss": 1.0304,
+      "step": 6803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001392919326755659,
+      "loss": 1.026,
+      "step": 6804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001392830037055226,
+      "loss": 1.1173,
+      "step": 6805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013927407473547927,
+      "loss": 0.9378,
+      "step": 6806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013926514576543598,
+      "loss": 0.9975,
+      "step": 6807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013925621679539266,
+      "loss": 0.9952,
+      "step": 6808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013924728782534936,
+      "loss": 1.034,
+      "step": 6809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013923835885530606,
+      "loss": 1.0515,
+      "step": 6810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013922942988526274,
+      "loss": 1.0283,
+      "step": 6811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013922050091521944,
+      "loss": 0.9226,
+      "step": 6812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013921157194517612,
+      "loss": 0.9994,
+      "step": 6813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013920264297513283,
+      "loss": 0.9954,
+      "step": 6814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001391937140050895,
+      "loss": 1.1428,
+      "step": 6815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001391847850350462,
+      "loss": 1.0408,
+      "step": 6816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001391758560650029,
+      "loss": 1.1245,
+      "step": 6817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001391669270949596,
+      "loss": 1.004,
+      "step": 6818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001391579981249163,
+      "loss": 0.9106,
+      "step": 6819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000139149069154873,
+      "loss": 1.0503,
+      "step": 6820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013914014018482967,
+      "loss": 1.1533,
+      "step": 6821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013913121121478638,
+      "loss": 1.0976,
+      "step": 6822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013912228224474308,
+      "loss": 0.9847,
+      "step": 6823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013911335327469976,
+      "loss": 1.0346,
+      "step": 6824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013910442430465646,
+      "loss": 1.0429,
+      "step": 6825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013909549533461317,
+      "loss": 0.9049,
+      "step": 6826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013908656636456984,
+      "loss": 1.0585,
+      "step": 6827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013907763739452655,
+      "loss": 1.0791,
+      "step": 6828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013906870842448325,
+      "loss": 1.0536,
+      "step": 6829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013905977945443993,
+      "loss": 0.9929,
+      "step": 6830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013905085048439663,
+      "loss": 1.0002,
+      "step": 6831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013904192151435334,
+      "loss": 1.0255,
+      "step": 6832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00013903299254431,
+      "loss": 1.3033,
+      "step": 6833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013902406357426672,
+      "loss": 0.9519,
+      "step": 6834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013901513460422342,
+      "loss": 1.084,
+      "step": 6835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001390062056341801,
+      "loss": 1.0076,
+      "step": 6836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001389972766641368,
+      "loss": 1.0125,
+      "step": 6837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001389883476940935,
+      "loss": 1.0797,
+      "step": 6838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013897941872405018,
+      "loss": 1.0891,
+      "step": 6839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001389704897540069,
+      "loss": 1.067,
+      "step": 6840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001389615607839636,
+      "loss": 1.0392,
+      "step": 6841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013895263181392027,
+      "loss": 1.0348,
+      "step": 6842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013894370284387697,
+      "loss": 1.0799,
+      "step": 6843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013893477387383368,
+      "loss": 1.0125,
+      "step": 6844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013892584490379035,
+      "loss": 1.0372,
+      "step": 6845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013891691593374706,
+      "loss": 1.1506,
+      "step": 6846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013890798696370376,
+      "loss": 1.0464,
+      "step": 6847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013889905799366044,
+      "loss": 1.0479,
+      "step": 6848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013889012902361714,
+      "loss": 1.0423,
+      "step": 6849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013888120005357385,
+      "loss": 0.9453,
+      "step": 6850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013887227108353052,
+      "loss": 0.8423,
+      "step": 6851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013886334211348723,
+      "loss": 1.1825,
+      "step": 6852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013885441314344393,
+      "loss": 1.0201,
+      "step": 6853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001388454841734006,
+      "loss": 1.1081,
+      "step": 6854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001388365552033573,
+      "loss": 1.1137,
+      "step": 6855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000138827626233314,
+      "loss": 1.006,
+      "step": 6856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001388186972632707,
+      "loss": 1.2046,
+      "step": 6857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013880976829322737,
+      "loss": 0.9601,
+      "step": 6858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013880083932318407,
+      "loss": 1.1153,
+      "step": 6859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013879191035314075,
+      "loss": 1.1541,
+      "step": 6860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013878298138309745,
+      "loss": 1.0299,
+      "step": 6861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013877405241305416,
+      "loss": 1.1441,
+      "step": 6862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013876512344301084,
+      "loss": 1.0669,
+      "step": 6863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013875619447296754,
+      "loss": 1.0384,
+      "step": 6864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013874726550292424,
+      "loss": 1.0244,
+      "step": 6865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013873833653288092,
+      "loss": 1.1242,
+      "step": 6866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013872940756283762,
+      "loss": 1.0661,
+      "step": 6867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013872047859279433,
+      "loss": 1.0051,
+      "step": 6868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000138711549622751,
+      "loss": 0.9821,
+      "step": 6869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001387026206527077,
+      "loss": 1.0974,
+      "step": 6870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001386936916826644,
+      "loss": 1.0597,
+      "step": 6871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001386847627126211,
+      "loss": 1.0825,
+      "step": 6872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001386758337425778,
+      "loss": 1.0378,
+      "step": 6873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001386669047725345,
+      "loss": 1.0311,
+      "step": 6874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013865797580249118,
+      "loss": 0.9931,
+      "step": 6875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013864904683244788,
+      "loss": 1.0392,
+      "step": 6876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013864011786240458,
+      "loss": 1.08,
+      "step": 6877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013863118889236126,
+      "loss": 1.1193,
+      "step": 6878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013862225992231796,
+      "loss": 1.1028,
+      "step": 6879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013861333095227467,
+      "loss": 1.0149,
+      "step": 6880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013860440198223135,
+      "loss": 0.9347,
+      "step": 6881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013859547301218805,
+      "loss": 0.9216,
+      "step": 6882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013858654404214475,
+      "loss": 1.1276,
+      "step": 6883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013857761507210143,
+      "loss": 1.0076,
+      "step": 6884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013856868610205813,
+      "loss": 0.9554,
+      "step": 6885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013855975713201484,
+      "loss": 1.139,
+      "step": 6886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013855082816197152,
+      "loss": 1.0277,
+      "step": 6887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013854189919192822,
+      "loss": 0.9698,
+      "step": 6888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013853297022188492,
+      "loss": 0.9778,
+      "step": 6889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001385240412518416,
+      "loss": 1.0956,
+      "step": 6890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001385151122817983,
+      "loss": 1.0265,
+      "step": 6891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000138506183311755,
+      "loss": 1.018,
+      "step": 6892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013849725434171169,
+      "loss": 1.049,
+      "step": 6893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001384883253716684,
+      "loss": 1.0027,
+      "step": 6894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001384793964016251,
+      "loss": 0.9306,
+      "step": 6895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013847046743158177,
+      "loss": 0.948,
+      "step": 6896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 0.9983,
+      "step": 6897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013845260949149518,
+      "loss": 0.9834,
+      "step": 6898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013844368052145186,
+      "loss": 1.1324,
+      "step": 6899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013843475155140856,
+      "loss": 0.9641,
+      "step": 6900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013842582258136524,
+      "loss": 0.9627,
+      "step": 6901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013841689361132194,
+      "loss": 1.0323,
+      "step": 6902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013840796464127864,
+      "loss": 1.0607,
+      "step": 6903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013839903567123532,
+      "loss": 0.9563,
+      "step": 6904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00013839010670119203,
+      "loss": 1.081,
+      "step": 6905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001383811777311487,
+      "loss": 1.08,
+      "step": 6906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001383722487611054,
+      "loss": 1.0554,
+      "step": 6907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001383633197910621,
+      "loss": 1.0138,
+      "step": 6908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001383543908210188,
+      "loss": 1.0679,
+      "step": 6909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001383454618509755,
+      "loss": 0.993,
+      "step": 6910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001383365328809322,
+      "loss": 1.063,
+      "step": 6911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013832760391088887,
+      "loss": 1.0358,
+      "step": 6912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013831867494084558,
+      "loss": 1.0241,
+      "step": 6913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013830974597080228,
+      "loss": 1.0255,
+      "step": 6914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013830081700075896,
+      "loss": 1.0441,
+      "step": 6915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013829188803071566,
+      "loss": 1.0108,
+      "step": 6916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013828295906067237,
+      "loss": 1.0146,
+      "step": 6917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013827403009062904,
+      "loss": 1.1634,
+      "step": 6918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013826510112058575,
+      "loss": 0.8954,
+      "step": 6919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013825617215054245,
+      "loss": 0.9718,
+      "step": 6920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013824724318049913,
+      "loss": 1.0017,
+      "step": 6921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013823831421045583,
+      "loss": 1.1243,
+      "step": 6922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013822938524041254,
+      "loss": 1.0706,
+      "step": 6923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001382204562703692,
+      "loss": 1.0588,
+      "step": 6924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013821152730032592,
+      "loss": 1.0813,
+      "step": 6925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013820259833028262,
+      "loss": 0.9677,
+      "step": 6926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001381936693602393,
+      "loss": 0.9398,
+      "step": 6927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000138184740390196,
+      "loss": 1.0697,
+      "step": 6928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001381758114201527,
+      "loss": 0.9698,
+      "step": 6929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013816688245010938,
+      "loss": 0.9523,
+      "step": 6930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013815795348006609,
+      "loss": 1.0053,
+      "step": 6931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001381490245100228,
+      "loss": 1.0111,
+      "step": 6932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013814009553997947,
+      "loss": 1.0062,
+      "step": 6933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013813116656993617,
+      "loss": 1.1445,
+      "step": 6934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013812223759989288,
+      "loss": 1.0079,
+      "step": 6935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013811330862984955,
+      "loss": 0.9313,
+      "step": 6936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013810437965980626,
+      "loss": 1.0917,
+      "step": 6937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013809545068976296,
+      "loss": 1.0363,
+      "step": 6938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013808652171971964,
+      "loss": 0.9862,
+      "step": 6939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013807759274967634,
+      "loss": 1.1039,
+      "step": 6940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013806866377963305,
+      "loss": 0.9452,
+      "step": 6941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013805973480958972,
+      "loss": 1.1056,
+      "step": 6942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013805080583954643,
+      "loss": 1.2099,
+      "step": 6943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013804187686950313,
+      "loss": 1.0512,
+      "step": 6944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001380329478994598,
+      "loss": 1.025,
+      "step": 6945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001380240189294165,
+      "loss": 1.0715,
+      "step": 6946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001380150899593732,
+      "loss": 0.9988,
+      "step": 6947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001380061609893299,
+      "loss": 1.1593,
+      "step": 6948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013799723201928657,
+      "loss": 1.0054,
+      "step": 6949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013798830304924327,
+      "loss": 1.0837,
+      "step": 6950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013797937407919995,
+      "loss": 0.9691,
+      "step": 6951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013797044510915665,
+      "loss": 0.9106,
+      "step": 6952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013796151613911336,
+      "loss": 1.093,
+      "step": 6953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013795258716907004,
+      "loss": 0.9771,
+      "step": 6954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013794365819902674,
+      "loss": 1.0146,
+      "step": 6955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013793472922898344,
+      "loss": 0.8364,
+      "step": 6956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013792580025894012,
+      "loss": 0.9607,
+      "step": 6957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013791687128889682,
+      "loss": 1.0783,
+      "step": 6958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013790794231885353,
+      "loss": 1.0963,
+      "step": 6959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001378990133488102,
+      "loss": 0.9668,
+      "step": 6960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001378900843787669,
+      "loss": 0.8539,
+      "step": 6961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001378811554087236,
+      "loss": 1.0079,
+      "step": 6962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001378722264386803,
+      "loss": 1.1057,
+      "step": 6963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000137863297468637,
+      "loss": 1.0378,
+      "step": 6964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001378543684985937,
+      "loss": 1.0591,
+      "step": 6965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013784543952855038,
+      "loss": 0.9339,
+      "step": 6966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013783651055850708,
+      "loss": 1.0252,
+      "step": 6967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013782758158846378,
+      "loss": 1.1431,
+      "step": 6968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013781865261842046,
+      "loss": 0.9584,
+      "step": 6969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013780972364837716,
+      "loss": 1.0548,
+      "step": 6970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013780079467833387,
+      "loss": 1.0655,
+      "step": 6971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013779186570829055,
+      "loss": 0.8371,
+      "step": 6972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013778293673824725,
+      "loss": 1.0564,
+      "step": 6973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013777400776820395,
+      "loss": 1.0501,
+      "step": 6974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013776507879816063,
+      "loss": 1.0248,
+      "step": 6975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013775614982811733,
+      "loss": 1.1066,
+      "step": 6976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013774722085807404,
+      "loss": 1.0196,
+      "step": 6977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013773829188803072,
+      "loss": 1.0321,
+      "step": 6978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013772936291798742,
+      "loss": 1.0336,
+      "step": 6979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00013772043394794412,
+      "loss": 0.9595,
+      "step": 6980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001377115049779008,
+      "loss": 1.0243,
+      "step": 6981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001377025760078575,
+      "loss": 1.0624,
+      "step": 6982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001376936470378142,
+      "loss": 0.9406,
+      "step": 6983
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013768471806777089,
+      "loss": 1.0548,
+      "step": 6984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001376757890977276,
+      "loss": 1.0429,
+      "step": 6985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001376668601276843,
+      "loss": 1.0323,
+      "step": 6986
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013765793115764097,
+      "loss": 1.1705,
+      "step": 6987
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013764900218759767,
+      "loss": 1.0583,
+      "step": 6988
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013764007321755438,
+      "loss": 1.0265,
+      "step": 6989
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013763114424751106,
+      "loss": 1.0975,
+      "step": 6990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013762221527746776,
+      "loss": 1.0612,
+      "step": 6991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013761328630742444,
+      "loss": 1.1106,
+      "step": 6992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013760435733738114,
+      "loss": 1.0806,
+      "step": 6993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013759542836733784,
+      "loss": 0.9853,
+      "step": 6994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013758649939729452,
+      "loss": 1.044,
+      "step": 6995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013757757042725123,
+      "loss": 1.2028,
+      "step": 6996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001375686414572079,
+      "loss": 0.9935,
+      "step": 6997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001375597124871646,
+      "loss": 1.1071,
+      "step": 6998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001375507835171213,
+      "loss": 1.1986,
+      "step": 6999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000137541854547078,
+      "loss": 1.0287,
+      "step": 7000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001375329255770347,
+      "loss": 1.0584,
+      "step": 7001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001375239966069914,
+      "loss": 0.9997,
+      "step": 7002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013751506763694807,
+      "loss": 1.0616,
+      "step": 7003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013750613866690478,
+      "loss": 1.0706,
+      "step": 7004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013749720969686148,
+      "loss": 1.1079,
+      "step": 7005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013748828072681816,
+      "loss": 0.9914,
+      "step": 7006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013747935175677486,
+      "loss": 1.0204,
+      "step": 7007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013747042278673156,
+      "loss": 0.9682,
+      "step": 7008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013746149381668824,
+      "loss": 0.9953,
+      "step": 7009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013745256484664495,
+      "loss": 1.1242,
+      "step": 7010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013744363587660165,
+      "loss": 1.1259,
+      "step": 7011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013743470690655833,
+      "loss": 1.2495,
+      "step": 7012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013742577793651503,
+      "loss": 1.2394,
+      "step": 7013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013741684896647173,
+      "loss": 1.0686,
+      "step": 7014
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001374079199964284,
+      "loss": 1.2027,
+      "step": 7015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013739899102638512,
+      "loss": 1.117,
+      "step": 7016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013739006205634182,
+      "loss": 1.0764,
+      "step": 7017
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001373811330862985,
+      "loss": 1.0661,
+      "step": 7018
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001373722041162552,
+      "loss": 0.9603,
+      "step": 7019
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001373632751462119,
+      "loss": 0.9818,
+      "step": 7020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013735434617616858,
+      "loss": 0.9444,
+      "step": 7021
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013734541720612529,
+      "loss": 1.0844,
+      "step": 7022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000137336488236082,
+      "loss": 1.0233,
+      "step": 7023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013732755926603867,
+      "loss": 1.0403,
+      "step": 7024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013731863029599537,
+      "loss": 1.0076,
+      "step": 7025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013730970132595207,
+      "loss": 1.016,
+      "step": 7026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013730077235590875,
+      "loss": 1.0002,
+      "step": 7027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013729184338586546,
+      "loss": 1.0255,
+      "step": 7028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013728291441582216,
+      "loss": 1.113,
+      "step": 7029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013727398544577884,
+      "loss": 0.9549,
+      "step": 7030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013726505647573554,
+      "loss": 1.1372,
+      "step": 7031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013725612750569224,
+      "loss": 1.0909,
+      "step": 7032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013724719853564892,
+      "loss": 1.0499,
+      "step": 7033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013723826956560563,
+      "loss": 1.0106,
+      "step": 7034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013722934059556233,
+      "loss": 0.9856,
+      "step": 7035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000137220411625519,
+      "loss": 1.1494,
+      "step": 7036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001372114826554757,
+      "loss": 1.1416,
+      "step": 7037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001372025536854324,
+      "loss": 0.9031,
+      "step": 7038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001371936247153891,
+      "loss": 1.0391,
+      "step": 7039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013718469574534577,
+      "loss": 1.0251,
+      "step": 7040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013717576677530247,
+      "loss": 1.1671,
+      "step": 7041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013716683780525915,
+      "loss": 1.1371,
+      "step": 7042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013715790883521585,
+      "loss": 1.0753,
+      "step": 7043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013714897986517256,
+      "loss": 1.0578,
+      "step": 7044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013714005089512923,
+      "loss": 1.0025,
+      "step": 7045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013713112192508594,
+      "loss": 0.9237,
+      "step": 7046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013712219295504264,
+      "loss": 0.8938,
+      "step": 7047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013711326398499932,
+      "loss": 1.0698,
+      "step": 7048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013710433501495602,
+      "loss": 1.1883,
+      "step": 7049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013709540604491273,
+      "loss": 1.0533,
+      "step": 7050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001370864770748694,
+      "loss": 0.9814,
+      "step": 7051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001370775481048261,
+      "loss": 1.0348,
+      "step": 7052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001370686191347828,
+      "loss": 0.9474,
+      "step": 7053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001370596901647395,
+      "loss": 0.9542,
+      "step": 7054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001370507611946962,
+      "loss": 1.0395,
+      "step": 7055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001370418322246529,
+      "loss": 0.8833,
+      "step": 7056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00013703290325460957,
+      "loss": 0.9768,
+      "step": 7057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013702397428456628,
+      "loss": 1.1629,
+      "step": 7058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013701504531452298,
+      "loss": 1.0568,
+      "step": 7059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013700611634447966,
+      "loss": 1.0204,
+      "step": 7060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013699718737443636,
+      "loss": 0.9188,
+      "step": 7061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013698825840439307,
+      "loss": 1.0009,
+      "step": 7062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013697932943434974,
+      "loss": 1.0559,
+      "step": 7063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013697040046430645,
+      "loss": 1.0445,
+      "step": 7064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013696147149426315,
+      "loss": 1.1097,
+      "step": 7065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013695254252421983,
+      "loss": 1.0466,
+      "step": 7066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013694361355417653,
+      "loss": 0.9368,
+      "step": 7067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013693468458413324,
+      "loss": 1.1922,
+      "step": 7068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013692575561408991,
+      "loss": 1.1026,
+      "step": 7069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013691682664404662,
+      "loss": 1.0021,
+      "step": 7070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013690789767400332,
+      "loss": 1.1687,
+      "step": 7071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013689896870396,
+      "loss": 1.1398,
+      "step": 7072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001368900397339167,
+      "loss": 1.0942,
+      "step": 7073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001368811107638734,
+      "loss": 0.9549,
+      "step": 7074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013687218179383008,
+      "loss": 1.0152,
+      "step": 7075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001368632528237868,
+      "loss": 1.1098,
+      "step": 7076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001368543238537435,
+      "loss": 1.0324,
+      "step": 7077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013684539488370017,
+      "loss": 1.0277,
+      "step": 7078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013683646591365687,
+      "loss": 0.9332,
+      "step": 7079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013682753694361358,
+      "loss": 1.0029,
+      "step": 7080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013681860797357025,
+      "loss": 1.0654,
+      "step": 7081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013680967900352696,
+      "loss": 1.106,
+      "step": 7082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013680075003348364,
+      "loss": 1.052,
+      "step": 7083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013679182106344034,
+      "loss": 1.1864,
+      "step": 7084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013678289209339704,
+      "loss": 1.0205,
+      "step": 7085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013677396312335372,
+      "loss": 1.1683,
+      "step": 7086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013676503415331042,
+      "loss": 1.0725,
+      "step": 7087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001367561051832671,
+      "loss": 1.0156,
+      "step": 7088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001367471762132238,
+      "loss": 1.0341,
+      "step": 7089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013673824724318048,
+      "loss": 0.9825,
+      "step": 7090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001367293182731372,
+      "loss": 0.9719,
+      "step": 7091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001367203893030939,
+      "loss": 1.0802,
+      "step": 7092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013671146033305057,
+      "loss": 1.0704,
+      "step": 7093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013670253136300727,
+      "loss": 1.1358,
+      "step": 7094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013669360239296398,
+      "loss": 1.1077,
+      "step": 7095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013668467342292065,
+      "loss": 1.046,
+      "step": 7096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013667574445287736,
+      "loss": 0.9433,
+      "step": 7097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013666681548283406,
+      "loss": 1.1301,
+      "step": 7098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013665788651279076,
+      "loss": 1.0232,
+      "step": 7099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013664895754274744,
+      "loss": 0.9844,
+      "step": 7100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013664002857270415,
+      "loss": 1.1096,
+      "step": 7101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013663109960266085,
+      "loss": 1.0062,
+      "step": 7102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013662217063261753,
+      "loss": 1.2293,
+      "step": 7103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013661324166257423,
+      "loss": 0.9896,
+      "step": 7104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013660431269253093,
+      "loss": 0.9408,
+      "step": 7105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001365953837224876,
+      "loss": 1.142,
+      "step": 7106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013658645475244432,
+      "loss": 0.9851,
+      "step": 7107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013657752578240102,
+      "loss": 0.972,
+      "step": 7108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001365685968123577,
+      "loss": 0.9407,
+      "step": 7109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001365596678423144,
+      "loss": 1.0954,
+      "step": 7110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001365507388722711,
+      "loss": 0.9924,
+      "step": 7111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013654180990222778,
+      "loss": 1.0851,
+      "step": 7112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013653288093218449,
+      "loss": 1.0084,
+      "step": 7113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001365239519621412,
+      "loss": 0.984,
+      "step": 7114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013651502299209787,
+      "loss": 1.0722,
+      "step": 7115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013650609402205457,
+      "loss": 1.1047,
+      "step": 7116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013649716505201127,
+      "loss": 0.966,
+      "step": 7117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013648823608196795,
+      "loss": 1.1555,
+      "step": 7118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013647930711192466,
+      "loss": 1.1358,
+      "step": 7119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013647037814188136,
+      "loss": 1.0371,
+      "step": 7120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013646144917183804,
+      "loss": 1.1819,
+      "step": 7121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013645252020179474,
+      "loss": 1.0979,
+      "step": 7122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013644359123175144,
+      "loss": 1.0917,
+      "step": 7123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013643466226170812,
+      "loss": 1.0908,
+      "step": 7124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013642573329166483,
+      "loss": 0.9685,
+      "step": 7125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013641680432162153,
+      "loss": 1.0403,
+      "step": 7126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001364078753515782,
+      "loss": 0.9915,
+      "step": 7127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001363989463815349,
+      "loss": 1.0155,
+      "step": 7128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001363900174114916,
+      "loss": 1.0711,
+      "step": 7129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001363810884414483,
+      "loss": 1.0869,
+      "step": 7130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013637215947140497,
+      "loss": 1.172,
+      "step": 7131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00013636323050136167,
+      "loss": 0.9688,
+      "step": 7132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013635430153131835,
+      "loss": 1.0071,
+      "step": 7133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013634537256127505,
+      "loss": 0.9115,
+      "step": 7134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013633644359123176,
+      "loss": 1.0766,
+      "step": 7135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013632751462118843,
+      "loss": 1.019,
+      "step": 7136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013631858565114514,
+      "loss": 1.0188,
+      "step": 7137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013630965668110184,
+      "loss": 0.9869,
+      "step": 7138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013630072771105852,
+      "loss": 1.1078,
+      "step": 7139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013629179874101522,
+      "loss": 1.0347,
+      "step": 7140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013628286977097193,
+      "loss": 1.0737,
+      "step": 7141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001362739408009286,
+      "loss": 1.1072,
+      "step": 7142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001362650118308853,
+      "loss": 1.1334,
+      "step": 7143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000136256082860842,
+      "loss": 0.9901,
+      "step": 7144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001362471538907987,
+      "loss": 0.9742,
+      "step": 7145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001362382249207554,
+      "loss": 0.9949,
+      "step": 7146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001362292959507121,
+      "loss": 1.1691,
+      "step": 7147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013622036698066877,
+      "loss": 1.0923,
+      "step": 7148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013621143801062548,
+      "loss": 1.0821,
+      "step": 7149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013620250904058218,
+      "loss": 1.084,
+      "step": 7150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013619358007053886,
+      "loss": 0.9979,
+      "step": 7151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013618465110049556,
+      "loss": 0.9901,
+      "step": 7152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013617572213045227,
+      "loss": 0.985,
+      "step": 7153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013616679316040894,
+      "loss": 0.9877,
+      "step": 7154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013615786419036565,
+      "loss": 1.0133,
+      "step": 7155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013614893522032235,
+      "loss": 1.0906,
+      "step": 7156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013614000625027903,
+      "loss": 0.982,
+      "step": 7157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013613107728023573,
+      "loss": 1.0125,
+      "step": 7158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013612214831019244,
+      "loss": 1.0534,
+      "step": 7159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013611321934014911,
+      "loss": 1.0199,
+      "step": 7160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013610429037010582,
+      "loss": 1.0669,
+      "step": 7161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013609536140006252,
+      "loss": 1.0363,
+      "step": 7162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001360864324300192,
+      "loss": 0.9613,
+      "step": 7163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001360775034599759,
+      "loss": 1.0364,
+      "step": 7164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001360685744899326,
+      "loss": 1.0453,
+      "step": 7165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013605964551988928,
+      "loss": 0.9967,
+      "step": 7166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000136050716549846,
+      "loss": 1.036,
+      "step": 7167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001360417875798027,
+      "loss": 0.9883,
+      "step": 7168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013603285860975937,
+      "loss": 1.0007,
+      "step": 7169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013602392963971607,
+      "loss": 1.0865,
+      "step": 7170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013601500066967278,
+      "loss": 1.505,
+      "step": 7171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013600607169962945,
+      "loss": 1.0586,
+      "step": 7172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013599714272958616,
+      "loss": 0.9903,
+      "step": 7173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013598821375954284,
+      "loss": 0.976,
+      "step": 7174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013597928478949954,
+      "loss": 1.1234,
+      "step": 7175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013597035581945624,
+      "loss": 0.8717,
+      "step": 7176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013596142684941292,
+      "loss": 1.1385,
+      "step": 7177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013595249787936962,
+      "loss": 1.1636,
+      "step": 7178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001359435689093263,
+      "loss": 1.1042,
+      "step": 7179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000135934639939283,
+      "loss": 0.918,
+      "step": 7180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013592571096923968,
+      "loss": 0.9744,
+      "step": 7181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013591678199919639,
+      "loss": 1.0892,
+      "step": 7182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001359078530291531,
+      "loss": 0.9829,
+      "step": 7183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013589892405910977,
+      "loss": 1.0943,
+      "step": 7184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013588999508906647,
+      "loss": 0.9393,
+      "step": 7185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013588106611902318,
+      "loss": 0.9196,
+      "step": 7186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013587213714897985,
+      "loss": 0.9988,
+      "step": 7187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013586320817893656,
+      "loss": 0.8243,
+      "step": 7188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013585427920889326,
+      "loss": 1.0699,
+      "step": 7189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013584535023884994,
+      "loss": 1.0475,
+      "step": 7190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013583642126880664,
+      "loss": 1.168,
+      "step": 7191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013582749229876335,
+      "loss": 1.0536,
+      "step": 7192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013581856332872002,
+      "loss": 1.124,
+      "step": 7193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013580963435867673,
+      "loss": 1.0264,
+      "step": 7194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013580070538863343,
+      "loss": 1.035,
+      "step": 7195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013579177641859013,
+      "loss": 1.0693,
+      "step": 7196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001357828474485468,
+      "loss": 1.0427,
+      "step": 7197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013577391847850352,
+      "loss": 1.0403,
+      "step": 7198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013576498950846022,
+      "loss": 1.0445,
+      "step": 7199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001357560605384169,
+      "loss": 0.982,
+      "step": 7200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001357471315683736,
+      "loss": 1.0525,
+      "step": 7201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001357382025983303,
+      "loss": 1.1459,
+      "step": 7202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013572927362828698,
+      "loss": 1.051,
+      "step": 7203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013572034465824369,
+      "loss": 1.1068,
+      "step": 7204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001357114156882004,
+      "loss": 1.1343,
+      "step": 7205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013570248671815707,
+      "loss": 1.0055,
+      "step": 7206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00013569355774811377,
+      "loss": 1.1026,
+      "step": 7207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013568462877807047,
+      "loss": 1.0567,
+      "step": 7208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013567569980802715,
+      "loss": 1.1557,
+      "step": 7209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013566677083798386,
+      "loss": 0.9868,
+      "step": 7210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013565784186794056,
+      "loss": 1.0035,
+      "step": 7211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013564891289789724,
+      "loss": 0.9675,
+      "step": 7212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013563998392785394,
+      "loss": 0.9672,
+      "step": 7213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013563105495781064,
+      "loss": 0.9771,
+      "step": 7214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013562212598776732,
+      "loss": 1.1228,
+      "step": 7215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013561319701772403,
+      "loss": 1.1885,
+      "step": 7216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013560426804768073,
+      "loss": 1.1251,
+      "step": 7217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001355953390776374,
+      "loss": 1.0011,
+      "step": 7218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001355864101075941,
+      "loss": 1.0203,
+      "step": 7219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001355774811375508,
+      "loss": 1.0617,
+      "step": 7220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001355685521675075,
+      "loss": 1.2122,
+      "step": 7221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013555962319746417,
+      "loss": 1.0126,
+      "step": 7222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013555069422742087,
+      "loss": 1.1885,
+      "step": 7223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013554176525737755,
+      "loss": 0.9164,
+      "step": 7224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013553283628733425,
+      "loss": 0.9008,
+      "step": 7225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013552390731729096,
+      "loss": 1.0371,
+      "step": 7226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013551497834724763,
+      "loss": 1.096,
+      "step": 7227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013550604937720434,
+      "loss": 0.9651,
+      "step": 7228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013549712040716104,
+      "loss": 1.0337,
+      "step": 7229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013548819143711772,
+      "loss": 1.0187,
+      "step": 7230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013547926246707442,
+      "loss": 1.0122,
+      "step": 7231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013547033349703113,
+      "loss": 0.9593,
+      "step": 7232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001354614045269878,
+      "loss": 1.0007,
+      "step": 7233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001354524755569445,
+      "loss": 0.9918,
+      "step": 7234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001354435465869012,
+      "loss": 0.9583,
+      "step": 7235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001354346176168579,
+      "loss": 1.0417,
+      "step": 7236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001354256886468146,
+      "loss": 0.9438,
+      "step": 7237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001354167596767713,
+      "loss": 0.9727,
+      "step": 7238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013540783070672797,
+      "loss": 1.2376,
+      "step": 7239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013539890173668468,
+      "loss": 1.1301,
+      "step": 7240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013538997276664138,
+      "loss": 1.1892,
+      "step": 7241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013538104379659806,
+      "loss": 1.0355,
+      "step": 7242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013537211482655476,
+      "loss": 1.0973,
+      "step": 7243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013536318585651147,
+      "loss": 0.9804,
+      "step": 7244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013535425688646814,
+      "loss": 1.0122,
+      "step": 7245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013534532791642485,
+      "loss": 0.8708,
+      "step": 7246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013533639894638155,
+      "loss": 0.9997,
+      "step": 7247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013532746997633823,
+      "loss": 1.1924,
+      "step": 7248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013531854100629493,
+      "loss": 1.1696,
+      "step": 7249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013530961203625164,
+      "loss": 1.0274,
+      "step": 7250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013530068306620831,
+      "loss": 1.1287,
+      "step": 7251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013529175409616502,
+      "loss": 0.9674,
+      "step": 7252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013528282512612172,
+      "loss": 1.0233,
+      "step": 7253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001352738961560784,
+      "loss": 1.0299,
+      "step": 7254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001352649671860351,
+      "loss": 1.0935,
+      "step": 7255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001352560382159918,
+      "loss": 1.0626,
+      "step": 7256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013524710924594848,
+      "loss": 0.9275,
+      "step": 7257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001352381802759052,
+      "loss": 1.0066,
+      "step": 7258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001352292513058619,
+      "loss": 1.121,
+      "step": 7259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013522032233581857,
+      "loss": 1.139,
+      "step": 7260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013521139336577527,
+      "loss": 0.9864,
+      "step": 7261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013520246439573198,
+      "loss": 1.0953,
+      "step": 7262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013519353542568865,
+      "loss": 1.0071,
+      "step": 7263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013518460645564536,
+      "loss": 1.0075,
+      "step": 7264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013517567748560203,
+      "loss": 0.9944,
+      "step": 7265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013516674851555874,
+      "loss": 0.9839,
+      "step": 7266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013515781954551544,
+      "loss": 0.9857,
+      "step": 7267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013514889057547212,
+      "loss": 0.9643,
+      "step": 7268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013513996160542882,
+      "loss": 1.0455,
+      "step": 7269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001351310326353855,
+      "loss": 1.1872,
+      "step": 7270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001351221036653422,
+      "loss": 1.2456,
+      "step": 7271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013511317469529888,
+      "loss": 0.9978,
+      "step": 7272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013510424572525559,
+      "loss": 1.1016,
+      "step": 7273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001350953167552123,
+      "loss": 1.0287,
+      "step": 7274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013508638778516897,
+      "loss": 1.1142,
+      "step": 7275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013507745881512567,
+      "loss": 1.1027,
+      "step": 7276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013506852984508237,
+      "loss": 1.1626,
+      "step": 7277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013505960087503905,
+      "loss": 0.93,
+      "step": 7278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013505067190499576,
+      "loss": 0.9741,
+      "step": 7279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013504174293495246,
+      "loss": 1.0081,
+      "step": 7280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00013503281396490914,
+      "loss": 0.92,
+      "step": 7281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013502388499486584,
+      "loss": 1.0077,
+      "step": 7282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013501495602482254,
+      "loss": 1.1109,
+      "step": 7283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013500602705477922,
+      "loss": 1.1914,
+      "step": 7284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013499709808473593,
+      "loss": 1.0848,
+      "step": 7285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013498816911469263,
+      "loss": 0.9556,
+      "step": 7286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001349792401446493,
+      "loss": 0.9188,
+      "step": 7287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000134970311174606,
+      "loss": 1.0128,
+      "step": 7288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013496138220456271,
+      "loss": 1.1479,
+      "step": 7289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001349524532345194,
+      "loss": 1.0457,
+      "step": 7290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001349435242644761,
+      "loss": 1.0807,
+      "step": 7291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001349345952944328,
+      "loss": 1.0698,
+      "step": 7292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001349256663243895,
+      "loss": 1.0915,
+      "step": 7293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013491673735434618,
+      "loss": 1.0433,
+      "step": 7294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013490780838430288,
+      "loss": 0.9129,
+      "step": 7295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001348988794142596,
+      "loss": 1.0799,
+      "step": 7296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013488995044421627,
+      "loss": 1.0691,
+      "step": 7297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013488102147417297,
+      "loss": 0.8815,
+      "step": 7298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013487209250412967,
+      "loss": 1.0034,
+      "step": 7299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013486316353408635,
+      "loss": 1.0328,
+      "step": 7300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013485423456404305,
+      "loss": 0.9511,
+      "step": 7301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013484530559399976,
+      "loss": 1.012,
+      "step": 7302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013483637662395644,
+      "loss": 1.0132,
+      "step": 7303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013482744765391314,
+      "loss": 1.042,
+      "step": 7304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013481851868386984,
+      "loss": 1.0807,
+      "step": 7305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013480958971382652,
+      "loss": 1.0569,
+      "step": 7306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013480066074378322,
+      "loss": 1.041,
+      "step": 7307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013479173177373993,
+      "loss": 0.948,
+      "step": 7308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001347828028036966,
+      "loss": 1.1374,
+      "step": 7309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001347738738336533,
+      "loss": 0.9235,
+      "step": 7310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013476494486361,
+      "loss": 1.156,
+      "step": 7311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001347560158935667,
+      "loss": 1.029,
+      "step": 7312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013474708692352337,
+      "loss": 1.1986,
+      "step": 7313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013473815795348007,
+      "loss": 0.8906,
+      "step": 7314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013472922898343675,
+      "loss": 0.9843,
+      "step": 7315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013472030001339345,
+      "loss": 0.999,
+      "step": 7316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013471137104335016,
+      "loss": 1.1361,
+      "step": 7317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013470244207330683,
+      "loss": 1.0079,
+      "step": 7318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013469351310326354,
+      "loss": 1.0722,
+      "step": 7319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013468458413322024,
+      "loss": 1.1207,
+      "step": 7320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013467565516317692,
+      "loss": 1.0826,
+      "step": 7321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013466672619313362,
+      "loss": 1.0057,
+      "step": 7322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013465779722309033,
+      "loss": 0.9611,
+      "step": 7323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000134648868253047,
+      "loss": 1.1132,
+      "step": 7324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001346399392830037,
+      "loss": 0.9484,
+      "step": 7325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001346310103129604,
+      "loss": 1.0202,
+      "step": 7326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001346220813429171,
+      "loss": 1.116,
+      "step": 7327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001346131523728738,
+      "loss": 0.9271,
+      "step": 7328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001346042234028305,
+      "loss": 1.0396,
+      "step": 7329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013459529443278717,
+      "loss": 1.0062,
+      "step": 7330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013458636546274388,
+      "loss": 1.0072,
+      "step": 7331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013457743649270058,
+      "loss": 1.0648,
+      "step": 7332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013456850752265726,
+      "loss": 1.0161,
+      "step": 7333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013455957855261396,
+      "loss": 1.0505,
+      "step": 7334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013455064958257067,
+      "loss": 0.9844,
+      "step": 7335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013454172061252734,
+      "loss": 1.0799,
+      "step": 7336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013453279164248405,
+      "loss": 1.1014,
+      "step": 7337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013452386267244075,
+      "loss": 0.9244,
+      "step": 7338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013451493370239743,
+      "loss": 1.0113,
+      "step": 7339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013450600473235413,
+      "loss": 1.0999,
+      "step": 7340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013449707576231084,
+      "loss": 1.0169,
+      "step": 7341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013448814679226751,
+      "loss": 1.0071,
+      "step": 7342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013447921782222422,
+      "loss": 0.9541,
+      "step": 7343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013447028885218092,
+      "loss": 1.1982,
+      "step": 7344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001344613598821376,
+      "loss": 1.1083,
+      "step": 7345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001344524309120943,
+      "loss": 1.1025,
+      "step": 7346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000134443501942051,
+      "loss": 0.9868,
+      "step": 7347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013443457297200768,
+      "loss": 1.0773,
+      "step": 7348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001344256440019644,
+      "loss": 1.1038,
+      "step": 7349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001344167150319211,
+      "loss": 1.0902,
+      "step": 7350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013440778606187777,
+      "loss": 1.0016,
+      "step": 7351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013439885709183447,
+      "loss": 1.0053,
+      "step": 7352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013438992812179118,
+      "loss": 1.0878,
+      "step": 7353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013438099915174785,
+      "loss": 1.0447,
+      "step": 7354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013437207018170456,
+      "loss": 1.1418,
+      "step": 7355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00013436314121166123,
+      "loss": 1.0357,
+      "step": 7356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013435421224161794,
+      "loss": 1.0927,
+      "step": 7357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013434528327157462,
+      "loss": 1.1253,
+      "step": 7358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013433635430153132,
+      "loss": 1.0098,
+      "step": 7359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013432742533148802,
+      "loss": 0.9875,
+      "step": 7360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001343184963614447,
+      "loss": 1.084,
+      "step": 7361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001343095673914014,
+      "loss": 1.1252,
+      "step": 7362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013430063842135808,
+      "loss": 1.009,
+      "step": 7363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013429170945131479,
+      "loss": 1.1481,
+      "step": 7364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001342827804812715,
+      "loss": 1.1005,
+      "step": 7365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013427385151122817,
+      "loss": 0.9279,
+      "step": 7366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013426492254118487,
+      "loss": 1.037,
+      "step": 7367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013425599357114157,
+      "loss": 1.0149,
+      "step": 7368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013424706460109825,
+      "loss": 1.0527,
+      "step": 7369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013423813563105496,
+      "loss": 1.1215,
+      "step": 7370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013422920666101166,
+      "loss": 1.09,
+      "step": 7371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013422027769096834,
+      "loss": 0.9614,
+      "step": 7372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013421134872092504,
+      "loss": 1.0021,
+      "step": 7373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013420241975088174,
+      "loss": 0.9336,
+      "step": 7374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013419349078083842,
+      "loss": 1.12,
+      "step": 7375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013418456181079513,
+      "loss": 1.0025,
+      "step": 7376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013417563284075183,
+      "loss": 0.9479,
+      "step": 7377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001341667038707085,
+      "loss": 1.0163,
+      "step": 7378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001341577749006652,
+      "loss": 1.0143,
+      "step": 7379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013414884593062191,
+      "loss": 0.9765,
+      "step": 7380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001341399169605786,
+      "loss": 1.0895,
+      "step": 7381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001341309879905353,
+      "loss": 1.0135,
+      "step": 7382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000134122059020492,
+      "loss": 0.9617,
+      "step": 7383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013411313005044868,
+      "loss": 1.1094,
+      "step": 7384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013410420108040538,
+      "loss": 0.8682,
+      "step": 7385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013409527211036208,
+      "loss": 1.0912,
+      "step": 7386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013408634314031876,
+      "loss": 1.0011,
+      "step": 7387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013407741417027547,
+      "loss": 0.9036,
+      "step": 7388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013406848520023217,
+      "loss": 1.2064,
+      "step": 7389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013405955623018887,
+      "loss": 0.9817,
+      "step": 7390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013405062726014555,
+      "loss": 1.0699,
+      "step": 7391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013404169829010225,
+      "loss": 0.9173,
+      "step": 7392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013403276932005896,
+      "loss": 1.1086,
+      "step": 7393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013402384035001564,
+      "loss": 1.0002,
+      "step": 7394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013401491137997234,
+      "loss": 1.0192,
+      "step": 7395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013400598240992904,
+      "loss": 1.0856,
+      "step": 7396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013399705343988572,
+      "loss": 1.0863,
+      "step": 7397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013398812446984242,
+      "loss": 1.1736,
+      "step": 7398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013397919549979913,
+      "loss": 1.1081,
+      "step": 7399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001339702665297558,
+      "loss": 1.0034,
+      "step": 7400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001339613375597125,
+      "loss": 1.0735,
+      "step": 7401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013395240858966919,
+      "loss": 1.0717,
+      "step": 7402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001339434796196259,
+      "loss": 1.0438,
+      "step": 7403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013393455064958257,
+      "loss": 1.0613,
+      "step": 7404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013392562167953927,
+      "loss": 0.9931,
+      "step": 7405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013391669270949595,
+      "loss": 0.9221,
+      "step": 7406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013390776373945265,
+      "loss": 1.0169,
+      "step": 7407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013389883476940936,
+      "loss": 0.9961,
+      "step": 7408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013388990579936603,
+      "loss": 1.0185,
+      "step": 7409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013388097682932274,
+      "loss": 0.9875,
+      "step": 7410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013387204785927944,
+      "loss": 1.1939,
+      "step": 7411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013386311888923612,
+      "loss": 1.0268,
+      "step": 7412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013385418991919282,
+      "loss": 0.9374,
+      "step": 7413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013384526094914953,
+      "loss": 1.0479,
+      "step": 7414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001338363319791062,
+      "loss": 1.0225,
+      "step": 7415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001338274030090629,
+      "loss": 0.9456,
+      "step": 7416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001338184740390196,
+      "loss": 1.0245,
+      "step": 7417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001338095450689763,
+      "loss": 0.9957,
+      "step": 7418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000133800616098933,
+      "loss": 0.9636,
+      "step": 7419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001337916871288897,
+      "loss": 1.1898,
+      "step": 7420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013378275815884637,
+      "loss": 1.1231,
+      "step": 7421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013377382918880308,
+      "loss": 1.1044,
+      "step": 7422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013376490021875978,
+      "loss": 1.0054,
+      "step": 7423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013375597124871646,
+      "loss": 1.1118,
+      "step": 7424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013374704227867316,
+      "loss": 1.083,
+      "step": 7425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013373811330862987,
+      "loss": 1.089,
+      "step": 7426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013372918433858654,
+      "loss": 0.9204,
+      "step": 7427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013372025536854325,
+      "loss": 1.0581,
+      "step": 7428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013371132639849995,
+      "loss": 0.9362,
+      "step": 7429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013370239742845663,
+      "loss": 0.8887,
+      "step": 7430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00013369346845841333,
+      "loss": 1.079,
+      "step": 7431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013368453948837004,
+      "loss": 0.8989,
+      "step": 7432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001336756105183267,
+      "loss": 1.0655,
+      "step": 7433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013366668154828342,
+      "loss": 0.9719,
+      "step": 7434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013365775257824012,
+      "loss": 0.9966,
+      "step": 7435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001336488236081968,
+      "loss": 1.0388,
+      "step": 7436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001336398946381535,
+      "loss": 1.0434,
+      "step": 7437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001336309656681102,
+      "loss": 1.1059,
+      "step": 7438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013362203669806688,
+      "loss": 1.0375,
+      "step": 7439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001336131077280236,
+      "loss": 0.98,
+      "step": 7440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001336041787579803,
+      "loss": 1.019,
+      "step": 7441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013359524978793697,
+      "loss": 1.0407,
+      "step": 7442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013358632081789367,
+      "loss": 1.0626,
+      "step": 7443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013357739184785038,
+      "loss": 1.0532,
+      "step": 7444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013356846287780705,
+      "loss": 1.007,
+      "step": 7445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013355953390776376,
+      "loss": 1.0139,
+      "step": 7446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013355060493772043,
+      "loss": 1.0471,
+      "step": 7447
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013354167596767714,
+      "loss": 1.0688,
+      "step": 7448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013353274699763382,
+      "loss": 1.1067,
+      "step": 7449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013352381802759052,
+      "loss": 1.1304,
+      "step": 7450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013351488905754722,
+      "loss": 0.9402,
+      "step": 7451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001335059600875039,
+      "loss": 1.0092,
+      "step": 7452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001334970311174606,
+      "loss": 1.1452,
+      "step": 7453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013348810214741728,
+      "loss": 1.1159,
+      "step": 7454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013347917317737399,
+      "loss": 1.0863,
+      "step": 7455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001334702442073307,
+      "loss": 1.0752,
+      "step": 7456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013346131523728737,
+      "loss": 1.0815,
+      "step": 7457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013345238626724407,
+      "loss": 1.1133,
+      "step": 7458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013344345729720077,
+      "loss": 0.9917,
+      "step": 7459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013343452832715745,
+      "loss": 1.0238,
+      "step": 7460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013342559935711416,
+      "loss": 1.0754,
+      "step": 7461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013341667038707086,
+      "loss": 1.0503,
+      "step": 7462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013340774141702754,
+      "loss": 0.9515,
+      "step": 7463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013339881244698424,
+      "loss": 0.8987,
+      "step": 7464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013338988347694094,
+      "loss": 1.0288,
+      "step": 7465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013338095450689762,
+      "loss": 1.0504,
+      "step": 7466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013337202553685433,
+      "loss": 1.0718,
+      "step": 7467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013336309656681103,
+      "loss": 1.192,
+      "step": 7468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001333541675967677,
+      "loss": 0.8932,
+      "step": 7469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001333452386267244,
+      "loss": 0.9893,
+      "step": 7470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013333630965668111,
+      "loss": 0.9959,
+      "step": 7471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001333273806866378,
+      "loss": 1.0916,
+      "step": 7472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001333184517165945,
+      "loss": 1.0413,
+      "step": 7473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001333095227465512,
+      "loss": 1.0701,
+      "step": 7474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013330059377650788,
+      "loss": 0.9826,
+      "step": 7475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013329166480646458,
+      "loss": 1.0134,
+      "step": 7476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013328273583642128,
+      "loss": 1.0326,
+      "step": 7477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013327380686637796,
+      "loss": 1.0633,
+      "step": 7478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013326487789633467,
+      "loss": 1.0948,
+      "step": 7479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013325594892629137,
+      "loss": 0.9889,
+      "step": 7480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013324701995624805,
+      "loss": 1.0013,
+      "step": 7481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013323809098620475,
+      "loss": 1.0386,
+      "step": 7482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013322916201616145,
+      "loss": 0.9054,
+      "step": 7483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013322023304611813,
+      "loss": 0.9575,
+      "step": 7484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013321130407607484,
+      "loss": 1.0628,
+      "step": 7485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013320237510603154,
+      "loss": 0.982,
+      "step": 7486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013319344613598822,
+      "loss": 0.9998,
+      "step": 7487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013318451716594492,
+      "loss": 1.0985,
+      "step": 7488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013317558819590162,
+      "loss": 0.8862,
+      "step": 7489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013316665922585833,
+      "loss": 1.2213,
+      "step": 7490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000133157730255815,
+      "loss": 1.1082,
+      "step": 7491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001331488012857717,
+      "loss": 0.9131,
+      "step": 7492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013313987231572839,
+      "loss": 1.1547,
+      "step": 7493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001331309433456851,
+      "loss": 1.1842,
+      "step": 7494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013312201437564177,
+      "loss": 1.0066,
+      "step": 7495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013311308540559847,
+      "loss": 0.9725,
+      "step": 7496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013310415643555515,
+      "loss": 0.9996,
+      "step": 7497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013309522746551185,
+      "loss": 1.0555,
+      "step": 7498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013308629849546856,
+      "loss": 1.0587,
+      "step": 7499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013307736952542523,
+      "loss": 0.9823,
+      "step": 7500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013306844055538194,
+      "loss": 1.0764,
+      "step": 7501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013305951158533864,
+      "loss": 1.0035,
+      "step": 7502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013305058261529532,
+      "loss": 1.0566,
+      "step": 7503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013304165364525202,
+      "loss": 1.0245,
+      "step": 7504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00013303272467520873,
+      "loss": 1.058,
+      "step": 7505
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001330237957051654,
+      "loss": 0.9789,
+      "step": 7506
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001330148667351221,
+      "loss": 1.0408,
+      "step": 7507
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001330059377650788,
+      "loss": 1.1193,
+      "step": 7508
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001329970087950355,
+      "loss": 1.0335,
+      "step": 7509
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001329880798249922,
+      "loss": 1.0625,
+      "step": 7510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001329791508549489,
+      "loss": 1.0038,
+      "step": 7511
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013297022188490557,
+      "loss": 0.9718,
+      "step": 7512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013296129291486228,
+      "loss": 1.0215,
+      "step": 7513
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013295236394481898,
+      "loss": 1.029,
+      "step": 7514
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013294343497477566,
+      "loss": 1.0397,
+      "step": 7515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013293450600473236,
+      "loss": 1.0338,
+      "step": 7516
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013292557703468907,
+      "loss": 1.459,
+      "step": 7517
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013291664806464574,
+      "loss": 1.1045,
+      "step": 7518
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013290771909460245,
+      "loss": 1.0922,
+      "step": 7519
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013289879012455915,
+      "loss": 1.0163,
+      "step": 7520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013288986115451583,
+      "loss": 1.1617,
+      "step": 7521
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013288093218447253,
+      "loss": 1.1011,
+      "step": 7522
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013287200321442924,
+      "loss": 1.0521,
+      "step": 7523
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001328630742443859,
+      "loss": 1.0082,
+      "step": 7524
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013285414527434262,
+      "loss": 1.0246,
+      "step": 7525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013284521630429932,
+      "loss": 1.1787,
+      "step": 7526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000132836287334256,
+      "loss": 1.0355,
+      "step": 7527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001328273583642127,
+      "loss": 1.0029,
+      "step": 7528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001328184293941694,
+      "loss": 1.0494,
+      "step": 7529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013280950042412608,
+      "loss": 1.1324,
+      "step": 7530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001328005714540828,
+      "loss": 0.991,
+      "step": 7531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001327916424840395,
+      "loss": 1.0405,
+      "step": 7532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013278271351399617,
+      "loss": 1.0423,
+      "step": 7533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013277378454395287,
+      "loss": 0.9757,
+      "step": 7534
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013276485557390958,
+      "loss": 1.0097,
+      "step": 7535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013275592660386625,
+      "loss": 0.9248,
+      "step": 7536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013274699763382296,
+      "loss": 1.0014,
+      "step": 7537
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013273806866377963,
+      "loss": 0.9456,
+      "step": 7538
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013272913969373634,
+      "loss": 1.0047,
+      "step": 7539
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013272021072369301,
+      "loss": 0.9589,
+      "step": 7540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013271128175364972,
+      "loss": 1.0712,
+      "step": 7541
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013270235278360642,
+      "loss": 1.1743,
+      "step": 7542
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001326934238135631,
+      "loss": 0.9119,
+      "step": 7543
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001326844948435198,
+      "loss": 0.9809,
+      "step": 7544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013267556587347648,
+      "loss": 1.0678,
+      "step": 7545
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013266663690343318,
+      "loss": 0.9388,
+      "step": 7546
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001326577079333899,
+      "loss": 1.0299,
+      "step": 7547
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013264877896334657,
+      "loss": 1.111,
+      "step": 7548
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013263984999330327,
+      "loss": 1.0461,
+      "step": 7549
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013263092102325997,
+      "loss": 0.9326,
+      "step": 7550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013262199205321665,
+      "loss": 1.2637,
+      "step": 7551
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013261306308317335,
+      "loss": 1.0775,
+      "step": 7552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013260413411313006,
+      "loss": 1.0588,
+      "step": 7553
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013259520514308674,
+      "loss": 0.9331,
+      "step": 7554
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013258627617304344,
+      "loss": 1.0749,
+      "step": 7555
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013257734720300014,
+      "loss": 1.0235,
+      "step": 7556
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013256841823295682,
+      "loss": 0.9517,
+      "step": 7557
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013255948926291352,
+      "loss": 0.8707,
+      "step": 7558
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013255056029287023,
+      "loss": 1.2476,
+      "step": 7559
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001325416313228269,
+      "loss": 0.9441,
+      "step": 7560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001325327023527836,
+      "loss": 1.1178,
+      "step": 7561
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013252377338274031,
+      "loss": 1.0888,
+      "step": 7562
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000132514844412697,
+      "loss": 1.0197,
+      "step": 7563
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001325059154426537,
+      "loss": 1.1118,
+      "step": 7564
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001324969864726104,
+      "loss": 0.9326,
+      "step": 7565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013248805750256708,
+      "loss": 1.1022,
+      "step": 7566
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013247912853252378,
+      "loss": 1.177,
+      "step": 7567
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013247019956248048,
+      "loss": 1.0954,
+      "step": 7568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013246127059243716,
+      "loss": 1.1125,
+      "step": 7569
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013245234162239386,
+      "loss": 1.0581,
+      "step": 7570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013244341265235057,
+      "loss": 1.0287,
+      "step": 7571
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013243448368230725,
+      "loss": 0.9409,
+      "step": 7572
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013242555471226395,
+      "loss": 1.0609,
+      "step": 7573
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013241662574222065,
+      "loss": 0.9692,
+      "step": 7574
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013240769677217733,
+      "loss": 1.072,
+      "step": 7575
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013239876780213403,
+      "loss": 0.9372,
+      "step": 7576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013238983883209074,
+      "loss": 0.9505,
+      "step": 7577
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013238090986204742,
+      "loss": 1.0227,
+      "step": 7578
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013237198089200412,
+      "loss": 1.0544,
+      "step": 7579
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00013236305192196082,
+      "loss": 1.0775,
+      "step": 7580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001323541229519175,
+      "loss": 1.0665,
+      "step": 7581
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001323451939818742,
+      "loss": 0.9641,
+      "step": 7582
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001323362650118309,
+      "loss": 1.0567,
+      "step": 7583
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013232733604178759,
+      "loss": 0.9742,
+      "step": 7584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001323184070717443,
+      "loss": 0.8993,
+      "step": 7585
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013230947810170097,
+      "loss": 1.043,
+      "step": 7586
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013230054913165767,
+      "loss": 1.1614,
+      "step": 7587
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013229162016161435,
+      "loss": 0.9328,
+      "step": 7588
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013228269119157105,
+      "loss": 1.1428,
+      "step": 7589
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013227376222152776,
+      "loss": 1.0917,
+      "step": 7590
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013226483325148443,
+      "loss": 1.1106,
+      "step": 7591
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013225590428144114,
+      "loss": 1.0067,
+      "step": 7592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013224697531139784,
+      "loss": 1.0636,
+      "step": 7593
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013223804634135452,
+      "loss": 1.0715,
+      "step": 7594
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013222911737131122,
+      "loss": 1.0392,
+      "step": 7595
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013222018840126793,
+      "loss": 1.0161,
+      "step": 7596
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001322112594312246,
+      "loss": 1.0351,
+      "step": 7597
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001322023304611813,
+      "loss": 1.1028,
+      "step": 7598
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000132193401491138,
+      "loss": 0.9716,
+      "step": 7599
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001321844725210947,
+      "loss": 1.1848,
+      "step": 7600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001321755435510514,
+      "loss": 0.8903,
+      "step": 7601
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001321666145810081,
+      "loss": 1.1306,
+      "step": 7602
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013215768561096477,
+      "loss": 0.9781,
+      "step": 7603
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013214875664092148,
+      "loss": 1.0041,
+      "step": 7604
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013213982767087818,
+      "loss": 1.0658,
+      "step": 7605
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013213089870083486,
+      "loss": 1.0683,
+      "step": 7606
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013212196973079156,
+      "loss": 0.9804,
+      "step": 7607
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013211304076074827,
+      "loss": 1.1504,
+      "step": 7608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013210411179070494,
+      "loss": 0.9939,
+      "step": 7609
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013209518282066165,
+      "loss": 1.0505,
+      "step": 7610
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013208625385061835,
+      "loss": 1.1404,
+      "step": 7611
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013207732488057503,
+      "loss": 1.2547,
+      "step": 7612
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013206839591053173,
+      "loss": 1.0163,
+      "step": 7613
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013205946694048844,
+      "loss": 1.0151,
+      "step": 7614
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001320505379704451,
+      "loss": 0.9495,
+      "step": 7615
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013204160900040182,
+      "loss": 0.9852,
+      "step": 7616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013203268003035852,
+      "loss": 0.9618,
+      "step": 7617
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001320237510603152,
+      "loss": 1.0441,
+      "step": 7618
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001320148220902719,
+      "loss": 0.9746,
+      "step": 7619
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001320058931202286,
+      "loss": 1.2609,
+      "step": 7620
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013199696415018528,
+      "loss": 0.9886,
+      "step": 7621
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000131988035180142,
+      "loss": 0.959,
+      "step": 7622
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001319791062100987,
+      "loss": 0.9499,
+      "step": 7623
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013197017724005537,
+      "loss": 1.112,
+      "step": 7624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013196124827001207,
+      "loss": 1.0993,
+      "step": 7625
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013195231929996878,
+      "loss": 1.066,
+      "step": 7626
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013194339032992545,
+      "loss": 1.0048,
+      "step": 7627
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013193446135988216,
+      "loss": 1.0986,
+      "step": 7628
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013192553238983883,
+      "loss": 1.0168,
+      "step": 7629
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013191660341979554,
+      "loss": 1.0857,
+      "step": 7630
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013190767444975221,
+      "loss": 0.987,
+      "step": 7631
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013189874547970892,
+      "loss": 1.1023,
+      "step": 7632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013188981650966562,
+      "loss": 1.0423,
+      "step": 7633
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001318808875396223,
+      "loss": 1.218,
+      "step": 7634
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000131871958569579,
+      "loss": 0.966,
+      "step": 7635
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013186302959953568,
+      "loss": 1.1205,
+      "step": 7636
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013185410062949238,
+      "loss": 0.8831,
+      "step": 7637
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001318451716594491,
+      "loss": 1.0852,
+      "step": 7638
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013183624268940577,
+      "loss": 1.0332,
+      "step": 7639
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013182731371936247,
+      "loss": 0.9919,
+      "step": 7640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013181838474931917,
+      "loss": 0.985,
+      "step": 7641
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013180945577927585,
+      "loss": 1.1471,
+      "step": 7642
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013180052680923255,
+      "loss": 1.0336,
+      "step": 7643
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013179159783918926,
+      "loss": 0.9673,
+      "step": 7644
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013178266886914594,
+      "loss": 1.024,
+      "step": 7645
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013177373989910264,
+      "loss": 0.9844,
+      "step": 7646
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013176481092905934,
+      "loss": 1.0006,
+      "step": 7647
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013175588195901602,
+      "loss": 0.977,
+      "step": 7648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013174695298897272,
+      "loss": 1.0266,
+      "step": 7649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00013173802401892943,
+      "loss": 1.043,
+      "step": 7650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001317290950488861,
+      "loss": 1.0884,
+      "step": 7651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001317201660788428,
+      "loss": 1.1205,
+      "step": 7652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001317112371087995,
+      "loss": 1.0031,
+      "step": 7653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001317023081387562,
+      "loss": 1.1214,
+      "step": 7654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001316933791687129,
+      "loss": 1.084,
+      "step": 7655
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001316844501986696,
+      "loss": 1.0439,
+      "step": 7656
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013167552122862628,
+      "loss": 1.0627,
+      "step": 7657
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013166659225858298,
+      "loss": 1.0074,
+      "step": 7658
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013165766328853968,
+      "loss": 1.0676,
+      "step": 7659
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013164873431849636,
+      "loss": 1.0354,
+      "step": 7660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013163980534845306,
+      "loss": 1.1704,
+      "step": 7661
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013163087637840977,
+      "loss": 1.0244,
+      "step": 7662
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013162194740836645,
+      "loss": 1.0604,
+      "step": 7663
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013161301843832315,
+      "loss": 1.0555,
+      "step": 7664
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013160408946827985,
+      "loss": 1.1301,
+      "step": 7665
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013159516049823653,
+      "loss": 1.0136,
+      "step": 7666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013158623152819323,
+      "loss": 1.0548,
+      "step": 7667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013157730255814994,
+      "loss": 0.9964,
+      "step": 7668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013156837358810662,
+      "loss": 1.0538,
+      "step": 7669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013155944461806332,
+      "loss": 0.9224,
+      "step": 7670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013155051564802002,
+      "loss": 1.0212,
+      "step": 7671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001315415866779767,
+      "loss": 1.0,
+      "step": 7672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001315326577079334,
+      "loss": 0.9254,
+      "step": 7673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001315237287378901,
+      "loss": 1.0107,
+      "step": 7674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013151479976784679,
+      "loss": 0.9993,
+      "step": 7675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001315058707978035,
+      "loss": 1.0857,
+      "step": 7676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013149694182776017,
+      "loss": 0.9245,
+      "step": 7677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013148801285771687,
+      "loss": 1.271,
+      "step": 7678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013147908388767355,
+      "loss": 1.1026,
+      "step": 7679
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013147015491763025,
+      "loss": 0.979,
+      "step": 7680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013146122594758693,
+      "loss": 1.0435,
+      "step": 7681
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013145229697754363,
+      "loss": 0.9801,
+      "step": 7682
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013144336800750034,
+      "loss": 1.0132,
+      "step": 7683
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013143443903745704,
+      "loss": 1.0251,
+      "step": 7684
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013142551006741372,
+      "loss": 1.0053,
+      "step": 7685
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013141658109737042,
+      "loss": 0.9486,
+      "step": 7686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013140765212732713,
+      "loss": 0.9284,
+      "step": 7687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001313987231572838,
+      "loss": 1.0959,
+      "step": 7688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001313897941872405,
+      "loss": 1.0939,
+      "step": 7689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001313808652171972,
+      "loss": 1.3107,
+      "step": 7690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001313719362471539,
+      "loss": 1.1875,
+      "step": 7691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001313630072771106,
+      "loss": 1.0648,
+      "step": 7692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001313540783070673,
+      "loss": 0.9265,
+      "step": 7693
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013134514933702397,
+      "loss": 1.1292,
+      "step": 7694
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013133622036698068,
+      "loss": 1.0193,
+      "step": 7695
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013132729139693738,
+      "loss": 1.0128,
+      "step": 7696
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013131836242689406,
+      "loss": 1.0633,
+      "step": 7697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013130943345685076,
+      "loss": 1.0162,
+      "step": 7698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013130050448680747,
+      "loss": 1.098,
+      "step": 7699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013129157551676414,
+      "loss": 0.8869,
+      "step": 7700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013128264654672085,
+      "loss": 1.0313,
+      "step": 7701
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013127371757667755,
+      "loss": 1.0459,
+      "step": 7702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013126478860663423,
+      "loss": 0.9373,
+      "step": 7703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013125585963659093,
+      "loss": 1.106,
+      "step": 7704
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013124693066654764,
+      "loss": 1.0753,
+      "step": 7705
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001312380016965043,
+      "loss": 1.0956,
+      "step": 7706
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013122907272646102,
+      "loss": 1.1591,
+      "step": 7707
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013122014375641772,
+      "loss": 1.0222,
+      "step": 7708
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001312112147863744,
+      "loss": 1.0964,
+      "step": 7709
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001312022858163311,
+      "loss": 1.0969,
+      "step": 7710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001311933568462878,
+      "loss": 1.035,
+      "step": 7711
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013118442787624448,
+      "loss": 1.1679,
+      "step": 7712
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013117549890620119,
+      "loss": 1.0225,
+      "step": 7713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001311665699361579,
+      "loss": 1.0616,
+      "step": 7714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013115764096611457,
+      "loss": 1.1649,
+      "step": 7715
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013114871199607127,
+      "loss": 1.142,
+      "step": 7716
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013113978302602798,
+      "loss": 1.0212,
+      "step": 7717
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013113085405598465,
+      "loss": 0.9528,
+      "step": 7718
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013112192508594136,
+      "loss": 1.0888,
+      "step": 7719
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013111299611589803,
+      "loss": 0.9531,
+      "step": 7720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013110406714585474,
+      "loss": 0.9862,
+      "step": 7721
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013109513817581141,
+      "loss": 1.089,
+      "step": 7722
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013108620920576812,
+      "loss": 1.0506,
+      "step": 7723
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013107728023572482,
+      "loss": 0.9328,
+      "step": 7724
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001310683512656815,
+      "loss": 1.0428,
+      "step": 7725
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001310594222956382,
+      "loss": 1.0805,
+      "step": 7726
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013105049332559488,
+      "loss": 1.0223,
+      "step": 7727
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00013104156435555158,
+      "loss": 1.0254,
+      "step": 7728
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001310326353855083,
+      "loss": 0.9289,
+      "step": 7729
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013102370641546496,
+      "loss": 1.0728,
+      "step": 7730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013101477744542167,
+      "loss": 1.0513,
+      "step": 7731
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013100584847537837,
+      "loss": 0.9788,
+      "step": 7732
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013099691950533505,
+      "loss": 1.059,
+      "step": 7733
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013098799053529175,
+      "loss": 0.9232,
+      "step": 7734
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013097906156524846,
+      "loss": 1.0969,
+      "step": 7735
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013097013259520513,
+      "loss": 1.0868,
+      "step": 7736
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013096120362516184,
+      "loss": 0.9829,
+      "step": 7737
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013095227465511854,
+      "loss": 1.0565,
+      "step": 7738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013094334568507522,
+      "loss": 1.1164,
+      "step": 7739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013093441671503192,
+      "loss": 1.0691,
+      "step": 7740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013092548774498863,
+      "loss": 0.9744,
+      "step": 7741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001309165587749453,
+      "loss": 0.9682,
+      "step": 7742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000130907629804902,
+      "loss": 1.0368,
+      "step": 7743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001308987008348587,
+      "loss": 0.958,
+      "step": 7744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001308897718648154,
+      "loss": 0.9593,
+      "step": 7745
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001308808428947721,
+      "loss": 1.0672,
+      "step": 7746
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001308719139247288,
+      "loss": 1.0267,
+      "step": 7747
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013086298495468547,
+      "loss": 0.9907,
+      "step": 7748
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013085405598464218,
+      "loss": 1.003,
+      "step": 7749
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013084512701459888,
+      "loss": 1.0093,
+      "step": 7750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013083619804455556,
+      "loss": 0.9711,
+      "step": 7751
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013082726907451226,
+      "loss": 0.9685,
+      "step": 7752
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013081834010446897,
+      "loss": 1.0264,
+      "step": 7753
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013080941113442564,
+      "loss": 1.1281,
+      "step": 7754
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013080048216438235,
+      "loss": 0.9355,
+      "step": 7755
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013079155319433905,
+      "loss": 0.9643,
+      "step": 7756
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013078262422429573,
+      "loss": 1.0731,
+      "step": 7757
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013077369525425243,
+      "loss": 1.0152,
+      "step": 7758
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013076476628420914,
+      "loss": 1.1781,
+      "step": 7759
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013075583731416581,
+      "loss": 1.0553,
+      "step": 7760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013074690834412252,
+      "loss": 1.0267,
+      "step": 7761
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013073797937407922,
+      "loss": 0.908,
+      "step": 7762
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001307290504040359,
+      "loss": 1.0449,
+      "step": 7763
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001307201214339926,
+      "loss": 1.0229,
+      "step": 7764
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001307111924639493,
+      "loss": 1.0863,
+      "step": 7765
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013070226349390598,
+      "loss": 1.1224,
+      "step": 7766
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001306933345238627,
+      "loss": 1.1504,
+      "step": 7767
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013068440555381937,
+      "loss": 1.0608,
+      "step": 7768
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013067547658377607,
+      "loss": 0.9696,
+      "step": 7769
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013066654761373275,
+      "loss": 1.0865,
+      "step": 7770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013065761864368945,
+      "loss": 0.9916,
+      "step": 7771
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013064868967364613,
+      "loss": 1.0311,
+      "step": 7772
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013063976070360283,
+      "loss": 1.028,
+      "step": 7773
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013063083173355954,
+      "loss": 0.9459,
+      "step": 7774
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001306219027635162,
+      "loss": 1.0162,
+      "step": 7775
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013061297379347292,
+      "loss": 1.0553,
+      "step": 7776
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013060404482342962,
+      "loss": 1.0915,
+      "step": 7777
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001305951158533863,
+      "loss": 1.0316,
+      "step": 7778
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000130586186883343,
+      "loss": 1.0575,
+      "step": 7779
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001305772579132997,
+      "loss": 0.8695,
+      "step": 7780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013056832894325638,
+      "loss": 0.979,
+      "step": 7781
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001305593999732131,
+      "loss": 1.1326,
+      "step": 7782
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001305504710031698,
+      "loss": 1.003,
+      "step": 7783
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001305415420331265,
+      "loss": 1.0019,
+      "step": 7784
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013053261306308317,
+      "loss": 0.9538,
+      "step": 7785
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013052368409303988,
+      "loss": 1.1213,
+      "step": 7786
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013051475512299658,
+      "loss": 1.1123,
+      "step": 7787
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013050582615295326,
+      "loss": 0.922,
+      "step": 7788
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013049689718290996,
+      "loss": 1.0877,
+      "step": 7789
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013048796821286666,
+      "loss": 1.0164,
+      "step": 7790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013047903924282334,
+      "loss": 1.1401,
+      "step": 7791
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013047011027278005,
+      "loss": 0.9461,
+      "step": 7792
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013046118130273675,
+      "loss": 1.079,
+      "step": 7793
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013045225233269343,
+      "loss": 1.0079,
+      "step": 7794
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013044332336265013,
+      "loss": 1.0771,
+      "step": 7795
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013043439439260683,
+      "loss": 1.0397,
+      "step": 7796
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001304254654225635,
+      "loss": 1.1438,
+      "step": 7797
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013041653645252022,
+      "loss": 1.1383,
+      "step": 7798
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013040760748247692,
+      "loss": 1.0683,
+      "step": 7799
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001303986785124336,
+      "loss": 0.9641,
+      "step": 7800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001303897495423903,
+      "loss": 1.1481,
+      "step": 7801
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000130380820572347,
+      "loss": 1.0931,
+      "step": 7802
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013037189160230368,
+      "loss": 1.178,
+      "step": 7803
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00013036296263226039,
+      "loss": 1.0203,
+      "step": 7804
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001303540336622171,
+      "loss": 1.0495,
+      "step": 7805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013034510469217377,
+      "loss": 1.2022,
+      "step": 7806
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013033617572213047,
+      "loss": 0.949,
+      "step": 7807
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013032724675208717,
+      "loss": 0.9459,
+      "step": 7808
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013031831778204385,
+      "loss": 1.0549,
+      "step": 7809
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013030938881200056,
+      "loss": 1.006,
+      "step": 7810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013030045984195723,
+      "loss": 1.0209,
+      "step": 7811
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013029153087191394,
+      "loss": 0.9522,
+      "step": 7812
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013028260190187061,
+      "loss": 0.9061,
+      "step": 7813
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013027367293182732,
+      "loss": 0.9848,
+      "step": 7814
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013026474396178402,
+      "loss": 1.1039,
+      "step": 7815
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001302558149917407,
+      "loss": 1.0041,
+      "step": 7816
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001302468860216974,
+      "loss": 1.0455,
+      "step": 7817
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013023795705165408,
+      "loss": 1.0402,
+      "step": 7818
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013022902808161078,
+      "loss": 0.8785,
+      "step": 7819
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001302200991115675,
+      "loss": 0.9779,
+      "step": 7820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013021117014152416,
+      "loss": 1.0558,
+      "step": 7821
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013020224117148087,
+      "loss": 1.0608,
+      "step": 7822
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013019331220143757,
+      "loss": 1.0101,
+      "step": 7823
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013018438323139425,
+      "loss": 0.9884,
+      "step": 7824
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013017545426135095,
+      "loss": 1.102,
+      "step": 7825
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013016652529130766,
+      "loss": 0.9897,
+      "step": 7826
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013015759632126433,
+      "loss": 1.0719,
+      "step": 7827
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013014866735122104,
+      "loss": 1.0576,
+      "step": 7828
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013013973838117774,
+      "loss": 1.0298,
+      "step": 7829
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013013080941113442,
+      "loss": 0.9753,
+      "step": 7830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013012188044109112,
+      "loss": 1.1402,
+      "step": 7831
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013011295147104783,
+      "loss": 1.0049,
+      "step": 7832
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001301040225010045,
+      "loss": 0.9413,
+      "step": 7833
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001300950935309612,
+      "loss": 1.0316,
+      "step": 7834
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001300861645609179,
+      "loss": 1.0053,
+      "step": 7835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001300772355908746,
+      "loss": 1.039,
+      "step": 7836
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001300683066208313,
+      "loss": 1.0385,
+      "step": 7837
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000130059377650788,
+      "loss": 1.0632,
+      "step": 7838
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013005044868074467,
+      "loss": 1.124,
+      "step": 7839
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013004151971070138,
+      "loss": 1.2438,
+      "step": 7840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013003259074065808,
+      "loss": 0.9919,
+      "step": 7841
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013002366177061476,
+      "loss": 0.9895,
+      "step": 7842
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013001473280057146,
+      "loss": 1.0263,
+      "step": 7843
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00013000580383052817,
+      "loss": 0.955,
+      "step": 7844
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012999687486048484,
+      "loss": 1.0176,
+      "step": 7845
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012998794589044155,
+      "loss": 1.1296,
+      "step": 7846
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012997901692039825,
+      "loss": 1.0629,
+      "step": 7847
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012997008795035493,
+      "loss": 1.1686,
+      "step": 7848
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012996115898031163,
+      "loss": 1.1597,
+      "step": 7849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012995223001026834,
+      "loss": 1.0197,
+      "step": 7850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012994330104022501,
+      "loss": 1.0809,
+      "step": 7851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012993437207018172,
+      "loss": 1.0534,
+      "step": 7852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012992544310013842,
+      "loss": 1.0425,
+      "step": 7853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001299165141300951,
+      "loss": 1.0561,
+      "step": 7854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001299075851600518,
+      "loss": 0.9177,
+      "step": 7855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001298986561900085,
+      "loss": 1.1675,
+      "step": 7856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012988972721996518,
+      "loss": 1.0933,
+      "step": 7857
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001298807982499219,
+      "loss": 0.9986,
+      "step": 7858
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012987186927987857,
+      "loss": 1.085,
+      "step": 7859
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012986294030983527,
+      "loss": 1.1405,
+      "step": 7860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012985401133979195,
+      "loss": 1.0792,
+      "step": 7861
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012984508236974865,
+      "loss": 0.9688,
+      "step": 7862
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012983615339970533,
+      "loss": 0.9908,
+      "step": 7863
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012982722442966203,
+      "loss": 1.159,
+      "step": 7864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012981829545961874,
+      "loss": 1.1632,
+      "step": 7865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001298093664895754,
+      "loss": 1.0175,
+      "step": 7866
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012980043751953212,
+      "loss": 1.0895,
+      "step": 7867
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012979150854948882,
+      "loss": 0.9282,
+      "step": 7868
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001297825795794455,
+      "loss": 1.2705,
+      "step": 7869
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001297736506094022,
+      "loss": 1.0136,
+      "step": 7870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001297647216393589,
+      "loss": 0.9922,
+      "step": 7871
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012975579266931558,
+      "loss": 1.0527,
+      "step": 7872
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012974686369927229,
+      "loss": 1.0213,
+      "step": 7873
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000129737934729229,
+      "loss": 1.0201,
+      "step": 7874
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012972900575918567,
+      "loss": 1.1974,
+      "step": 7875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012972007678914237,
+      "loss": 1.0632,
+      "step": 7876
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012971114781909908,
+      "loss": 1.0297,
+      "step": 7877
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012970221884905575,
+      "loss": 1.0048,
+      "step": 7878
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00012969328987901246,
+      "loss": 1.1137,
+      "step": 7879
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012968436090896916,
+      "loss": 1.0245,
+      "step": 7880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012967543193892586,
+      "loss": 1.2561,
+      "step": 7881
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012966650296888254,
+      "loss": 0.8512,
+      "step": 7882
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012965757399883925,
+      "loss": 1.0339,
+      "step": 7883
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012964864502879595,
+      "loss": 1.0242,
+      "step": 7884
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012963971605875263,
+      "loss": 0.9307,
+      "step": 7885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012963078708870933,
+      "loss": 1.003,
+      "step": 7886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012962185811866603,
+      "loss": 1.1287,
+      "step": 7887
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001296129291486227,
+      "loss": 0.9908,
+      "step": 7888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012960400017857942,
+      "loss": 1.0592,
+      "step": 7889
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012959507120853612,
+      "loss": 1.0681,
+      "step": 7890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001295861422384928,
+      "loss": 0.9619,
+      "step": 7891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001295772132684495,
+      "loss": 1.0653,
+      "step": 7892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001295682842984062,
+      "loss": 1.0031,
+      "step": 7893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012955935532836288,
+      "loss": 1.02,
+      "step": 7894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012955042635831959,
+      "loss": 0.973,
+      "step": 7895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001295414973882763,
+      "loss": 1.1933,
+      "step": 7896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012953256841823297,
+      "loss": 0.9958,
+      "step": 7897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012952363944818967,
+      "loss": 1.0192,
+      "step": 7898
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012951471047814637,
+      "loss": 1.0785,
+      "step": 7899
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012950578150810305,
+      "loss": 1.0083,
+      "step": 7900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012949685253805976,
+      "loss": 1.073,
+      "step": 7901
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012948792356801643,
+      "loss": 0.9397,
+      "step": 7902
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012947899459797314,
+      "loss": 1.0945,
+      "step": 7903
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001294700656279298,
+      "loss": 1.0207,
+      "step": 7904
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012946113665788652,
+      "loss": 1.1687,
+      "step": 7905
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001294522076878432,
+      "loss": 1.0067,
+      "step": 7906
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001294432787177999,
+      "loss": 1.0714,
+      "step": 7907
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001294343497477566,
+      "loss": 0.8471,
+      "step": 7908
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012942542077771328,
+      "loss": 1.0592,
+      "step": 7909
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012941649180766998,
+      "loss": 1.0837,
+      "step": 7910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001294075628376267,
+      "loss": 1.0842,
+      "step": 7911
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012939863386758336,
+      "loss": 0.9964,
+      "step": 7912
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012938970489754007,
+      "loss": 0.9619,
+      "step": 7913
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012938077592749677,
+      "loss": 1.1151,
+      "step": 7914
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012937184695745345,
+      "loss": 1.0253,
+      "step": 7915
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012936291798741015,
+      "loss": 0.9405,
+      "step": 7916
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012935398901736686,
+      "loss": 1.0334,
+      "step": 7917
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012934506004732353,
+      "loss": 1.0778,
+      "step": 7918
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012933613107728024,
+      "loss": 0.9158,
+      "step": 7919
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012932720210723694,
+      "loss": 0.9436,
+      "step": 7920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012931827313719362,
+      "loss": 1.0517,
+      "step": 7921
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012930934416715032,
+      "loss": 0.9497,
+      "step": 7922
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012930041519710703,
+      "loss": 1.0072,
+      "step": 7923
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001292914862270637,
+      "loss": 1.015,
+      "step": 7924
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001292825572570204,
+      "loss": 0.9071,
+      "step": 7925
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001292736282869771,
+      "loss": 1.0196,
+      "step": 7926
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001292646993169338,
+      "loss": 1.048,
+      "step": 7927
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001292557703468905,
+      "loss": 0.9973,
+      "step": 7928
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001292468413768472,
+      "loss": 1.1353,
+      "step": 7929
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012923791240680387,
+      "loss": 1.1938,
+      "step": 7930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012922898343676058,
+      "loss": 1.2051,
+      "step": 7931
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012922005446671728,
+      "loss": 1.0423,
+      "step": 7932
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012921112549667396,
+      "loss": 1.1044,
+      "step": 7933
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012920219652663066,
+      "loss": 1.1509,
+      "step": 7934
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012919326755658737,
+      "loss": 1.323,
+      "step": 7935
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012918433858654404,
+      "loss": 1.0248,
+      "step": 7936
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012917540961650075,
+      "loss": 0.937,
+      "step": 7937
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012916648064645745,
+      "loss": 1.0278,
+      "step": 7938
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012915755167641413,
+      "loss": 1.0134,
+      "step": 7939
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012914862270637083,
+      "loss": 1.1422,
+      "step": 7940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012913969373632754,
+      "loss": 1.068,
+      "step": 7941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012913076476628421,
+      "loss": 1.0677,
+      "step": 7942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012912183579624092,
+      "loss": 1.0835,
+      "step": 7943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012911290682619762,
+      "loss": 1.1816,
+      "step": 7944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001291039778561543,
+      "loss": 1.0203,
+      "step": 7945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000129095048886111,
+      "loss": 1.0286,
+      "step": 7946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001290861199160677,
+      "loss": 1.1432,
+      "step": 7947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012907719094602438,
+      "loss": 1.0942,
+      "step": 7948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001290682619759811,
+      "loss": 0.9497,
+      "step": 7949
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012905933300593777,
+      "loss": 1.1083,
+      "step": 7950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012905040403589447,
+      "loss": 1.0808,
+      "step": 7951
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012904147506585115,
+      "loss": 0.8984,
+      "step": 7952
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00012903254609580785,
+      "loss": 1.0891,
+      "step": 7953
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012902361712576453,
+      "loss": 1.0256,
+      "step": 7954
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012901468815572123,
+      "loss": 1.0262,
+      "step": 7955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012900575918567794,
+      "loss": 0.9393,
+      "step": 7956
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001289968302156346,
+      "loss": 1.0187,
+      "step": 7957
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012898790124559132,
+      "loss": 1.1085,
+      "step": 7958
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012897897227554802,
+      "loss": 1.0835,
+      "step": 7959
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001289700433055047,
+      "loss": 1.0821,
+      "step": 7960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001289611143354614,
+      "loss": 1.0548,
+      "step": 7961
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001289521853654181,
+      "loss": 1.0153,
+      "step": 7962
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012894325639537478,
+      "loss": 0.9874,
+      "step": 7963
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012893432742533149,
+      "loss": 0.9372,
+      "step": 7964
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001289253984552882,
+      "loss": 1.1019,
+      "step": 7965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012891646948524487,
+      "loss": 0.8962,
+      "step": 7966
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012890754051520157,
+      "loss": 1.0445,
+      "step": 7967
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012889861154515828,
+      "loss": 1.0367,
+      "step": 7968
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012888968257511495,
+      "loss": 0.9657,
+      "step": 7969
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012888075360507166,
+      "loss": 1.1596,
+      "step": 7970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012887182463502836,
+      "loss": 1.0915,
+      "step": 7971
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012886289566498504,
+      "loss": 1.0454,
+      "step": 7972
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012885396669494174,
+      "loss": 1.129,
+      "step": 7973
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012884503772489845,
+      "loss": 1.045,
+      "step": 7974
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012883610875485512,
+      "loss": 1.0646,
+      "step": 7975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012882717978481183,
+      "loss": 1.4156,
+      "step": 7976
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012881825081476853,
+      "loss": 1.012,
+      "step": 7977
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012880932184472523,
+      "loss": 1.0136,
+      "step": 7978
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001288003928746819,
+      "loss": 1.0115,
+      "step": 7979
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012879146390463862,
+      "loss": 0.9636,
+      "step": 7980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012878253493459532,
+      "loss": 1.0788,
+      "step": 7981
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000128773605964552,
+      "loss": 0.9544,
+      "step": 7982
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001287646769945087,
+      "loss": 1.0823,
+      "step": 7983
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001287557480244654,
+      "loss": 0.9022,
+      "step": 7984
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012874681905442208,
+      "loss": 1.1179,
+      "step": 7985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012873789008437878,
+      "loss": 1.0465,
+      "step": 7986
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001287289611143355,
+      "loss": 1.0123,
+      "step": 7987
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012872003214429217,
+      "loss": 1.0251,
+      "step": 7988
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012871110317424887,
+      "loss": 1.0138,
+      "step": 7989
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012870217420420557,
+      "loss": 1.1272,
+      "step": 7990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012869324523416225,
+      "loss": 1.0591,
+      "step": 7991
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012868431626411895,
+      "loss": 1.012,
+      "step": 7992
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012867538729407563,
+      "loss": 0.9995,
+      "step": 7993
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012866645832403234,
+      "loss": 1.009,
+      "step": 7994
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000128657529353989,
+      "loss": 1.0197,
+      "step": 7995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012864860038394572,
+      "loss": 1.0345,
+      "step": 7996
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001286396714139024,
+      "loss": 1.0405,
+      "step": 7997
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001286307424438591,
+      "loss": 0.9779,
+      "step": 7998
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001286218134738158,
+      "loss": 1.0639,
+      "step": 7999
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012861288450377248,
+      "loss": 0.9926,
+      "step": 8000
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 1.0322954654693604,
+      "eval_runtime": 3049.234,
+      "eval_samples_per_second": 6.167,
+      "eval_steps_per_second": 0.771,
+      "step": 8000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012860395553372918,
+      "loss": 0.954,
+      "step": 8001
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001285950265636859,
+      "loss": 1.2393,
+      "step": 8002
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012858609759364256,
+      "loss": 1.0434,
+      "step": 8003
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012857716862359927,
+      "loss": 1.1327,
+      "step": 8004
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012856823965355597,
+      "loss": 1.0596,
+      "step": 8005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012855931068351265,
+      "loss": 1.0033,
+      "step": 8006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012855038171346935,
+      "loss": 0.987,
+      "step": 8007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012854145274342606,
+      "loss": 1.0492,
+      "step": 8008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012853252377338273,
+      "loss": 1.0672,
+      "step": 8009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012852359480333944,
+      "loss": 1.0761,
+      "step": 8010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012851466583329614,
+      "loss": 1.0836,
+      "step": 8011
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012850573686325282,
+      "loss": 1.0387,
+      "step": 8012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012849680789320952,
+      "loss": 1.1085,
+      "step": 8013
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012848787892316623,
+      "loss": 1.0199,
+      "step": 8014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001284789499531229,
+      "loss": 1.0066,
+      "step": 8015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001284700209830796,
+      "loss": 0.9859,
+      "step": 8016
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001284610920130363,
+      "loss": 1.0416,
+      "step": 8017
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000128452163042993,
+      "loss": 0.9841,
+      "step": 8018
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001284432340729497,
+      "loss": 1.0201,
+      "step": 8019
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001284343051029064,
+      "loss": 0.9832,
+      "step": 8020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012842537613286307,
+      "loss": 1.1072,
+      "step": 8021
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012841644716281978,
+      "loss": 0.9139,
+      "step": 8022
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012840751819277648,
+      "loss": 1.1441,
+      "step": 8023
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012839858922273316,
+      "loss": 0.9616,
+      "step": 8024
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012838966025268986,
+      "loss": 1.0458,
+      "step": 8025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012838073128264657,
+      "loss": 0.8632,
+      "step": 8026
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012837180231260324,
+      "loss": 1.0525,
+      "step": 8027
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00012836287334255995,
+      "loss": 1.012,
+      "step": 8028
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012835394437251665,
+      "loss": 1.0591,
+      "step": 8029
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012834501540247333,
+      "loss": 1.0696,
+      "step": 8030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012833608643243003,
+      "loss": 0.9502,
+      "step": 8031
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012832715746238674,
+      "loss": 1.1671,
+      "step": 8032
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012831822849234341,
+      "loss": 0.9981,
+      "step": 8033
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012830929952230012,
+      "loss": 0.9636,
+      "step": 8034
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012830037055225682,
+      "loss": 1.1239,
+      "step": 8035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001282914415822135,
+      "loss": 1.026,
+      "step": 8036
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001282825126121702,
+      "loss": 0.932,
+      "step": 8037
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001282735836421269,
+      "loss": 1.1473,
+      "step": 8038
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012826465467208358,
+      "loss": 0.9946,
+      "step": 8039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001282557257020403,
+      "loss": 1.0371,
+      "step": 8040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012824679673199696,
+      "loss": 0.9975,
+      "step": 8041
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012823786776195367,
+      "loss": 1.152,
+      "step": 8042
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012822893879191035,
+      "loss": 1.0333,
+      "step": 8043
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012822000982186705,
+      "loss": 0.9607,
+      "step": 8044
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012821108085182373,
+      "loss": 0.9083,
+      "step": 8045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012820215188178043,
+      "loss": 1.0704,
+      "step": 8046
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012819322291173713,
+      "loss": 1.0823,
+      "step": 8047
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001281842939416938,
+      "loss": 0.9885,
+      "step": 8048
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012817536497165052,
+      "loss": 0.9887,
+      "step": 8049
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012816643600160722,
+      "loss": 1.0999,
+      "step": 8050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001281575070315639,
+      "loss": 1.1071,
+      "step": 8051
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001281485780615206,
+      "loss": 1.0257,
+      "step": 8052
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001281396490914773,
+      "loss": 0.9586,
+      "step": 8053
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012813072012143398,
+      "loss": 1.0034,
+      "step": 8054
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012812179115139069,
+      "loss": 0.9544,
+      "step": 8055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001281128621813474,
+      "loss": 0.9507,
+      "step": 8056
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012810393321130407,
+      "loss": 1.0004,
+      "step": 8057
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012809500424126077,
+      "loss": 1.0313,
+      "step": 8058
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012808607527121747,
+      "loss": 1.0055,
+      "step": 8059
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012807714630117415,
+      "loss": 1.0913,
+      "step": 8060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012806821733113086,
+      "loss": 1.1048,
+      "step": 8061
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012805928836108756,
+      "loss": 1.0315,
+      "step": 8062
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012805035939104424,
+      "loss": 0.9804,
+      "step": 8063
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012804143042100094,
+      "loss": 1.0511,
+      "step": 8064
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012803250145095764,
+      "loss": 1.2121,
+      "step": 8065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012802357248091432,
+      "loss": 1.2029,
+      "step": 8066
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012801464351087103,
+      "loss": 0.9622,
+      "step": 8067
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012800571454082773,
+      "loss": 1.0416,
+      "step": 8068
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001279967855707844,
+      "loss": 0.9557,
+      "step": 8069
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001279878566007411,
+      "loss": 1.0982,
+      "step": 8070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012797892763069781,
+      "loss": 1.0087,
+      "step": 8071
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001279699986606545,
+      "loss": 1.1279,
+      "step": 8072
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001279610696906112,
+      "loss": 1.0673,
+      "step": 8073
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001279521407205679,
+      "loss": 1.0505,
+      "step": 8074
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001279432117505246,
+      "loss": 0.9347,
+      "step": 8075
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012793428278048128,
+      "loss": 1.059,
+      "step": 8076
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012792535381043798,
+      "loss": 0.9423,
+      "step": 8077
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001279164248403947,
+      "loss": 0.9875,
+      "step": 8078
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012790749587035137,
+      "loss": 1.0845,
+      "step": 8079
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012789856690030807,
+      "loss": 1.042,
+      "step": 8080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012788963793026477,
+      "loss": 0.925,
+      "step": 8081
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012788070896022145,
+      "loss": 1.1348,
+      "step": 8082
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012787177999017815,
+      "loss": 0.9927,
+      "step": 8083
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012786285102013483,
+      "loss": 1.0191,
+      "step": 8084
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012785392205009154,
+      "loss": 0.9884,
+      "step": 8085
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001278449930800482,
+      "loss": 1.0545,
+      "step": 8086
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012783606411000492,
+      "loss": 1.1962,
+      "step": 8087
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001278271351399616,
+      "loss": 1.0366,
+      "step": 8088
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001278182061699183,
+      "loss": 1.0678,
+      "step": 8089
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000127809277199875,
+      "loss": 0.9911,
+      "step": 8090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012780034822983168,
+      "loss": 1.1023,
+      "step": 8091
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012779141925978838,
+      "loss": 1.0357,
+      "step": 8092
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001277824902897451,
+      "loss": 1.0248,
+      "step": 8093
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012777356131970176,
+      "loss": 0.9649,
+      "step": 8094
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012776463234965847,
+      "loss": 1.1302,
+      "step": 8095
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012775570337961517,
+      "loss": 1.0209,
+      "step": 8096
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012774677440957185,
+      "loss": 1.169,
+      "step": 8097
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012773784543952855,
+      "loss": 1.0649,
+      "step": 8098
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012772891646948526,
+      "loss": 1.0712,
+      "step": 8099
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012771998749944193,
+      "loss": 1.1551,
+      "step": 8100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012771105852939864,
+      "loss": 1.0299,
+      "step": 8101
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012770212955935534,
+      "loss": 1.0201,
+      "step": 8102
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00012769320058931202,
+      "loss": 0.9645,
+      "step": 8103
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012768427161926872,
+      "loss": 0.9741,
+      "step": 8104
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012767534264922543,
+      "loss": 1.1591,
+      "step": 8105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001276664136791821,
+      "loss": 1.118,
+      "step": 8106
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001276574847091388,
+      "loss": 1.0659,
+      "step": 8107
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001276485557390955,
+      "loss": 0.9271,
+      "step": 8108
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001276396267690522,
+      "loss": 0.9796,
+      "step": 8109
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001276306977990089,
+      "loss": 1.0578,
+      "step": 8110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001276217688289656,
+      "loss": 0.9494,
+      "step": 8111
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012761283985892227,
+      "loss": 0.988,
+      "step": 8112
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012760391088887898,
+      "loss": 1.071,
+      "step": 8113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012759498191883568,
+      "loss": 1.0799,
+      "step": 8114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012758605294879236,
+      "loss": 1.1414,
+      "step": 8115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012757712397874906,
+      "loss": 1.0098,
+      "step": 8116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012756819500870577,
+      "loss": 1.1861,
+      "step": 8117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012755926603866244,
+      "loss": 0.9394,
+      "step": 8118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012755033706861915,
+      "loss": 0.8767,
+      "step": 8119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012754140809857585,
+      "loss": 0.999,
+      "step": 8120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012753247912853253,
+      "loss": 1.0339,
+      "step": 8121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012752355015848923,
+      "loss": 0.9409,
+      "step": 8122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012751462118844594,
+      "loss": 0.9963,
+      "step": 8123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001275056922184026,
+      "loss": 1.0484,
+      "step": 8124
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012749676324835932,
+      "loss": 1.019,
+      "step": 8125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012748783427831602,
+      "loss": 1.2182,
+      "step": 8126
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001274789053082727,
+      "loss": 1.0122,
+      "step": 8127
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001274699763382294,
+      "loss": 1.0551,
+      "step": 8128
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012746104736818608,
+      "loss": 1.0542,
+      "step": 8129
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012745211839814278,
+      "loss": 1.0269,
+      "step": 8130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001274431894280995,
+      "loss": 1.1273,
+      "step": 8131
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012743426045805616,
+      "loss": 1.1656,
+      "step": 8132
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012742533148801287,
+      "loss": 0.9778,
+      "step": 8133
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012741640251796955,
+      "loss": 1.108,
+      "step": 8134
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012740747354792625,
+      "loss": 0.9504,
+      "step": 8135
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012739854457788293,
+      "loss": 1.0487,
+      "step": 8136
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012738961560783963,
+      "loss": 1.1201,
+      "step": 8137
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012738068663779633,
+      "loss": 1.0366,
+      "step": 8138
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000127371757667753,
+      "loss": 1.0623,
+      "step": 8139
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012736282869770972,
+      "loss": 0.9885,
+      "step": 8140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012735389972766642,
+      "loss": 1.1176,
+      "step": 8141
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001273449707576231,
+      "loss": 1.2351,
+      "step": 8142
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001273360417875798,
+      "loss": 1.0271,
+      "step": 8143
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001273271128175365,
+      "loss": 1.104,
+      "step": 8144
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012731818384749318,
+      "loss": 0.9708,
+      "step": 8145
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012730925487744989,
+      "loss": 0.9678,
+      "step": 8146
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001273003259074066,
+      "loss": 1.0456,
+      "step": 8147
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012729139693736327,
+      "loss": 0.9836,
+      "step": 8148
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012728246796731997,
+      "loss": 1.0813,
+      "step": 8149
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012727353899727667,
+      "loss": 1.1904,
+      "step": 8150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012726461002723335,
+      "loss": 0.973,
+      "step": 8151
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012725568105719006,
+      "loss": 1.0893,
+      "step": 8152
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012724675208714676,
+      "loss": 0.997,
+      "step": 8153
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012723782311710344,
+      "loss": 0.9786,
+      "step": 8154
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012722889414706014,
+      "loss": 1.0708,
+      "step": 8155
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012721996517701684,
+      "loss": 0.9889,
+      "step": 8156
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012721103620697352,
+      "loss": 1.121,
+      "step": 8157
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012720210723693023,
+      "loss": 0.9686,
+      "step": 8158
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012719317826688693,
+      "loss": 1.0488,
+      "step": 8159
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001271842492968436,
+      "loss": 1.0759,
+      "step": 8160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001271753203268003,
+      "loss": 1.0868,
+      "step": 8161
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012716639135675701,
+      "loss": 1.0923,
+      "step": 8162
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001271574623867137,
+      "loss": 1.2643,
+      "step": 8163
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001271485334166704,
+      "loss": 0.9315,
+      "step": 8164
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001271396044466271,
+      "loss": 1.0317,
+      "step": 8165
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012713067547658378,
+      "loss": 1.0828,
+      "step": 8166
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012712174650654048,
+      "loss": 1.0722,
+      "step": 8167
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012711281753649718,
+      "loss": 1.1091,
+      "step": 8168
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012710388856645386,
+      "loss": 1.0553,
+      "step": 8169
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012709495959641057,
+      "loss": 1.0777,
+      "step": 8170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012708603062636727,
+      "loss": 1.1412,
+      "step": 8171
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012707710165632395,
+      "loss": 1.0682,
+      "step": 8172
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012706817268628065,
+      "loss": 0.9917,
+      "step": 8173
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012705924371623735,
+      "loss": 1.0242,
+      "step": 8174
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012705031474619403,
+      "loss": 1.1835,
+      "step": 8175
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012704138577615074,
+      "loss": 0.9566,
+      "step": 8176
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001270324568061074,
+      "loss": 0.9213,
+      "step": 8177
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00012702352783606412,
+      "loss": 1.0928,
+      "step": 8178
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001270145988660208,
+      "loss": 1.1143,
+      "step": 8179
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001270056698959775,
+      "loss": 1.0995,
+      "step": 8180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001269967409259342,
+      "loss": 0.9497,
+      "step": 8181
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012698781195589088,
+      "loss": 1.0657,
+      "step": 8182
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012697888298584758,
+      "loss": 1.0198,
+      "step": 8183
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012696995401580429,
+      "loss": 1.0134,
+      "step": 8184
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012696102504576096,
+      "loss": 0.9952,
+      "step": 8185
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012695209607571767,
+      "loss": 1.1581,
+      "step": 8186
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012694316710567437,
+      "loss": 0.9337,
+      "step": 8187
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012693423813563105,
+      "loss": 1.0896,
+      "step": 8188
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012692530916558775,
+      "loss": 0.9206,
+      "step": 8189
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012691638019554446,
+      "loss": 0.9567,
+      "step": 8190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012690745122550113,
+      "loss": 0.9516,
+      "step": 8191
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012689852225545784,
+      "loss": 1.0619,
+      "step": 8192
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012688959328541454,
+      "loss": 1.0656,
+      "step": 8193
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012688066431537122,
+      "loss": 1.3056,
+      "step": 8194
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012687173534532792,
+      "loss": 1.0212,
+      "step": 8195
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012686280637528463,
+      "loss": 0.9857,
+      "step": 8196
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001268538774052413,
+      "loss": 1.0191,
+      "step": 8197
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000126844948435198,
+      "loss": 1.1229,
+      "step": 8198
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001268360194651547,
+      "loss": 0.9128,
+      "step": 8199
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001268270904951114,
+      "loss": 0.9477,
+      "step": 8200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001268181615250681,
+      "loss": 1.0493,
+      "step": 8201
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001268092325550248,
+      "loss": 1.0012,
+      "step": 8202
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012680030358498147,
+      "loss": 0.9585,
+      "step": 8203
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012679137461493818,
+      "loss": 1.234,
+      "step": 8204
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012678244564489488,
+      "loss": 1.0,
+      "step": 8205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012677351667485156,
+      "loss": 1.1354,
+      "step": 8206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012676458770480826,
+      "loss": 1.0273,
+      "step": 8207
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012675565873476497,
+      "loss": 1.0975,
+      "step": 8208
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012674672976472164,
+      "loss": 1.0372,
+      "step": 8209
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012673780079467835,
+      "loss": 1.0658,
+      "step": 8210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012672887182463505,
+      "loss": 1.0039,
+      "step": 8211
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012671994285459173,
+      "loss": 1.1677,
+      "step": 8212
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012671101388454843,
+      "loss": 1.0265,
+      "step": 8213
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012670208491450514,
+      "loss": 0.9792,
+      "step": 8214
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001266931559444618,
+      "loss": 1.1344,
+      "step": 8215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012668422697441852,
+      "loss": 1.1545,
+      "step": 8216
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012667529800437522,
+      "loss": 1.0203,
+      "step": 8217
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001266663690343319,
+      "loss": 0.9614,
+      "step": 8218
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001266574400642886,
+      "loss": 1.0753,
+      "step": 8219
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012664851109424528,
+      "loss": 1.0171,
+      "step": 8220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012663958212420198,
+      "loss": 0.9729,
+      "step": 8221
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001266306531541587,
+      "loss": 1.191,
+      "step": 8222
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012662172418411536,
+      "loss": 0.8749,
+      "step": 8223
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012661279521407207,
+      "loss": 0.86,
+      "step": 8224
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012660386624402874,
+      "loss": 0.9822,
+      "step": 8225
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012659493727398545,
+      "loss": 0.9321,
+      "step": 8226
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012658600830394213,
+      "loss": 1.0219,
+      "step": 8227
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012657707933389883,
+      "loss": 1.309,
+      "step": 8228
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012656815036385553,
+      "loss": 0.9273,
+      "step": 8229
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001265592213938122,
+      "loss": 1.0044,
+      "step": 8230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012655029242376891,
+      "loss": 0.9901,
+      "step": 8231
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012654136345372562,
+      "loss": 1.0031,
+      "step": 8232
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001265324344836823,
+      "loss": 1.1035,
+      "step": 8233
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000126523505513639,
+      "loss": 1.1777,
+      "step": 8234
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001265145765435957,
+      "loss": 1.1695,
+      "step": 8235
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012650564757355238,
+      "loss": 1.024,
+      "step": 8236
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012649671860350908,
+      "loss": 1.0642,
+      "step": 8237
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001264877896334658,
+      "loss": 1.0064,
+      "step": 8238
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012647886066342247,
+      "loss": 1.0671,
+      "step": 8239
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012646993169337917,
+      "loss": 0.9964,
+      "step": 8240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012646100272333587,
+      "loss": 1.0397,
+      "step": 8241
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012645207375329255,
+      "loss": 0.9806,
+      "step": 8242
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012644314478324925,
+      "loss": 1.1001,
+      "step": 8243
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012643421581320596,
+      "loss": 1.0995,
+      "step": 8244
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012642528684316264,
+      "loss": 0.9543,
+      "step": 8245
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012641635787311934,
+      "loss": 1.0434,
+      "step": 8246
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012640742890307604,
+      "loss": 0.9646,
+      "step": 8247
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012639849993303272,
+      "loss": 1.0353,
+      "step": 8248
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012638957096298942,
+      "loss": 1.0481,
+      "step": 8249
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00012638064199294613,
+      "loss": 1.058,
+      "step": 8250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001263717130229028,
+      "loss": 0.9469,
+      "step": 8251
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001263627840528595,
+      "loss": 0.995,
+      "step": 8252
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012635385508281621,
+      "loss": 1.1336,
+      "step": 8253
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001263449261127729,
+      "loss": 1.043,
+      "step": 8254
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001263359971427296,
+      "loss": 1.0622,
+      "step": 8255
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001263270681726863,
+      "loss": 0.9972,
+      "step": 8256
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012631813920264298,
+      "loss": 1.1498,
+      "step": 8257
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012630921023259968,
+      "loss": 0.96,
+      "step": 8258
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012630028126255638,
+      "loss": 1.0619,
+      "step": 8259
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012629135229251306,
+      "loss": 1.2329,
+      "step": 8260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012628242332246976,
+      "loss": 1.0383,
+      "step": 8261
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012627349435242647,
+      "loss": 0.9313,
+      "step": 8262
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012626456538238315,
+      "loss": 1.0778,
+      "step": 8263
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012625563641233985,
+      "loss": 1.0675,
+      "step": 8264
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012624670744229655,
+      "loss": 1.0603,
+      "step": 8265
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012623777847225323,
+      "loss": 0.9336,
+      "step": 8266
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012622884950220993,
+      "loss": 0.9432,
+      "step": 8267
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001262199205321666,
+      "loss": 1.0596,
+      "step": 8268
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012621099156212332,
+      "loss": 1.1041,
+      "step": 8269
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012620206259208,
+      "loss": 1.0575,
+      "step": 8270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001261931336220367,
+      "loss": 0.9722,
+      "step": 8271
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001261842046519934,
+      "loss": 0.9891,
+      "step": 8272
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012617527568195008,
+      "loss": 0.9785,
+      "step": 8273
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012616634671190678,
+      "loss": 1.0681,
+      "step": 8274
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012615741774186349,
+      "loss": 0.956,
+      "step": 8275
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012614848877182016,
+      "loss": 0.9447,
+      "step": 8276
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012613955980177687,
+      "loss": 0.952,
+      "step": 8277
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012613063083173357,
+      "loss": 0.945,
+      "step": 8278
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012612170186169025,
+      "loss": 1.1386,
+      "step": 8279
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012611277289164695,
+      "loss": 1.2251,
+      "step": 8280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012610384392160366,
+      "loss": 1.0671,
+      "step": 8281
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012609491495156033,
+      "loss": 0.9935,
+      "step": 8282
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012608598598151704,
+      "loss": 1.0781,
+      "step": 8283
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012607705701147374,
+      "loss": 0.9049,
+      "step": 8284
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012606812804143042,
+      "loss": 1.0294,
+      "step": 8285
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012605919907138712,
+      "loss": 1.0733,
+      "step": 8286
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012605027010134383,
+      "loss": 0.9693,
+      "step": 8287
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001260413411313005,
+      "loss": 1.0002,
+      "step": 8288
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001260324121612572,
+      "loss": 1.0004,
+      "step": 8289
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001260234831912139,
+      "loss": 1.1194,
+      "step": 8290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001260145542211706,
+      "loss": 0.9701,
+      "step": 8291
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001260056252511273,
+      "loss": 1.0154,
+      "step": 8292
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000125996696281084,
+      "loss": 1.2639,
+      "step": 8293
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012598776731104067,
+      "loss": 1.0282,
+      "step": 8294
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012597883834099738,
+      "loss": 1.0948,
+      "step": 8295
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012596990937095408,
+      "loss": 1.132,
+      "step": 8296
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012596098040091076,
+      "loss": 0.965,
+      "step": 8297
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012595205143086746,
+      "loss": 0.9841,
+      "step": 8298
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012594312246082417,
+      "loss": 0.9514,
+      "step": 8299
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012593419349078084,
+      "loss": 1.0567,
+      "step": 8300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012592526452073755,
+      "loss": 0.9089,
+      "step": 8301
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012591633555069425,
+      "loss": 1.0824,
+      "step": 8302
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012590740658065093,
+      "loss": 0.9902,
+      "step": 8303
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012589847761060763,
+      "loss": 1.0004,
+      "step": 8304
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012588954864056434,
+      "loss": 1.0184,
+      "step": 8305
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000125880619670521,
+      "loss": 1.0023,
+      "step": 8306
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012587169070047772,
+      "loss": 1.1789,
+      "step": 8307
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012586276173043442,
+      "loss": 1.1514,
+      "step": 8308
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001258538327603911,
+      "loss": 1.0426,
+      "step": 8309
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001258449037903478,
+      "loss": 1.0018,
+      "step": 8310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012583597482030448,
+      "loss": 1.0495,
+      "step": 8311
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012582704585026118,
+      "loss": 1.0285,
+      "step": 8312
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001258181168802179,
+      "loss": 1.0271,
+      "step": 8313
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012580918791017456,
+      "loss": 0.9625,
+      "step": 8314
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012580025894013127,
+      "loss": 0.997,
+      "step": 8315
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012579132997008794,
+      "loss": 0.9744,
+      "step": 8316
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012578240100004465,
+      "loss": 0.996,
+      "step": 8317
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012577347203000133,
+      "loss": 1.0202,
+      "step": 8318
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012576454305995803,
+      "loss": 1.0609,
+      "step": 8319
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012575561408991473,
+      "loss": 1.0149,
+      "step": 8320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001257466851198714,
+      "loss": 1.0106,
+      "step": 8321
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012573775614982811,
+      "loss": 1.037,
+      "step": 8322
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012572882717978482,
+      "loss": 0.9981,
+      "step": 8323
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001257198982097415,
+      "loss": 0.9466,
+      "step": 8324
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001257109692396982,
+      "loss": 1.0105,
+      "step": 8325
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001257020402696549,
+      "loss": 0.967,
+      "step": 8326
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00012569311129961158,
+      "loss": 0.9923,
+      "step": 8327
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012568418232956828,
+      "loss": 0.9834,
+      "step": 8328
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000125675253359525,
+      "loss": 1.1705,
+      "step": 8329
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012566632438948167,
+      "loss": 0.9983,
+      "step": 8330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012565739541943837,
+      "loss": 1.082,
+      "step": 8331
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012564846644939507,
+      "loss": 1.0493,
+      "step": 8332
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012563953747935175,
+      "loss": 0.9775,
+      "step": 8333
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012563060850930845,
+      "loss": 0.9926,
+      "step": 8334
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012562167953926516,
+      "loss": 1.0325,
+      "step": 8335
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012561275056922184,
+      "loss": 1.0554,
+      "step": 8336
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012560382159917854,
+      "loss": 1.0246,
+      "step": 8337
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012559489262913524,
+      "loss": 0.9535,
+      "step": 8338
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012558596365909192,
+      "loss": 1.09,
+      "step": 8339
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012557703468904862,
+      "loss": 1.1171,
+      "step": 8340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012556810571900533,
+      "loss": 1.0874,
+      "step": 8341
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000125559176748962,
+      "loss": 1.0102,
+      "step": 8342
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001255502477789187,
+      "loss": 1.0281,
+      "step": 8343
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001255413188088754,
+      "loss": 1.0354,
+      "step": 8344
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001255323898388321,
+      "loss": 0.9247,
+      "step": 8345
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001255234608687888,
+      "loss": 1.0041,
+      "step": 8346
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001255145318987455,
+      "loss": 1.0968,
+      "step": 8347
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012550560292870218,
+      "loss": 1.054,
+      "step": 8348
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012549667395865888,
+      "loss": 0.9781,
+      "step": 8349
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012548774498861558,
+      "loss": 1.0047,
+      "step": 8350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012547881601857226,
+      "loss": 0.9685,
+      "step": 8351
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012546988704852896,
+      "loss": 1.0115,
+      "step": 8352
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012546095807848567,
+      "loss": 1.0238,
+      "step": 8353
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012545202910844235,
+      "loss": 1.0673,
+      "step": 8354
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012544310013839905,
+      "loss": 1.0172,
+      "step": 8355
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012543417116835575,
+      "loss": 0.9862,
+      "step": 8356
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012542524219831243,
+      "loss": 0.8868,
+      "step": 8357
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012541631322826913,
+      "loss": 1.0863,
+      "step": 8358
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001254073842582258,
+      "loss": 1.1599,
+      "step": 8359
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012539845528818252,
+      "loss": 1.0527,
+      "step": 8360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001253895263181392,
+      "loss": 1.0497,
+      "step": 8361
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001253805973480959,
+      "loss": 1.0727,
+      "step": 8362
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001253716683780526,
+      "loss": 0.9962,
+      "step": 8363
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012536273940800928,
+      "loss": 0.938,
+      "step": 8364
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012535381043796598,
+      "loss": 0.9576,
+      "step": 8365
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012534488146792266,
+      "loss": 1.0836,
+      "step": 8366
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012533595249787936,
+      "loss": 1.0458,
+      "step": 8367
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012532702352783607,
+      "loss": 0.9922,
+      "step": 8368
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012531809455779277,
+      "loss": 1.0558,
+      "step": 8369
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012530916558774945,
+      "loss": 0.9787,
+      "step": 8370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012530023661770615,
+      "loss": 1.0832,
+      "step": 8371
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012529130764766286,
+      "loss": 1.1873,
+      "step": 8372
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012528237867761953,
+      "loss": 1.0443,
+      "step": 8373
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012527344970757624,
+      "loss": 1.0524,
+      "step": 8374
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012526452073753294,
+      "loss": 1.0491,
+      "step": 8375
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012525559176748962,
+      "loss": 1.132,
+      "step": 8376
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012524666279744632,
+      "loss": 0.9739,
+      "step": 8377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012523773382740303,
+      "loss": 1.1169,
+      "step": 8378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001252288048573597,
+      "loss": 1.1332,
+      "step": 8379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001252198758873164,
+      "loss": 0.9714,
+      "step": 8380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001252109469172731,
+      "loss": 1.1208,
+      "step": 8381
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001252020179472298,
+      "loss": 1.0406,
+      "step": 8382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001251930889771865,
+      "loss": 0.9517,
+      "step": 8383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001251841600071432,
+      "loss": 1.0808,
+      "step": 8384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012517523103709987,
+      "loss": 0.9799,
+      "step": 8385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012516630206705658,
+      "loss": 1.0329,
+      "step": 8386
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012515737309701328,
+      "loss": 1.0293,
+      "step": 8387
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012514844412696996,
+      "loss": 1.0419,
+      "step": 8388
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012513951515692666,
+      "loss": 1.0909,
+      "step": 8389
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012513058618688337,
+      "loss": 0.9458,
+      "step": 8390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012512165721684004,
+      "loss": 1.0741,
+      "step": 8391
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012511272824679675,
+      "loss": 0.9777,
+      "step": 8392
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012510379927675345,
+      "loss": 0.9476,
+      "step": 8393
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012509487030671013,
+      "loss": 1.0054,
+      "step": 8394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012508594133666683,
+      "loss": 0.9159,
+      "step": 8395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012507701236662354,
+      "loss": 1.1161,
+      "step": 8396
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001250680833965802,
+      "loss": 0.973,
+      "step": 8397
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012505915442653692,
+      "loss": 1.0194,
+      "step": 8398
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012505022545649362,
+      "loss": 1.1242,
+      "step": 8399
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001250412964864503,
+      "loss": 1.037,
+      "step": 8400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000125032367516407,
+      "loss": 0.9912,
+      "step": 8401
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00012502343854636368,
+      "loss": 1.0386,
+      "step": 8402
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012501450957632038,
+      "loss": 1.0186,
+      "step": 8403
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012500558060627709,
+      "loss": 1.0281,
+      "step": 8404
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012499665163623376,
+      "loss": 0.9731,
+      "step": 8405
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012498772266619047,
+      "loss": 0.9324,
+      "step": 8406
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012497879369614714,
+      "loss": 0.9428,
+      "step": 8407
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012496986472610385,
+      "loss": 0.9996,
+      "step": 8408
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012496093575606053,
+      "loss": 1.0774,
+      "step": 8409
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012495200678601723,
+      "loss": 0.9288,
+      "step": 8410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012494307781597393,
+      "loss": 1.207,
+      "step": 8411
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001249341488459306,
+      "loss": 0.9996,
+      "step": 8412
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012492521987588731,
+      "loss": 1.094,
+      "step": 8413
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012491629090584402,
+      "loss": 0.9348,
+      "step": 8414
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001249073619358007,
+      "loss": 1.1335,
+      "step": 8415
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001248984329657574,
+      "loss": 1.2264,
+      "step": 8416
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001248895039957141,
+      "loss": 1.0481,
+      "step": 8417
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012488057502567078,
+      "loss": 0.9739,
+      "step": 8418
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012487164605562748,
+      "loss": 1.0735,
+      "step": 8419
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001248627170855842,
+      "loss": 1.1485,
+      "step": 8420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012485378811554087,
+      "loss": 0.8404,
+      "step": 8421
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012484485914549757,
+      "loss": 0.9288,
+      "step": 8422
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012483593017545427,
+      "loss": 0.8816,
+      "step": 8423
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012482700120541095,
+      "loss": 0.9407,
+      "step": 8424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012481807223536765,
+      "loss": 0.932,
+      "step": 8425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012480914326532436,
+      "loss": 0.9376,
+      "step": 8426
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012480021429528104,
+      "loss": 1.0377,
+      "step": 8427
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012479128532523774,
+      "loss": 1.3985,
+      "step": 8428
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012478235635519444,
+      "loss": 1.0829,
+      "step": 8429
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012477342738515112,
+      "loss": 1.1622,
+      "step": 8430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012476449841510782,
+      "loss": 1.0651,
+      "step": 8431
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012475556944506453,
+      "loss": 1.045,
+      "step": 8432
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001247466404750212,
+      "loss": 1.0266,
+      "step": 8433
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001247377115049779,
+      "loss": 0.9412,
+      "step": 8434
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001247287825349346,
+      "loss": 1.191,
+      "step": 8435
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001247198535648913,
+      "loss": 1.1474,
+      "step": 8436
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000124710924594848,
+      "loss": 1.0513,
+      "step": 8437
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001247019956248047,
+      "loss": 1.1312,
+      "step": 8438
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012469306665476138,
+      "loss": 0.9612,
+      "step": 8439
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012468413768471808,
+      "loss": 1.05,
+      "step": 8440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012467520871467478,
+      "loss": 1.1284,
+      "step": 8441
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012466627974463146,
+      "loss": 0.9446,
+      "step": 8442
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012465735077458816,
+      "loss": 1.1027,
+      "step": 8443
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012464842180454487,
+      "loss": 1.0089,
+      "step": 8444
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012463949283450155,
+      "loss": 0.9384,
+      "step": 8445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012463056386445825,
+      "loss": 0.8843,
+      "step": 8446
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012462163489441495,
+      "loss": 1.0114,
+      "step": 8447
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012461270592437163,
+      "loss": 0.9837,
+      "step": 8448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012460377695432833,
+      "loss": 1.0461,
+      "step": 8449
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000124594847984285,
+      "loss": 1.0074,
+      "step": 8450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012458591901424172,
+      "loss": 1.0327,
+      "step": 8451
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001245769900441984,
+      "loss": 1.1218,
+      "step": 8452
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001245680610741551,
+      "loss": 0.9128,
+      "step": 8453
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012455913210411177,
+      "loss": 0.9563,
+      "step": 8454
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012455020313406848,
+      "loss": 1.2408,
+      "step": 8455
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012454127416402518,
+      "loss": 1.0001,
+      "step": 8456
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012453234519398186,
+      "loss": 1.2755,
+      "step": 8457
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012452341622393856,
+      "loss": 0.8936,
+      "step": 8458
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012451448725389527,
+      "loss": 1.0302,
+      "step": 8459
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012450555828385194,
+      "loss": 1.0386,
+      "step": 8460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012449662931380865,
+      "loss": 1.0673,
+      "step": 8461
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012448770034376535,
+      "loss": 0.8876,
+      "step": 8462
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012447877137372203,
+      "loss": 1.0424,
+      "step": 8463
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012446984240367873,
+      "loss": 1.1172,
+      "step": 8464
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012446091343363544,
+      "loss": 0.922,
+      "step": 8465
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012445198446359214,
+      "loss": 0.9958,
+      "step": 8466
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012444305549354882,
+      "loss": 0.9808,
+      "step": 8467
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012443412652350552,
+      "loss": 0.9634,
+      "step": 8468
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012442519755346223,
+      "loss": 0.9761,
+      "step": 8469
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001244162685834189,
+      "loss": 1.0165,
+      "step": 8470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001244073396133756,
+      "loss": 0.9524,
+      "step": 8471
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001243984106433323,
+      "loss": 1.01,
+      "step": 8472
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000124389481673289,
+      "loss": 1.168,
+      "step": 8473
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001243805527032457,
+      "loss": 0.9685,
+      "step": 8474
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001243716237332024,
+      "loss": 0.9713,
+      "step": 8475
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00012436269476315907,
+      "loss": 1.1778,
+      "step": 8476
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012435376579311578,
+      "loss": 0.9685,
+      "step": 8477
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012434483682307248,
+      "loss": 0.9796,
+      "step": 8478
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012433590785302916,
+      "loss": 0.9909,
+      "step": 8479
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012432697888298586,
+      "loss": 1.0078,
+      "step": 8480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012431804991294256,
+      "loss": 0.9897,
+      "step": 8481
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012430912094289924,
+      "loss": 1.0836,
+      "step": 8482
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012430019197285595,
+      "loss": 1.2933,
+      "step": 8483
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012429126300281265,
+      "loss": 1.0961,
+      "step": 8484
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012428233403276933,
+      "loss": 1.0726,
+      "step": 8485
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012427340506272603,
+      "loss": 1.0078,
+      "step": 8486
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012426447609268273,
+      "loss": 1.1365,
+      "step": 8487
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001242555471226394,
+      "loss": 1.0953,
+      "step": 8488
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012424661815259612,
+      "loss": 1.0458,
+      "step": 8489
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012423768918255282,
+      "loss": 1.2407,
+      "step": 8490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001242287602125095,
+      "loss": 0.9967,
+      "step": 8491
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001242198312424662,
+      "loss": 0.9041,
+      "step": 8492
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012421090227242288,
+      "loss": 1.0136,
+      "step": 8493
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012420197330237958,
+      "loss": 0.9632,
+      "step": 8494
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012419304433233629,
+      "loss": 1.1165,
+      "step": 8495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012418411536229296,
+      "loss": 0.9039,
+      "step": 8496
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012417518639224967,
+      "loss": 1.2222,
+      "step": 8497
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012416625742220634,
+      "loss": 1.1008,
+      "step": 8498
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012415732845216305,
+      "loss": 1.0211,
+      "step": 8499
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012414839948211972,
+      "loss": 0.9563,
+      "step": 8500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012413947051207643,
+      "loss": 1.0866,
+      "step": 8501
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012413054154203313,
+      "loss": 1.069,
+      "step": 8502
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001241216125719898,
+      "loss": 1.0497,
+      "step": 8503
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012411268360194651,
+      "loss": 0.9797,
+      "step": 8504
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012410375463190322,
+      "loss": 1.188,
+      "step": 8505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001240948256618599,
+      "loss": 0.975,
+      "step": 8506
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001240858966918166,
+      "loss": 0.9272,
+      "step": 8507
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001240769677217733,
+      "loss": 1.0231,
+      "step": 8508
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012406803875172998,
+      "loss": 0.951,
+      "step": 8509
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012405910978168668,
+      "loss": 1.1212,
+      "step": 8510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001240501808116434,
+      "loss": 1.1522,
+      "step": 8511
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012404125184160006,
+      "loss": 1.1036,
+      "step": 8512
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012403232287155677,
+      "loss": 1.0555,
+      "step": 8513
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012402339390151347,
+      "loss": 1.0671,
+      "step": 8514
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012401446493147015,
+      "loss": 0.9801,
+      "step": 8515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012400553596142685,
+      "loss": 1.153,
+      "step": 8516
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012399660699138356,
+      "loss": 1.1355,
+      "step": 8517
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012398767802134023,
+      "loss": 1.0865,
+      "step": 8518
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012397874905129694,
+      "loss": 1.0414,
+      "step": 8519
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012396982008125364,
+      "loss": 1.1408,
+      "step": 8520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012396089111121032,
+      "loss": 1.0442,
+      "step": 8521
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012395196214116702,
+      "loss": 0.9419,
+      "step": 8522
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012394303317112373,
+      "loss": 1.1357,
+      "step": 8523
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001239341042010804,
+      "loss": 1.1325,
+      "step": 8524
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001239251752310371,
+      "loss": 0.8946,
+      "step": 8525
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001239162462609938,
+      "loss": 1.11,
+      "step": 8526
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001239073172909505,
+      "loss": 1.005,
+      "step": 8527
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001238983883209072,
+      "loss": 0.9392,
+      "step": 8528
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001238894593508639,
+      "loss": 1.0323,
+      "step": 8529
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012388053038082057,
+      "loss": 1.0002,
+      "step": 8530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012387160141077728,
+      "loss": 1.0281,
+      "step": 8531
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012386267244073398,
+      "loss": 1.0721,
+      "step": 8532
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012385374347069066,
+      "loss": 1.159,
+      "step": 8533
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012384481450064736,
+      "loss": 0.9935,
+      "step": 8534
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012383588553060407,
+      "loss": 1.0641,
+      "step": 8535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012382695656056074,
+      "loss": 1.0502,
+      "step": 8536
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012381802759051745,
+      "loss": 1.0521,
+      "step": 8537
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012380909862047415,
+      "loss": 1.0862,
+      "step": 8538
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012380016965043083,
+      "loss": 0.8714,
+      "step": 8539
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012379124068038753,
+      "loss": 1.1851,
+      "step": 8540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001237823117103442,
+      "loss": 0.9978,
+      "step": 8541
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012377338274030091,
+      "loss": 1.1368,
+      "step": 8542
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001237644537702576,
+      "loss": 1.0244,
+      "step": 8543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001237555248002143,
+      "loss": 1.2156,
+      "step": 8544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012374659583017097,
+      "loss": 1.046,
+      "step": 8545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012373766686012768,
+      "loss": 0.9321,
+      "step": 8546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012372873789008438,
+      "loss": 0.9806,
+      "step": 8547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012371980892004106,
+      "loss": 1.1625,
+      "step": 8548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012371087994999776,
+      "loss": 1.0245,
+      "step": 8549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012370195097995447,
+      "loss": 0.9942,
+      "step": 8550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00012369302200991114,
+      "loss": 1.0337,
+      "step": 8551
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012368409303986785,
+      "loss": 0.9803,
+      "step": 8552
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012367516406982455,
+      "loss": 1.0304,
+      "step": 8553
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012366623509978123,
+      "loss": 1.0988,
+      "step": 8554
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012365730612973793,
+      "loss": 0.8649,
+      "step": 8555
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012364837715969464,
+      "loss": 0.9822,
+      "step": 8556
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001236394481896513,
+      "loss": 0.9965,
+      "step": 8557
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012363051921960802,
+      "loss": 0.9904,
+      "step": 8558
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012362159024956472,
+      "loss": 1.0433,
+      "step": 8559
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001236126612795214,
+      "loss": 1.0941,
+      "step": 8560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001236037323094781,
+      "loss": 1.0427,
+      "step": 8561
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001235948033394348,
+      "loss": 1.0451,
+      "step": 8562
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012358587436939148,
+      "loss": 1.1118,
+      "step": 8563
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001235769453993482,
+      "loss": 0.9718,
+      "step": 8564
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001235680164293049,
+      "loss": 1.0043,
+      "step": 8565
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001235590874592616,
+      "loss": 1.0897,
+      "step": 8566
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012355015848921827,
+      "loss": 1.0092,
+      "step": 8567
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012354122951917498,
+      "loss": 1.076,
+      "step": 8568
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012353230054913168,
+      "loss": 1.0417,
+      "step": 8569
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012352337157908836,
+      "loss": 1.036,
+      "step": 8570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012351444260904506,
+      "loss": 1.0567,
+      "step": 8571
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012350551363900176,
+      "loss": 1.0245,
+      "step": 8572
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012349658466895844,
+      "loss": 1.108,
+      "step": 8573
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012348765569891515,
+      "loss": 0.9571,
+      "step": 8574
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012347872672887185,
+      "loss": 1.0987,
+      "step": 8575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012346979775882853,
+      "loss": 0.9438,
+      "step": 8576
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012346086878878523,
+      "loss": 1.0189,
+      "step": 8577
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012345193981874193,
+      "loss": 0.9245,
+      "step": 8578
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001234430108486986,
+      "loss": 1.1505,
+      "step": 8579
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012343408187865532,
+      "loss": 0.9944,
+      "step": 8580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012342515290861202,
+      "loss": 0.8633,
+      "step": 8581
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001234162239385687,
+      "loss": 1.0691,
+      "step": 8582
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001234072949685254,
+      "loss": 1.1758,
+      "step": 8583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012339836599848208,
+      "loss": 1.086,
+      "step": 8584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012338943702843878,
+      "loss": 1.0528,
+      "step": 8585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012338050805839546,
+      "loss": 1.0253,
+      "step": 8586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012337157908835216,
+      "loss": 1.0038,
+      "step": 8587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012336265011830887,
+      "loss": 1.2323,
+      "step": 8588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012335372114826554,
+      "loss": 1.0278,
+      "step": 8589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012334479217822225,
+      "loss": 0.9975,
+      "step": 8590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012333586320817892,
+      "loss": 1.0519,
+      "step": 8591
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012332693423813563,
+      "loss": 1.0206,
+      "step": 8592
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012331800526809233,
+      "loss": 1.1151,
+      "step": 8593
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000123309076298049,
+      "loss": 1.028,
+      "step": 8594
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001233001473280057,
+      "loss": 1.0312,
+      "step": 8595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012329121835796242,
+      "loss": 1.1044,
+      "step": 8596
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001232822893879191,
+      "loss": 1.1742,
+      "step": 8597
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001232733604178758,
+      "loss": 0.9387,
+      "step": 8598
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001232644314478325,
+      "loss": 0.9705,
+      "step": 8599
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012325550247778918,
+      "loss": 1.0057,
+      "step": 8600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012324657350774588,
+      "loss": 1.0999,
+      "step": 8601
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001232376445377026,
+      "loss": 1.0873,
+      "step": 8602
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012322871556765926,
+      "loss": 1.0371,
+      "step": 8603
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012321978659761597,
+      "loss": 0.994,
+      "step": 8604
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012321085762757267,
+      "loss": 1.0753,
+      "step": 8605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012320192865752935,
+      "loss": 0.9708,
+      "step": 8606
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012319299968748605,
+      "loss": 1.0494,
+      "step": 8607
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012318407071744276,
+      "loss": 0.9084,
+      "step": 8608
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012317514174739943,
+      "loss": 1.0184,
+      "step": 8609
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012316621277735614,
+      "loss": 0.9532,
+      "step": 8610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012315728380731284,
+      "loss": 0.8628,
+      "step": 8611
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012314835483726952,
+      "loss": 1.07,
+      "step": 8612
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012313942586722622,
+      "loss": 1.0596,
+      "step": 8613
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012313049689718293,
+      "loss": 0.9855,
+      "step": 8614
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001231215679271396,
+      "loss": 1.0473,
+      "step": 8615
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001231126389570963,
+      "loss": 1.0897,
+      "step": 8616
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000123103709987053,
+      "loss": 1.0608,
+      "step": 8617
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001230947810170097,
+      "loss": 1.0629,
+      "step": 8618
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001230858520469664,
+      "loss": 1.1878,
+      "step": 8619
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.9754,
+      "step": 8620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012306799410687977,
+      "loss": 1.0083,
+      "step": 8621
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012305906513683648,
+      "loss": 1.0521,
+      "step": 8622
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012305013616679318,
+      "loss": 1.0804,
+      "step": 8623
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012304120719674986,
+      "loss": 1.0438,
+      "step": 8624
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012303227822670656,
+      "loss": 1.1506,
+      "step": 8625
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00012302334925666327,
+      "loss": 1.1108,
+      "step": 8626
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012301442028661994,
+      "loss": 1.0324,
+      "step": 8627
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012300549131657665,
+      "loss": 0.9804,
+      "step": 8628
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012299656234653335,
+      "loss": 1.06,
+      "step": 8629
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012298763337649003,
+      "loss": 1.1498,
+      "step": 8630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012297870440644673,
+      "loss": 1.0591,
+      "step": 8631
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001229697754364034,
+      "loss": 1.0456,
+      "step": 8632
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012296084646636011,
+      "loss": 1.0314,
+      "step": 8633
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001229519174963168,
+      "loss": 1.0135,
+      "step": 8634
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001229429885262735,
+      "loss": 1.1038,
+      "step": 8635
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012293405955623017,
+      "loss": 1.123,
+      "step": 8636
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012292513058618688,
+      "loss": 0.9528,
+      "step": 8637
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012291620161614358,
+      "loss": 1.0233,
+      "step": 8638
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012290727264610026,
+      "loss": 0.9997,
+      "step": 8639
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012289834367605696,
+      "loss": 1.0811,
+      "step": 8640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012288941470601367,
+      "loss": 0.9568,
+      "step": 8641
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012288048573597034,
+      "loss": 1.0844,
+      "step": 8642
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012287155676592705,
+      "loss": 1.04,
+      "step": 8643
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012286262779588375,
+      "loss": 1.0277,
+      "step": 8644
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012285369882584043,
+      "loss": 1.1548,
+      "step": 8645
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012284476985579713,
+      "loss": 0.987,
+      "step": 8646
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012283584088575384,
+      "loss": 1.0368,
+      "step": 8647
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001228269119157105,
+      "loss": 1.0106,
+      "step": 8648
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012281798294566722,
+      "loss": 1.1128,
+      "step": 8649
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012280905397562392,
+      "loss": 0.9979,
+      "step": 8650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001228001250055806,
+      "loss": 1.0306,
+      "step": 8651
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001227911960355373,
+      "loss": 0.9146,
+      "step": 8652
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000122782267065494,
+      "loss": 0.9834,
+      "step": 8653
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012277333809545068,
+      "loss": 1.0754,
+      "step": 8654
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012276440912540739,
+      "loss": 1.038,
+      "step": 8655
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001227554801553641,
+      "loss": 1.023,
+      "step": 8656
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012274655118532077,
+      "loss": 0.9489,
+      "step": 8657
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012273762221527747,
+      "loss": 1.144,
+      "step": 8658
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012272869324523418,
+      "loss": 1.0186,
+      "step": 8659
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012271976427519085,
+      "loss": 1.0065,
+      "step": 8660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012271083530514756,
+      "loss": 0.9812,
+      "step": 8661
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012270190633510426,
+      "loss": 1.1043,
+      "step": 8662
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012269297736506096,
+      "loss": 1.0748,
+      "step": 8663
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012268404839501764,
+      "loss": 0.9506,
+      "step": 8664
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012267511942497435,
+      "loss": 1.114,
+      "step": 8665
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012266619045493105,
+      "loss": 1.1945,
+      "step": 8666
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012265726148488773,
+      "loss": 0.9615,
+      "step": 8667
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012264833251484443,
+      "loss": 0.9929,
+      "step": 8668
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012263940354480113,
+      "loss": 1.1547,
+      "step": 8669
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001226304745747578,
+      "loss": 1.1003,
+      "step": 8670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012262154560471452,
+      "loss": 1.012,
+      "step": 8671
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012261261663467122,
+      "loss": 1.0159,
+      "step": 8672
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001226036876646279,
+      "loss": 0.8935,
+      "step": 8673
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001225947586945846,
+      "loss": 1.066,
+      "step": 8674
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012258582972454128,
+      "loss": 0.9886,
+      "step": 8675
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012257690075449798,
+      "loss": 1.0374,
+      "step": 8676
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012256797178445466,
+      "loss": 0.955,
+      "step": 8677
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012255904281441136,
+      "loss": 1.1889,
+      "step": 8678
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012255011384436807,
+      "loss": 1.1174,
+      "step": 8679
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012254118487432474,
+      "loss": 1.0315,
+      "step": 8680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012253225590428145,
+      "loss": 0.9889,
+      "step": 8681
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012252332693423812,
+      "loss": 0.9801,
+      "step": 8682
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012251439796419483,
+      "loss": 1.2371,
+      "step": 8683
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012250546899415153,
+      "loss": 1.1043,
+      "step": 8684
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001224965400241082,
+      "loss": 1.0538,
+      "step": 8685
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001224876110540649,
+      "loss": 1.0458,
+      "step": 8686
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012247868208402162,
+      "loss": 0.9919,
+      "step": 8687
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001224697531139783,
+      "loss": 1.0535,
+      "step": 8688
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000122460824143935,
+      "loss": 0.9681,
+      "step": 8689
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001224518951738917,
+      "loss": 0.9373,
+      "step": 8690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012244296620384838,
+      "loss": 0.9066,
+      "step": 8691
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012243403723380508,
+      "loss": 0.9604,
+      "step": 8692
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001224251082637618,
+      "loss": 1.0894,
+      "step": 8693
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012241617929371846,
+      "loss": 1.1242,
+      "step": 8694
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012240725032367517,
+      "loss": 1.0101,
+      "step": 8695
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012239832135363187,
+      "loss": 1.0684,
+      "step": 8696
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012238939238358855,
+      "loss": 1.0426,
+      "step": 8697
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012238046341354525,
+      "loss": 0.9985,
+      "step": 8698
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012237153444350196,
+      "loss": 1.1897,
+      "step": 8699
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00012236260547345863,
+      "loss": 1.1196,
+      "step": 8700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012235367650341534,
+      "loss": 0.945,
+      "step": 8701
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012234474753337204,
+      "loss": 0.9797,
+      "step": 8702
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012233581856332872,
+      "loss": 1.0939,
+      "step": 8703
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012232688959328542,
+      "loss": 1.128,
+      "step": 8704
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012231796062324213,
+      "loss": 1.0625,
+      "step": 8705
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001223090316531988,
+      "loss": 0.9997,
+      "step": 8706
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001223001026831555,
+      "loss": 1.1697,
+      "step": 8707
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001222911737131122,
+      "loss": 1.1237,
+      "step": 8708
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001222822447430689,
+      "loss": 1.0764,
+      "step": 8709
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001222733157730256,
+      "loss": 1.0594,
+      "step": 8710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001222643868029823,
+      "loss": 1.1321,
+      "step": 8711
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012225545783293897,
+      "loss": 0.8945,
+      "step": 8712
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012224652886289568,
+      "loss": 1.0213,
+      "step": 8713
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012223759989285238,
+      "loss": 0.9053,
+      "step": 8714
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012222867092280906,
+      "loss": 1.115,
+      "step": 8715
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012221974195276576,
+      "loss": 1.0568,
+      "step": 8716
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012221081298272247,
+      "loss": 1.0029,
+      "step": 8717
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012220188401267914,
+      "loss": 0.9686,
+      "step": 8718
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012219295504263585,
+      "loss": 1.0985,
+      "step": 8719
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012218402607259255,
+      "loss": 1.0662,
+      "step": 8720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012217509710254923,
+      "loss": 1.0004,
+      "step": 8721
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012216616813250593,
+      "loss": 1.0611,
+      "step": 8722
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001221572391624626,
+      "loss": 1.0504,
+      "step": 8723
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012214831019241931,
+      "loss": 1.1251,
+      "step": 8724
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000122139381222376,
+      "loss": 1.1916,
+      "step": 8725
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001221304522523327,
+      "loss": 0.8756,
+      "step": 8726
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012212152328228937,
+      "loss": 1.0831,
+      "step": 8727
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012211259431224608,
+      "loss": 1.1474,
+      "step": 8728
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012210366534220278,
+      "loss": 1.2123,
+      "step": 8729
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012209473637215946,
+      "loss": 1.0936,
+      "step": 8730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012208580740211616,
+      "loss": 0.9612,
+      "step": 8731
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012207687843207286,
+      "loss": 1.0231,
+      "step": 8732
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012206794946202956,
+      "loss": 1.0383,
+      "step": 8733
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012205902049198626,
+      "loss": 1.013,
+      "step": 8734
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012205009152194295,
+      "loss": 0.9184,
+      "step": 8735
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012204116255189964,
+      "loss": 1.0241,
+      "step": 8736
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012203223358185633,
+      "loss": 1.0756,
+      "step": 8737
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012202330461181303,
+      "loss": 0.9309,
+      "step": 8738
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012201437564176971,
+      "loss": 1.0765,
+      "step": 8739
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012200544667172642,
+      "loss": 1.0996,
+      "step": 8740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012199651770168312,
+      "loss": 0.9638,
+      "step": 8741
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001219875887316398,
+      "loss": 1.1722,
+      "step": 8742
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001219786597615965,
+      "loss": 1.007,
+      "step": 8743
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001219697307915532,
+      "loss": 1.194,
+      "step": 8744
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012196080182150988,
+      "loss": 1.0766,
+      "step": 8745
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012195187285146659,
+      "loss": 1.0097,
+      "step": 8746
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012194294388142329,
+      "loss": 0.9794,
+      "step": 8747
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012193401491137997,
+      "loss": 1.0815,
+      "step": 8748
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012192508594133667,
+      "loss": 0.9901,
+      "step": 8749
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012191615697129337,
+      "loss": 1.161,
+      "step": 8750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012190722800125005,
+      "loss": 1.1424,
+      "step": 8751
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012189829903120676,
+      "loss": 1.1054,
+      "step": 8752
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012188937006116346,
+      "loss": 0.967,
+      "step": 8753
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012188044109112014,
+      "loss": 0.9118,
+      "step": 8754
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012187151212107684,
+      "loss": 0.9862,
+      "step": 8755
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012186258315103353,
+      "loss": 1.0221,
+      "step": 8756
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012185365418099022,
+      "loss": 1.0527,
+      "step": 8757
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012184472521094693,
+      "loss": 1.0413,
+      "step": 8758
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012183579624090362,
+      "loss": 1.0501,
+      "step": 8759
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012182686727086032,
+      "loss": 0.9784,
+      "step": 8760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000121817938300817,
+      "loss": 1.1025,
+      "step": 8761
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001218090093307737,
+      "loss": 1.0474,
+      "step": 8762
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001218000803607304,
+      "loss": 1.0303,
+      "step": 8763
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012179115139068708,
+      "loss": 0.9879,
+      "step": 8764
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012178222242064379,
+      "loss": 0.9405,
+      "step": 8765
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012177329345060049,
+      "loss": 0.991,
+      "step": 8766
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012176436448055717,
+      "loss": 1.0765,
+      "step": 8767
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012175543551051387,
+      "loss": 1.0027,
+      "step": 8768
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012174650654047058,
+      "loss": 1.0728,
+      "step": 8769
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012173757757042725,
+      "loss": 0.9264,
+      "step": 8770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012172864860038396,
+      "loss": 1.0572,
+      "step": 8771
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012171971963034066,
+      "loss": 1.0328,
+      "step": 8772
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012171079066029734,
+      "loss": 0.9783,
+      "step": 8773
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012170186169025404,
+      "loss": 1.2056,
+      "step": 8774
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00012169293272021075,
+      "loss": 1.1727,
+      "step": 8775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012168400375016742,
+      "loss": 1.2127,
+      "step": 8776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012167507478012413,
+      "loss": 1.1074,
+      "step": 8777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012166614581008082,
+      "loss": 1.0233,
+      "step": 8778
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012165721684003751,
+      "loss": 0.9564,
+      "step": 8779
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001216482878699942,
+      "loss": 1.1046,
+      "step": 8780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001216393588999509,
+      "loss": 0.9594,
+      "step": 8781
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012163042992990759,
+      "loss": 1.0563,
+      "step": 8782
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012162150095986428,
+      "loss": 1.0596,
+      "step": 8783
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012161257198982099,
+      "loss": 0.9944,
+      "step": 8784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012160364301977766,
+      "loss": 1.0507,
+      "step": 8785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012159471404973437,
+      "loss": 1.0622,
+      "step": 8786
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012158578507969107,
+      "loss": 0.9891,
+      "step": 8787
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012157685610964775,
+      "loss": 0.9996,
+      "step": 8788
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012156792713960445,
+      "loss": 0.915,
+      "step": 8789
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012155899816956116,
+      "loss": 1.0519,
+      "step": 8790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012155006919951783,
+      "loss": 0.9841,
+      "step": 8791
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012154114022947454,
+      "loss": 1.1062,
+      "step": 8792
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012153221125943124,
+      "loss": 0.9842,
+      "step": 8793
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012152328228938792,
+      "loss": 1.1604,
+      "step": 8794
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012151435331934462,
+      "loss": 0.9921,
+      "step": 8795
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012150542434930133,
+      "loss": 1.0333,
+      "step": 8796
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000121496495379258,
+      "loss": 1.0184,
+      "step": 8797
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012148756640921471,
+      "loss": 1.1788,
+      "step": 8798
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012147863743917141,
+      "loss": 1.0503,
+      "step": 8799
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012146970846912809,
+      "loss": 1.0637,
+      "step": 8800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012146077949908479,
+      "loss": 1.0699,
+      "step": 8801
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012145185052904148,
+      "loss": 1.0148,
+      "step": 8802
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012144292155899817,
+      "loss": 1.1347,
+      "step": 8803
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012143399258895486,
+      "loss": 1.0751,
+      "step": 8804
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012142506361891157,
+      "loss": 1.1275,
+      "step": 8805
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012141613464886825,
+      "loss": 1.1007,
+      "step": 8806
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012140720567882495,
+      "loss": 1.0465,
+      "step": 8807
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012139827670878165,
+      "loss": 1.088,
+      "step": 8808
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012138934773873833,
+      "loss": 1.0329,
+      "step": 8809
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012138041876869503,
+      "loss": 0.9912,
+      "step": 8810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012137148979865174,
+      "loss": 1.0265,
+      "step": 8811
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012136256082860842,
+      "loss": 1.253,
+      "step": 8812
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012135363185856512,
+      "loss": 0.8208,
+      "step": 8813
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012134470288852182,
+      "loss": 1.0869,
+      "step": 8814
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001213357739184785,
+      "loss": 0.987,
+      "step": 8815
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001213268449484352,
+      "loss": 1.042,
+      "step": 8816
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012131791597839191,
+      "loss": 1.1096,
+      "step": 8817
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012130898700834859,
+      "loss": 1.1061,
+      "step": 8818
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012130005803830529,
+      "loss": 0.9639,
+      "step": 8819
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012129112906826199,
+      "loss": 1.0857,
+      "step": 8820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012128220009821867,
+      "loss": 0.9716,
+      "step": 8821
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012127327112817537,
+      "loss": 1.0497,
+      "step": 8822
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012126434215813208,
+      "loss": 1.0304,
+      "step": 8823
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012125541318808876,
+      "loss": 1.0701,
+      "step": 8824
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012124648421804546,
+      "loss": 1.1551,
+      "step": 8825
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012123755524800215,
+      "loss": 1.0444,
+      "step": 8826
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012122862627795884,
+      "loss": 0.9643,
+      "step": 8827
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012121969730791553,
+      "loss": 1.0161,
+      "step": 8828
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012121076833787223,
+      "loss": 1.0481,
+      "step": 8829
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012120183936782891,
+      "loss": 0.9933,
+      "step": 8830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012119291039778562,
+      "loss": 1.0532,
+      "step": 8831
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012118398142774232,
+      "loss": 1.1058,
+      "step": 8832
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000121175052457699,
+      "loss": 0.9998,
+      "step": 8833
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001211661234876557,
+      "loss": 0.9803,
+      "step": 8834
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001211571945176124,
+      "loss": 1.0269,
+      "step": 8835
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012114826554756908,
+      "loss": 1.0758,
+      "step": 8836
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012113933657752579,
+      "loss": 1.0797,
+      "step": 8837
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012113040760748249,
+      "loss": 0.9951,
+      "step": 8838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012112147863743917,
+      "loss": 0.9566,
+      "step": 8839
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012111254966739587,
+      "loss": 1.0781,
+      "step": 8840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012110362069735257,
+      "loss": 1.0053,
+      "step": 8841
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012109469172730925,
+      "loss": 1.0582,
+      "step": 8842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012108576275726596,
+      "loss": 1.0488,
+      "step": 8843
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012107683378722266,
+      "loss": 1.0051,
+      "step": 8844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012106790481717934,
+      "loss": 1.0517,
+      "step": 8845
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012105897584713604,
+      "loss": 0.9484,
+      "step": 8846
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012105004687709273,
+      "loss": 1.0503,
+      "step": 8847
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012104111790704942,
+      "loss": 1.0469,
+      "step": 8848
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012103218893700613,
+      "loss": 1.1857,
+      "step": 8849
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00012102325996696282,
+      "loss": 0.9255,
+      "step": 8850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001210143309969195,
+      "loss": 1.2626,
+      "step": 8851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001210054020268762,
+      "loss": 0.9652,
+      "step": 8852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001209964730568329,
+      "loss": 0.8914,
+      "step": 8853
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012098754408678958,
+      "loss": 0.8451,
+      "step": 8854
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012097861511674628,
+      "loss": 1.1128,
+      "step": 8855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012096968614670299,
+      "loss": 0.9866,
+      "step": 8856
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012096075717665966,
+      "loss": 1.0195,
+      "step": 8857
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012095182820661637,
+      "loss": 1.1161,
+      "step": 8858
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012094289923657307,
+      "loss": 1.0743,
+      "step": 8859
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012093397026652977,
+      "loss": 1.0637,
+      "step": 8860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012092504129648645,
+      "loss": 0.9839,
+      "step": 8861
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012091611232644316,
+      "loss": 0.8876,
+      "step": 8862
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012090718335639986,
+      "loss": 1.1132,
+      "step": 8863
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012089825438635654,
+      "loss": 1.002,
+      "step": 8864
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012088932541631324,
+      "loss": 1.2428,
+      "step": 8865
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012088039644626994,
+      "loss": 1.1147,
+      "step": 8866
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012087146747622662,
+      "loss": 1.0339,
+      "step": 8867
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012086253850618333,
+      "loss": 1.0026,
+      "step": 8868
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012085360953614002,
+      "loss": 1.0551,
+      "step": 8869
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012084468056609671,
+      "loss": 0.9526,
+      "step": 8870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001208357515960534,
+      "loss": 0.9539,
+      "step": 8871
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001208268226260101,
+      "loss": 1.0328,
+      "step": 8872
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012081789365596679,
+      "loss": 1.1922,
+      "step": 8873
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012080896468592348,
+      "loss": 1.0097,
+      "step": 8874
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012080003571588019,
+      "loss": 1.1896,
+      "step": 8875
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012079110674583686,
+      "loss": 1.0198,
+      "step": 8876
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012078217777579357,
+      "loss": 1.0801,
+      "step": 8877
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012077324880575027,
+      "loss": 0.9745,
+      "step": 8878
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012076431983570695,
+      "loss": 0.9024,
+      "step": 8879
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012075539086566365,
+      "loss": 0.9988,
+      "step": 8880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012074646189562036,
+      "loss": 0.9213,
+      "step": 8881
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012073753292557703,
+      "loss": 1.0455,
+      "step": 8882
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012072860395553374,
+      "loss": 1.0494,
+      "step": 8883
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012071967498549044,
+      "loss": 0.9291,
+      "step": 8884
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012071074601544712,
+      "loss": 1.0087,
+      "step": 8885
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012070181704540382,
+      "loss": 0.963,
+      "step": 8886
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012069288807536053,
+      "loss": 1.0864,
+      "step": 8887
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001206839591053172,
+      "loss": 1.0294,
+      "step": 8888
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012067503013527391,
+      "loss": 0.9362,
+      "step": 8889
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012066610116523061,
+      "loss": 1.0454,
+      "step": 8890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012065717219518729,
+      "loss": 1.1218,
+      "step": 8891
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012064824322514399,
+      "loss": 1.0688,
+      "step": 8892
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012063931425510068,
+      "loss": 0.9746,
+      "step": 8893
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012063038528505737,
+      "loss": 1.1348,
+      "step": 8894
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012062145631501406,
+      "loss": 1.0829,
+      "step": 8895
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012061252734497077,
+      "loss": 0.9583,
+      "step": 8896
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012060359837492744,
+      "loss": 1.0316,
+      "step": 8897
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012059466940488415,
+      "loss": 0.986,
+      "step": 8898
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012058574043484085,
+      "loss": 1.0747,
+      "step": 8899
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012057681146479753,
+      "loss": 1.2265,
+      "step": 8900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012056788249475423,
+      "loss": 1.0598,
+      "step": 8901
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012055895352471094,
+      "loss": 1.0755,
+      "step": 8902
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012055002455466761,
+      "loss": 1.0192,
+      "step": 8903
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012054109558462432,
+      "loss": 1.0336,
+      "step": 8904
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012053216661458102,
+      "loss": 1.2984,
+      "step": 8905
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001205232376445377,
+      "loss": 1.1673,
+      "step": 8906
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001205143086744944,
+      "loss": 1.0522,
+      "step": 8907
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012050537970445111,
+      "loss": 1.0536,
+      "step": 8908
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012049645073440778,
+      "loss": 1.1279,
+      "step": 8909
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012048752176436449,
+      "loss": 1.1083,
+      "step": 8910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012047859279432119,
+      "loss": 1.1022,
+      "step": 8911
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012046966382427787,
+      "loss": 0.9861,
+      "step": 8912
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012046073485423457,
+      "loss": 0.9873,
+      "step": 8913
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012045180588419128,
+      "loss": 1.0263,
+      "step": 8914
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012044287691414795,
+      "loss": 1.0172,
+      "step": 8915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012043394794410466,
+      "loss": 1.0956,
+      "step": 8916
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012042501897406135,
+      "loss": 0.947,
+      "step": 8917
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012041609000401804,
+      "loss": 1.0266,
+      "step": 8918
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012040716103397473,
+      "loss": 1.0963,
+      "step": 8919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012039823206393143,
+      "loss": 1.0049,
+      "step": 8920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012038930309388811,
+      "loss": 0.9754,
+      "step": 8921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012038037412384482,
+      "loss": 1.027,
+      "step": 8922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012037144515380152,
+      "loss": 1.1141,
+      "step": 8923
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001203625161837582,
+      "loss": 0.9549,
+      "step": 8924
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001203535872137149,
+      "loss": 1.0662,
+      "step": 8925
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001203446582436716,
+      "loss": 0.9269,
+      "step": 8926
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012033572927362828,
+      "loss": 0.9967,
+      "step": 8927
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012032680030358499,
+      "loss": 1.0407,
+      "step": 8928
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012031787133354169,
+      "loss": 1.0555,
+      "step": 8929
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012030894236349837,
+      "loss": 1.0644,
+      "step": 8930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012030001339345507,
+      "loss": 0.9435,
+      "step": 8931
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012029108442341177,
+      "loss": 1.1263,
+      "step": 8932
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012028215545336845,
+      "loss": 1.0268,
+      "step": 8933
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012027322648332516,
+      "loss": 1.0971,
+      "step": 8934
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012026429751328186,
+      "loss": 0.9222,
+      "step": 8935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012025536854323854,
+      "loss": 0.9819,
+      "step": 8936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012024643957319524,
+      "loss": 0.9762,
+      "step": 8937
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012023751060315193,
+      "loss": 1.0913,
+      "step": 8938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012022858163310862,
+      "loss": 1.0043,
+      "step": 8939
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012021965266306533,
+      "loss": 0.9429,
+      "step": 8940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012021072369302202,
+      "loss": 1.055,
+      "step": 8941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001202017947229787,
+      "loss": 1.0536,
+      "step": 8942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001201928657529354,
+      "loss": 1.0285,
+      "step": 8943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001201839367828921,
+      "loss": 1.0671,
+      "step": 8944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012017500781284878,
+      "loss": 1.1113,
+      "step": 8945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012016607884280548,
+      "loss": 1.0281,
+      "step": 8946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012015714987276219,
+      "loss": 0.9844,
+      "step": 8947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012014822090271886,
+      "loss": 1.0061,
+      "step": 8948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012013929193267557,
+      "loss": 1.0268,
+      "step": 8949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012013036296263227,
+      "loss": 0.9337,
+      "step": 8950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012012143399258895,
+      "loss": 1.0855,
+      "step": 8951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012011250502254565,
+      "loss": 1.0397,
+      "step": 8952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012010357605250236,
+      "loss": 0.972,
+      "step": 8953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012009464708245903,
+      "loss": 0.9541,
+      "step": 8954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012008571811241574,
+      "loss": 1.0719,
+      "step": 8955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012007678914237244,
+      "loss": 1.011,
+      "step": 8956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012006786017232914,
+      "loss": 1.1072,
+      "step": 8957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012005893120228582,
+      "loss": 1.0265,
+      "step": 8958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012005000223224253,
+      "loss": 1.0775,
+      "step": 8959
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012004107326219922,
+      "loss": 0.9818,
+      "step": 8960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001200321442921559,
+      "loss": 1.1101,
+      "step": 8961
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001200232153221126,
+      "loss": 0.9116,
+      "step": 8962
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001200142863520693,
+      "loss": 0.9553,
+      "step": 8963
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012000535738202598,
+      "loss": 1.0719,
+      "step": 8964
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011999642841198268,
+      "loss": 1.0206,
+      "step": 8965
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011998749944193939,
+      "loss": 0.9722,
+      "step": 8966
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011997857047189606,
+      "loss": 0.9694,
+      "step": 8967
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011996964150185277,
+      "loss": 1.0384,
+      "step": 8968
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011996071253180947,
+      "loss": 0.8616,
+      "step": 8969
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011995178356176615,
+      "loss": 1.0778,
+      "step": 8970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011994285459172285,
+      "loss": 1.0358,
+      "step": 8971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011993392562167956,
+      "loss": 0.991,
+      "step": 8972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011992499665163623,
+      "loss": 1.1065,
+      "step": 8973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011991606768159294,
+      "loss": 1.1033,
+      "step": 8974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011990713871154964,
+      "loss": 1.0464,
+      "step": 8975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011989820974150632,
+      "loss": 0.946,
+      "step": 8976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011988928077146302,
+      "loss": 1.0692,
+      "step": 8977
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011988035180141973,
+      "loss": 1.0567,
+      "step": 8978
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001198714228313764,
+      "loss": 0.9658,
+      "step": 8979
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011986249386133311,
+      "loss": 0.9437,
+      "step": 8980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011985356489128981,
+      "loss": 1.1017,
+      "step": 8981
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011984463592124649,
+      "loss": 1.058,
+      "step": 8982
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011983570695120319,
+      "loss": 0.9987,
+      "step": 8983
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011982677798115988,
+      "loss": 1.0715,
+      "step": 8984
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011981784901111657,
+      "loss": 0.8584,
+      "step": 8985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011980892004107326,
+      "loss": 1.047,
+      "step": 8986
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011979999107102997,
+      "loss": 1.0484,
+      "step": 8987
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011979106210098664,
+      "loss": 0.9529,
+      "step": 8988
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011978213313094335,
+      "loss": 1.0659,
+      "step": 8989
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011977320416090005,
+      "loss": 1.0112,
+      "step": 8990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011976427519085673,
+      "loss": 1.0009,
+      "step": 8991
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011975534622081343,
+      "loss": 1.1503,
+      "step": 8992
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011974641725077014,
+      "loss": 1.1149,
+      "step": 8993
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011973748828072681,
+      "loss": 0.9566,
+      "step": 8994
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011972855931068352,
+      "loss": 1.0184,
+      "step": 8995
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011971963034064022,
+      "loss": 1.0231,
+      "step": 8996
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001197107013705969,
+      "loss": 1.0709,
+      "step": 8997
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001197017724005536,
+      "loss": 1.0464,
+      "step": 8998
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00011969284343051031,
+      "loss": 1.0618,
+      "step": 8999
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011968391446046698,
+      "loss": 1.0071,
+      "step": 9000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011967498549042369,
+      "loss": 1.0095,
+      "step": 9001
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011966605652038039,
+      "loss": 0.9942,
+      "step": 9002
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011965712755033707,
+      "loss": 0.8843,
+      "step": 9003
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011964819858029377,
+      "loss": 0.9015,
+      "step": 9004
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011963926961025048,
+      "loss": 1.0209,
+      "step": 9005
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011963034064020715,
+      "loss": 1.0534,
+      "step": 9006
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011962141167016386,
+      "loss": 1.1207,
+      "step": 9007
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011961248270012055,
+      "loss": 1.1842,
+      "step": 9008
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011960355373007724,
+      "loss": 1.1073,
+      "step": 9009
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011959462476003393,
+      "loss": 1.1056,
+      "step": 9010
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011958569578999063,
+      "loss": 0.9849,
+      "step": 9011
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011957676681994731,
+      "loss": 1.1125,
+      "step": 9012
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011956783784990401,
+      "loss": 1.0635,
+      "step": 9013
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011955890887986072,
+      "loss": 1.0648,
+      "step": 9014
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001195499799098174,
+      "loss": 1.029,
+      "step": 9015
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001195410509397741,
+      "loss": 1.1016,
+      "step": 9016
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001195321219697308,
+      "loss": 1.0351,
+      "step": 9017
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011952319299968748,
+      "loss": 1.2698,
+      "step": 9018
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011951426402964418,
+      "loss": 1.0946,
+      "step": 9019
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011950533505960089,
+      "loss": 1.0685,
+      "step": 9020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011949640608955757,
+      "loss": 1.0559,
+      "step": 9021
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011948747711951427,
+      "loss": 1.0665,
+      "step": 9022
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011947854814947097,
+      "loss": 1.0539,
+      "step": 9023
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011946961917942765,
+      "loss": 1.0027,
+      "step": 9024
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011946069020938435,
+      "loss": 1.0098,
+      "step": 9025
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011945176123934106,
+      "loss": 0.9602,
+      "step": 9026
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011944283226929774,
+      "loss": 0.9757,
+      "step": 9027
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011943390329925444,
+      "loss": 1.002,
+      "step": 9028
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011942497432921113,
+      "loss": 1.0005,
+      "step": 9029
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011941604535916782,
+      "loss": 1.0721,
+      "step": 9030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011940711638912452,
+      "loss": 0.9732,
+      "step": 9031
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011939818741908122,
+      "loss": 1.137,
+      "step": 9032
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001193892584490379,
+      "loss": 1.1159,
+      "step": 9033
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001193803294789946,
+      "loss": 0.9862,
+      "step": 9034
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001193714005089513,
+      "loss": 0.9986,
+      "step": 9035
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011936247153890798,
+      "loss": 1.1259,
+      "step": 9036
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011935354256886468,
+      "loss": 1.0923,
+      "step": 9037
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011934461359882139,
+      "loss": 1.0079,
+      "step": 9038
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011933568462877806,
+      "loss": 0.9213,
+      "step": 9039
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011932675565873477,
+      "loss": 0.9733,
+      "step": 9040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011931782668869147,
+      "loss": 0.8846,
+      "step": 9041
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011930889771864815,
+      "loss": 1.0205,
+      "step": 9042
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011929996874860485,
+      "loss": 1.1536,
+      "step": 9043
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011929103977856156,
+      "loss": 0.9987,
+      "step": 9044
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011928211080851823,
+      "loss": 1.1571,
+      "step": 9045
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011927318183847494,
+      "loss": 0.9968,
+      "step": 9046
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011926425286843164,
+      "loss": 0.9889,
+      "step": 9047
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011925532389838832,
+      "loss": 1.0369,
+      "step": 9048
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011924639492834502,
+      "loss": 1.0469,
+      "step": 9049
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011923746595830173,
+      "loss": 1.0035,
+      "step": 9050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001192285369882584,
+      "loss": 1.0988,
+      "step": 9051
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001192196080182151,
+      "loss": 0.9576,
+      "step": 9052
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001192106790481718,
+      "loss": 1.0673,
+      "step": 9053
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001192017500781285,
+      "loss": 0.9514,
+      "step": 9054
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011919282110808518,
+      "loss": 1.0743,
+      "step": 9055
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011918389213804188,
+      "loss": 1.1189,
+      "step": 9056
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011917496316799859,
+      "loss": 1.0084,
+      "step": 9057
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011916603419795526,
+      "loss": 1.1515,
+      "step": 9058
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011915710522791197,
+      "loss": 1.0263,
+      "step": 9059
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011914817625786867,
+      "loss": 1.0835,
+      "step": 9060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011913924728782535,
+      "loss": 1.0474,
+      "step": 9061
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011913031831778205,
+      "loss": 0.9691,
+      "step": 9062
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011912138934773876,
+      "loss": 1.0168,
+      "step": 9063
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011911246037769543,
+      "loss": 1.0566,
+      "step": 9064
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011910353140765214,
+      "loss": 1.0033,
+      "step": 9065
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011909460243760884,
+      "loss": 1.0251,
+      "step": 9066
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011908567346756552,
+      "loss": 1.0843,
+      "step": 9067
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011907674449752222,
+      "loss": 0.9745,
+      "step": 9068
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011906781552747893,
+      "loss": 1.0202,
+      "step": 9069
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001190588865574356,
+      "loss": 1.0311,
+      "step": 9070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001190499575873923,
+      "loss": 0.9302,
+      "step": 9071
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011904102861734901,
+      "loss": 0.9548,
+      "step": 9072
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011903209964730569,
+      "loss": 1.0278,
+      "step": 9073
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00011902317067726239,
+      "loss": 0.9636,
+      "step": 9074
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011901424170721908,
+      "loss": 0.972,
+      "step": 9075
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011900531273717577,
+      "loss": 0.9644,
+      "step": 9076
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011899638376713246,
+      "loss": 1.1232,
+      "step": 9077
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011898745479708917,
+      "loss": 1.1317,
+      "step": 9078
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011897852582704584,
+      "loss": 0.9501,
+      "step": 9079
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011896959685700255,
+      "loss": 1.0537,
+      "step": 9080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011896066788695925,
+      "loss": 1.0196,
+      "step": 9081
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011895173891691593,
+      "loss": 1.0079,
+      "step": 9082
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011894280994687263,
+      "loss": 1.0155,
+      "step": 9083
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011893388097682934,
+      "loss": 1.3513,
+      "step": 9084
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011892495200678601,
+      "loss": 1.1558,
+      "step": 9085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011891602303674272,
+      "loss": 1.0939,
+      "step": 9086
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011890709406669942,
+      "loss": 0.9528,
+      "step": 9087
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001188981650966561,
+      "loss": 1.0836,
+      "step": 9088
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001188892361266128,
+      "loss": 1.019,
+      "step": 9089
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011888030715656951,
+      "loss": 1.0089,
+      "step": 9090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011887137818652618,
+      "loss": 1.0444,
+      "step": 9091
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011886244921648289,
+      "loss": 0.969,
+      "step": 9092
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011885352024643959,
+      "loss": 1.0501,
+      "step": 9093
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011884459127639627,
+      "loss": 0.8839,
+      "step": 9094
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011883566230635297,
+      "loss": 1.1315,
+      "step": 9095
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011882673333630968,
+      "loss": 0.9231,
+      "step": 9096
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011881780436626635,
+      "loss": 1.0443,
+      "step": 9097
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011880887539622306,
+      "loss": 1.197,
+      "step": 9098
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011879994642617975,
+      "loss": 1.0846,
+      "step": 9099
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011879101745613644,
+      "loss": 1.0378,
+      "step": 9100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011878208848609313,
+      "loss": 1.0228,
+      "step": 9101
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011877315951604983,
+      "loss": 1.016,
+      "step": 9102
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011876423054600651,
+      "loss": 1.1902,
+      "step": 9103
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011875530157596321,
+      "loss": 0.971,
+      "step": 9104
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011874637260591992,
+      "loss": 0.9409,
+      "step": 9105
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001187374436358766,
+      "loss": 1.0723,
+      "step": 9106
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001187285146658333,
+      "loss": 1.0895,
+      "step": 9107
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011871958569579,
+      "loss": 1.0796,
+      "step": 9108
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011871065672574668,
+      "loss": 0.9771,
+      "step": 9109
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011870172775570338,
+      "loss": 1.0893,
+      "step": 9110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011869279878566009,
+      "loss": 1.0037,
+      "step": 9111
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011868386981561677,
+      "loss": 0.9734,
+      "step": 9112
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011867494084557347,
+      "loss": 1.0137,
+      "step": 9113
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011866601187553017,
+      "loss": 1.0039,
+      "step": 9114
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011865708290548685,
+      "loss": 1.0429,
+      "step": 9115
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011864815393544355,
+      "loss": 0.9766,
+      "step": 9116
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011863922496540026,
+      "loss": 0.9297,
+      "step": 9117
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011863029599535694,
+      "loss": 1.1608,
+      "step": 9118
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011862136702531364,
+      "loss": 0.9752,
+      "step": 9119
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011861243805527033,
+      "loss": 1.0532,
+      "step": 9120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011860350908522702,
+      "loss": 0.9985,
+      "step": 9121
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011859458011518372,
+      "loss": 1.0516,
+      "step": 9122
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011858565114514041,
+      "loss": 0.9845,
+      "step": 9123
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001185767221750971,
+      "loss": 0.937,
+      "step": 9124
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001185677932050538,
+      "loss": 1.0619,
+      "step": 9125
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001185588642350105,
+      "loss": 1.0682,
+      "step": 9126
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011854993526496718,
+      "loss": 1.158,
+      "step": 9127
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011854100629492388,
+      "loss": 0.9647,
+      "step": 9128
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011853207732488058,
+      "loss": 0.9823,
+      "step": 9129
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011852314835483726,
+      "loss": 1.125,
+      "step": 9130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011851421938479397,
+      "loss": 0.9699,
+      "step": 9131
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011850529041475067,
+      "loss": 1.0432,
+      "step": 9132
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011849636144470735,
+      "loss": 1.1317,
+      "step": 9133
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011848743247466405,
+      "loss": 1.0562,
+      "step": 9134
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011847850350462075,
+      "loss": 1.0361,
+      "step": 9135
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011846957453457743,
+      "loss": 0.9685,
+      "step": 9136
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011846064556453414,
+      "loss": 0.9997,
+      "step": 9137
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011845171659449084,
+      "loss": 1.0465,
+      "step": 9138
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011844278762444752,
+      "loss": 1.113,
+      "step": 9139
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011843385865440422,
+      "loss": 1.0906,
+      "step": 9140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011842492968436092,
+      "loss": 1.0543,
+      "step": 9141
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001184160007143176,
+      "loss": 0.9923,
+      "step": 9142
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001184070717442743,
+      "loss": 0.9931,
+      "step": 9143
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000118398142774231,
+      "loss": 1.0624,
+      "step": 9144
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011838921380418769,
+      "loss": 1.1436,
+      "step": 9145
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011838028483414438,
+      "loss": 1.002,
+      "step": 9146
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011837135586410108,
+      "loss": 0.9989,
+      "step": 9147
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00011836242689405777,
+      "loss": 1.0068,
+      "step": 9148
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011835349792401446,
+      "loss": 1.0108,
+      "step": 9149
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011834456895397117,
+      "loss": 1.1341,
+      "step": 9150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011833563998392787,
+      "loss": 0.9385,
+      "step": 9151
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011832671101388455,
+      "loss": 1.1736,
+      "step": 9152
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011831778204384125,
+      "loss": 0.9449,
+      "step": 9153
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011830885307379796,
+      "loss": 0.9272,
+      "step": 9154
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011829992410375463,
+      "loss": 1.1267,
+      "step": 9155
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011829099513371134,
+      "loss": 1.0304,
+      "step": 9156
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011828206616366804,
+      "loss": 1.0889,
+      "step": 9157
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011827313719362472,
+      "loss": 0.9503,
+      "step": 9158
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011826420822358142,
+      "loss": 0.9082,
+      "step": 9159
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011825527925353813,
+      "loss": 0.9459,
+      "step": 9160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001182463502834948,
+      "loss": 0.9584,
+      "step": 9161
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001182374213134515,
+      "loss": 1.1107,
+      "step": 9162
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011822849234340821,
+      "loss": 0.8995,
+      "step": 9163
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011821956337336489,
+      "loss": 1.1131,
+      "step": 9164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011821063440332159,
+      "loss": 0.9871,
+      "step": 9165
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011820170543327828,
+      "loss": 1.02,
+      "step": 9166
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011819277646323497,
+      "loss": 0.9942,
+      "step": 9167
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011818384749319166,
+      "loss": 0.9745,
+      "step": 9168
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011817491852314837,
+      "loss": 1.1165,
+      "step": 9169
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011816598955310504,
+      "loss": 0.9894,
+      "step": 9170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011815706058306175,
+      "loss": 1.1083,
+      "step": 9171
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011814813161301845,
+      "loss": 0.9995,
+      "step": 9172
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011813920264297513,
+      "loss": 0.9916,
+      "step": 9173
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011813027367293183,
+      "loss": 1.0598,
+      "step": 9174
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011812134470288854,
+      "loss": 0.9583,
+      "step": 9175
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011811241573284521,
+      "loss": 1.0207,
+      "step": 9176
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011810348676280192,
+      "loss": 1.1081,
+      "step": 9177
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011809455779275862,
+      "loss": 1.0141,
+      "step": 9178
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001180856288227153,
+      "loss": 1.048,
+      "step": 9179
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000118076699852672,
+      "loss": 1.0731,
+      "step": 9180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001180677708826287,
+      "loss": 1.0721,
+      "step": 9181
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011805884191258538,
+      "loss": 0.8947,
+      "step": 9182
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011804991294254209,
+      "loss": 1.0443,
+      "step": 9183
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011804098397249879,
+      "loss": 0.9814,
+      "step": 9184
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011803205500245547,
+      "loss": 1.1104,
+      "step": 9185
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011802312603241217,
+      "loss": 1.1548,
+      "step": 9186
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011801419706236886,
+      "loss": 1.0004,
+      "step": 9187
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011800526809232555,
+      "loss": 0.9948,
+      "step": 9188
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011799633912228226,
+      "loss": 1.0214,
+      "step": 9189
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011798741015223895,
+      "loss": 0.9555,
+      "step": 9190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011797848118219564,
+      "loss": 1.0145,
+      "step": 9191
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011796955221215233,
+      "loss": 1.0966,
+      "step": 9192
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011796062324210903,
+      "loss": 1.0747,
+      "step": 9193
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011795169427206571,
+      "loss": 1.1094,
+      "step": 9194
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011794276530202241,
+      "loss": 1.0086,
+      "step": 9195
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011793383633197912,
+      "loss": 1.0742,
+      "step": 9196
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001179249073619358,
+      "loss": 1.02,
+      "step": 9197
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001179159783918925,
+      "loss": 1.0459,
+      "step": 9198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001179070494218492,
+      "loss": 0.9828,
+      "step": 9199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011789812045180588,
+      "loss": 1.0643,
+      "step": 9200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011788919148176258,
+      "loss": 1.019,
+      "step": 9201
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011788026251171929,
+      "loss": 1.1081,
+      "step": 9202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011787133354167596,
+      "loss": 1.0646,
+      "step": 9203
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011786240457163267,
+      "loss": 1.1289,
+      "step": 9204
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011785347560158937,
+      "loss": 1.0464,
+      "step": 9205
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011784454663154605,
+      "loss": 1.2411,
+      "step": 9206
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011783561766150275,
+      "loss": 0.9904,
+      "step": 9207
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011782668869145946,
+      "loss": 0.9926,
+      "step": 9208
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011781775972141613,
+      "loss": 1.1608,
+      "step": 9209
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011780883075137284,
+      "loss": 0.9816,
+      "step": 9210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011779990178132953,
+      "loss": 1.0629,
+      "step": 9211
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011779097281128622,
+      "loss": 1.0123,
+      "step": 9212
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011778204384124292,
+      "loss": 1.0893,
+      "step": 9213
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011777311487119961,
+      "loss": 1.1797,
+      "step": 9214
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001177641859011563,
+      "loss": 0.9734,
+      "step": 9215
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000117755256931113,
+      "loss": 1.095,
+      "step": 9216
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001177463279610697,
+      "loss": 1.0623,
+      "step": 9217
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011773739899102638,
+      "loss": 1.0792,
+      "step": 9218
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011772847002098308,
+      "loss": 1.0354,
+      "step": 9219
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011771954105093978,
+      "loss": 0.9965,
+      "step": 9220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011771061208089646,
+      "loss": 1.2107,
+      "step": 9221
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011770168311085317,
+      "loss": 1.0138,
+      "step": 9222
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00011769275414080987,
+      "loss": 1.0527,
+      "step": 9223
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011768382517076655,
+      "loss": 1.1418,
+      "step": 9224
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011767489620072325,
+      "loss": 0.7557,
+      "step": 9225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011766596723067995,
+      "loss": 0.9793,
+      "step": 9226
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011765703826063663,
+      "loss": 0.9203,
+      "step": 9227
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011764810929059334,
+      "loss": 1.0763,
+      "step": 9228
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011763918032055004,
+      "loss": 1.1038,
+      "step": 9229
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011763025135050672,
+      "loss": 1.0245,
+      "step": 9230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011762132238046342,
+      "loss": 1.1155,
+      "step": 9231
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011761239341042012,
+      "loss": 1.1117,
+      "step": 9232
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001176034644403768,
+      "loss": 1.0522,
+      "step": 9233
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001175945354703335,
+      "loss": 1.0712,
+      "step": 9234
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001175856065002902,
+      "loss": 1.1774,
+      "step": 9235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011757667753024689,
+      "loss": 1.0558,
+      "step": 9236
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011756774856020358,
+      "loss": 1.0946,
+      "step": 9237
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011755881959016028,
+      "loss": 1.0114,
+      "step": 9238
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011754989062011697,
+      "loss": 1.1081,
+      "step": 9239
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011754096165007366,
+      "loss": 1.0157,
+      "step": 9240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011753203268003037,
+      "loss": 0.9898,
+      "step": 9241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011752310370998704,
+      "loss": 1.0011,
+      "step": 9242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011751417473994375,
+      "loss": 0.9657,
+      "step": 9243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011750524576990045,
+      "loss": 0.9588,
+      "step": 9244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011749631679985713,
+      "loss": 1.1334,
+      "step": 9245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011748738782981383,
+      "loss": 1.0772,
+      "step": 9246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011747845885977054,
+      "loss": 0.9501,
+      "step": 9247
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011746952988972721,
+      "loss": 0.9685,
+      "step": 9248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011746060091968392,
+      "loss": 0.9355,
+      "step": 9249
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011745167194964062,
+      "loss": 1.0619,
+      "step": 9250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011744274297959732,
+      "loss": 0.9403,
+      "step": 9251
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000117433814009554,
+      "loss": 1.0384,
+      "step": 9252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001174248850395107,
+      "loss": 1.0016,
+      "step": 9253
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011741595606946741,
+      "loss": 1.0909,
+      "step": 9254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011740702709942409,
+      "loss": 1.1867,
+      "step": 9255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011739809812938079,
+      "loss": 0.9514,
+      "step": 9256
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011738916915933748,
+      "loss": 1.0556,
+      "step": 9257
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011738024018929417,
+      "loss": 0.972,
+      "step": 9258
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011737131121925086,
+      "loss": 1.0344,
+      "step": 9259
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011736238224920757,
+      "loss": 1.022,
+      "step": 9260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011735345327916424,
+      "loss": 1.1079,
+      "step": 9261
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011734452430912095,
+      "loss": 1.0825,
+      "step": 9262
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011733559533907765,
+      "loss": 1.1156,
+      "step": 9263
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011732666636903433,
+      "loss": 0.9765,
+      "step": 9264
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011731773739899103,
+      "loss": 0.9727,
+      "step": 9265
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011730880842894774,
+      "loss": 1.0911,
+      "step": 9266
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011729987945890441,
+      "loss": 1.1568,
+      "step": 9267
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011729095048886112,
+      "loss": 0.9425,
+      "step": 9268
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011728202151881782,
+      "loss": 1.035,
+      "step": 9269
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001172730925487745,
+      "loss": 1.0548,
+      "step": 9270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001172641635787312,
+      "loss": 0.9498,
+      "step": 9271
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001172552346086879,
+      "loss": 1.1724,
+      "step": 9272
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011724630563864458,
+      "loss": 0.9786,
+      "step": 9273
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011723737666860129,
+      "loss": 1.0234,
+      "step": 9274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011722844769855799,
+      "loss": 1.2156,
+      "step": 9275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011721951872851467,
+      "loss": 1.2018,
+      "step": 9276
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011721058975847137,
+      "loss": 1.018,
+      "step": 9277
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011720166078842806,
+      "loss": 1.0412,
+      "step": 9278
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011719273181838475,
+      "loss": 0.9723,
+      "step": 9279
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011718380284834146,
+      "loss": 1.0388,
+      "step": 9280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011717487387829815,
+      "loss": 1.0185,
+      "step": 9281
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011716594490825484,
+      "loss": 1.0501,
+      "step": 9282
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011715701593821153,
+      "loss": 0.9794,
+      "step": 9283
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011714808696816823,
+      "loss": 1.0442,
+      "step": 9284
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011713915799812491,
+      "loss": 1.093,
+      "step": 9285
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011713022902808161,
+      "loss": 1.028,
+      "step": 9286
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011712130005803832,
+      "loss": 0.9642,
+      "step": 9287
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000117112371087995,
+      "loss": 1.0656,
+      "step": 9288
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001171034421179517,
+      "loss": 0.951,
+      "step": 9289
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001170945131479084,
+      "loss": 1.0601,
+      "step": 9290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011708558417786508,
+      "loss": 0.9575,
+      "step": 9291
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011707665520782178,
+      "loss": 0.9963,
+      "step": 9292
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011706772623777849,
+      "loss": 0.9478,
+      "step": 9293
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011705879726773516,
+      "loss": 1.1937,
+      "step": 9294
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011704986829769187,
+      "loss": 1.0025,
+      "step": 9295
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011704093932764857,
+      "loss": 0.9939,
+      "step": 9296
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011703201035760525,
+      "loss": 0.9167,
+      "step": 9297
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00011702308138756195,
+      "loss": 0.9886,
+      "step": 9298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011701415241751866,
+      "loss": 1.0641,
+      "step": 9299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011700522344747533,
+      "loss": 0.911,
+      "step": 9300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011699629447743204,
+      "loss": 1.1006,
+      "step": 9301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011698736550738873,
+      "loss": 0.9932,
+      "step": 9302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011697843653734542,
+      "loss": 1.2174,
+      "step": 9303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011696950756730211,
+      "loss": 1.1142,
+      "step": 9304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011696057859725881,
+      "loss": 0.9687,
+      "step": 9305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001169516496272155,
+      "loss": 1.0163,
+      "step": 9306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001169427206571722,
+      "loss": 1.0172,
+      "step": 9307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001169337916871289,
+      "loss": 0.9661,
+      "step": 9308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011692486271708558,
+      "loss": 1.0797,
+      "step": 9309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011691593374704228,
+      "loss": 1.0168,
+      "step": 9310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011690700477699898,
+      "loss": 1.012,
+      "step": 9311
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011689807580695566,
+      "loss": 0.9849,
+      "step": 9312
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011688914683691237,
+      "loss": 1.0147,
+      "step": 9313
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011688021786686907,
+      "loss": 1.1168,
+      "step": 9314
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011687128889682575,
+      "loss": 1.0938,
+      "step": 9315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011686235992678245,
+      "loss": 0.997,
+      "step": 9316
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011685343095673915,
+      "loss": 0.9714,
+      "step": 9317
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011684450198669583,
+      "loss": 1.0007,
+      "step": 9318
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011683557301665254,
+      "loss": 1.0203,
+      "step": 9319
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011682664404660924,
+      "loss": 1.1211,
+      "step": 9320
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011681771507656592,
+      "loss": 1.2091,
+      "step": 9321
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011680878610652262,
+      "loss": 0.995,
+      "step": 9322
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011679985713647932,
+      "loss": 1.0433,
+      "step": 9323
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000116790928166436,
+      "loss": 1.0798,
+      "step": 9324
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001167819991963927,
+      "loss": 1.293,
+      "step": 9325
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001167730702263494,
+      "loss": 1.1041,
+      "step": 9326
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011676414125630609,
+      "loss": 1.026,
+      "step": 9327
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011675521228626278,
+      "loss": 1.0654,
+      "step": 9328
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011674628331621948,
+      "loss": 0.9632,
+      "step": 9329
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011673735434617617,
+      "loss": 1.0183,
+      "step": 9330
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011672842537613286,
+      "loss": 1.059,
+      "step": 9331
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011671949640608957,
+      "loss": 1.0166,
+      "step": 9332
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011671056743604624,
+      "loss": 1.08,
+      "step": 9333
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011670163846600295,
+      "loss": 1.0322,
+      "step": 9334
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011669270949595965,
+      "loss": 0.9183,
+      "step": 9335
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011668378052591633,
+      "loss": 1.1042,
+      "step": 9336
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011667485155587303,
+      "loss": 0.9574,
+      "step": 9337
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011666592258582974,
+      "loss": 0.9292,
+      "step": 9338
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011665699361578641,
+      "loss": 0.9416,
+      "step": 9339
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011664806464574312,
+      "loss": 1.0914,
+      "step": 9340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011663913567569982,
+      "loss": 1.032,
+      "step": 9341
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001166302067056565,
+      "loss": 0.9487,
+      "step": 9342
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001166212777356132,
+      "loss": 1.0058,
+      "step": 9343
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001166123487655699,
+      "loss": 1.057,
+      "step": 9344
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011660341979552658,
+      "loss": 1.0235,
+      "step": 9345
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011659449082548329,
+      "loss": 1.0225,
+      "step": 9346
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011658556185543999,
+      "loss": 1.1335,
+      "step": 9347
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011657663288539668,
+      "loss": 1.1383,
+      "step": 9348
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011656770391535337,
+      "loss": 0.9518,
+      "step": 9349
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011655877494531006,
+      "loss": 1.0068,
+      "step": 9350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011654984597526677,
+      "loss": 1.0159,
+      "step": 9351
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011654091700522344,
+      "loss": 1.0032,
+      "step": 9352
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011653198803518015,
+      "loss": 1.0317,
+      "step": 9353
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011652305906513685,
+      "loss": 1.0237,
+      "step": 9354
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011651413009509353,
+      "loss": 1.017,
+      "step": 9355
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011650520112505023,
+      "loss": 0.9814,
+      "step": 9356
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011649627215500694,
+      "loss": 0.9823,
+      "step": 9357
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011648734318496361,
+      "loss": 0.97,
+      "step": 9358
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011647841421492032,
+      "loss": 0.9253,
+      "step": 9359
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011646948524487702,
+      "loss": 1.0414,
+      "step": 9360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001164605562748337,
+      "loss": 1.0001,
+      "step": 9361
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001164516273047904,
+      "loss": 1.0061,
+      "step": 9362
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001164426983347471,
+      "loss": 1.0293,
+      "step": 9363
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011643376936470378,
+      "loss": 1.1157,
+      "step": 9364
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011642484039466049,
+      "loss": 1.0843,
+      "step": 9365
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011641591142461719,
+      "loss": 1.0157,
+      "step": 9366
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011640698245457387,
+      "loss": 0.9387,
+      "step": 9367
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011639805348453057,
+      "loss": 1.2052,
+      "step": 9368
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011638912451448726,
+      "loss": 0.9275,
+      "step": 9369
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011638019554444395,
+      "loss": 1.1211,
+      "step": 9370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011637126657440066,
+      "loss": 0.9815,
+      "step": 9371
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00011636233760435735,
+      "loss": 1.0121,
+      "step": 9372
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011635340863431404,
+      "loss": 1.007,
+      "step": 9373
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011634447966427073,
+      "loss": 1.0622,
+      "step": 9374
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011633555069422743,
+      "loss": 0.9556,
+      "step": 9375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011632662172418411,
+      "loss": 1.0163,
+      "step": 9376
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011631769275414081,
+      "loss": 1.0368,
+      "step": 9377
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011630876378409752,
+      "loss": 0.9763,
+      "step": 9378
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001162998348140542,
+      "loss": 1.1034,
+      "step": 9379
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001162909058440109,
+      "loss": 0.9275,
+      "step": 9380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001162819768739676,
+      "loss": 1.0986,
+      "step": 9381
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011627304790392428,
+      "loss": 0.9157,
+      "step": 9382
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011626411893388098,
+      "loss": 0.9743,
+      "step": 9383
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011625518996383769,
+      "loss": 1.0005,
+      "step": 9384
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011624626099379436,
+      "loss": 0.969,
+      "step": 9385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011623733202375107,
+      "loss": 1.0906,
+      "step": 9386
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011622840305370777,
+      "loss": 1.0388,
+      "step": 9387
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011621947408366445,
+      "loss": 1.0836,
+      "step": 9388
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011621054511362115,
+      "loss": 1.1163,
+      "step": 9389
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011620161614357786,
+      "loss": 0.9712,
+      "step": 9390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011619268717353453,
+      "loss": 1.0686,
+      "step": 9391
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011618375820349124,
+      "loss": 1.0344,
+      "step": 9392
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011617482923344793,
+      "loss": 0.8498,
+      "step": 9393
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011616590026340462,
+      "loss": 0.9877,
+      "step": 9394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011615697129336131,
+      "loss": 0.9555,
+      "step": 9395
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011614804232331801,
+      "loss": 1.041,
+      "step": 9396
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001161391133532747,
+      "loss": 0.9949,
+      "step": 9397
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001161301843832314,
+      "loss": 0.8968,
+      "step": 9398
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001161212554131881,
+      "loss": 1.1025,
+      "step": 9399
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011611232644314478,
+      "loss": 0.9886,
+      "step": 9400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011610339747310148,
+      "loss": 0.9244,
+      "step": 9401
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011609446850305818,
+      "loss": 0.9511,
+      "step": 9402
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011608553953301486,
+      "loss": 0.9556,
+      "step": 9403
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011607661056297156,
+      "loss": 0.9924,
+      "step": 9404
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011606768159292827,
+      "loss": 1.0495,
+      "step": 9405
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011605875262288495,
+      "loss": 1.0641,
+      "step": 9406
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011604982365284165,
+      "loss": 1.0498,
+      "step": 9407
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011604089468279835,
+      "loss": 1.0638,
+      "step": 9408
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011603196571275503,
+      "loss": 1.0229,
+      "step": 9409
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011602303674271173,
+      "loss": 1.0422,
+      "step": 9410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011601410777266844,
+      "loss": 1.1226,
+      "step": 9411
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011600517880262512,
+      "loss": 0.9757,
+      "step": 9412
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011599624983258182,
+      "loss": 1.0147,
+      "step": 9413
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011598732086253852,
+      "loss": 0.9348,
+      "step": 9414
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001159783918924952,
+      "loss": 1.1125,
+      "step": 9415
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001159694629224519,
+      "loss": 1.0597,
+      "step": 9416
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001159605339524086,
+      "loss": 1.0918,
+      "step": 9417
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011595160498236529,
+      "loss": 1.0219,
+      "step": 9418
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011594267601232198,
+      "loss": 1.0569,
+      "step": 9419
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011593374704227868,
+      "loss": 1.0218,
+      "step": 9420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011592481807223537,
+      "loss": 1.0444,
+      "step": 9421
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011591588910219206,
+      "loss": 1.1012,
+      "step": 9422
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011590696013214877,
+      "loss": 0.977,
+      "step": 9423
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011589803116210544,
+      "loss": 1.087,
+      "step": 9424
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011588910219206215,
+      "loss": 1.0518,
+      "step": 9425
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011588017322201885,
+      "loss": 0.9903,
+      "step": 9426
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011587124425197553,
+      "loss": 1.0485,
+      "step": 9427
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011586231528193223,
+      "loss": 1.0738,
+      "step": 9428
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011585338631188894,
+      "loss": 1.153,
+      "step": 9429
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011584445734184561,
+      "loss": 0.959,
+      "step": 9430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011583552837180232,
+      "loss": 1.0563,
+      "step": 9431
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011582659940175902,
+      "loss": 1.0672,
+      "step": 9432
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001158176704317157,
+      "loss": 1.1026,
+      "step": 9433
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001158087414616724,
+      "loss": 1.0998,
+      "step": 9434
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001157998124916291,
+      "loss": 1.0919,
+      "step": 9435
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011579088352158578,
+      "loss": 0.9956,
+      "step": 9436
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011578195455154249,
+      "loss": 1.0517,
+      "step": 9437
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011577302558149919,
+      "loss": 1.0487,
+      "step": 9438
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011576409661145587,
+      "loss": 0.9707,
+      "step": 9439
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011575516764141257,
+      "loss": 1.0425,
+      "step": 9440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011574623867136926,
+      "loss": 1.0118,
+      "step": 9441
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011573730970132595,
+      "loss": 1.21,
+      "step": 9442
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011572838073128264,
+      "loss": 0.8404,
+      "step": 9443
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011571945176123935,
+      "loss": 1.1039,
+      "step": 9444
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011571052279119605,
+      "loss": 1.0505,
+      "step": 9445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011570159382115273,
+      "loss": 1.0539,
+      "step": 9446
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00011569266485110943,
+      "loss": 0.923,
+      "step": 9447
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011568373588106614,
+      "loss": 1.0641,
+      "step": 9448
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011567480691102281,
+      "loss": 1.0557,
+      "step": 9449
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011566587794097952,
+      "loss": 1.0423,
+      "step": 9450
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011565694897093622,
+      "loss": 0.9778,
+      "step": 9451
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001156480200008929,
+      "loss": 0.9309,
+      "step": 9452
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001156390910308496,
+      "loss": 1.0291,
+      "step": 9453
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001156301620608063,
+      "loss": 0.9856,
+      "step": 9454
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011562123309076298,
+      "loss": 0.9518,
+      "step": 9455
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011561230412071969,
+      "loss": 1.071,
+      "step": 9456
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011560337515067639,
+      "loss": 0.9961,
+      "step": 9457
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011559444618063307,
+      "loss": 0.9164,
+      "step": 9458
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011558551721058977,
+      "loss": 1.0417,
+      "step": 9459
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011557658824054646,
+      "loss": 0.8451,
+      "step": 9460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011556765927050315,
+      "loss": 1.1682,
+      "step": 9461
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011555873030045986,
+      "loss": 1.073,
+      "step": 9462
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011554980133041655,
+      "loss": 0.9632,
+      "step": 9463
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011554087236037324,
+      "loss": 0.9963,
+      "step": 9464
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011553194339032993,
+      "loss": 0.9445,
+      "step": 9465
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011552301442028663,
+      "loss": 1.0582,
+      "step": 9466
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011551408545024331,
+      "loss": 1.0804,
+      "step": 9467
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011550515648020001,
+      "loss": 1.0322,
+      "step": 9468
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011549622751015672,
+      "loss": 0.9474,
+      "step": 9469
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001154872985401134,
+      "loss": 0.9731,
+      "step": 9470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001154783695700701,
+      "loss": 0.9846,
+      "step": 9471
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001154694406000268,
+      "loss": 1.1578,
+      "step": 9472
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011546051162998348,
+      "loss": 1.0485,
+      "step": 9473
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011545158265994018,
+      "loss": 1.0159,
+      "step": 9474
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011544265368989689,
+      "loss": 1.0277,
+      "step": 9475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011543372471985356,
+      "loss": 1.0626,
+      "step": 9476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011542479574981027,
+      "loss": 1.0321,
+      "step": 9477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011541586677976697,
+      "loss": 0.9556,
+      "step": 9478
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011540693780972365,
+      "loss": 0.9827,
+      "step": 9479
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011539800883968035,
+      "loss": 1.0375,
+      "step": 9480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011538907986963706,
+      "loss": 0.9706,
+      "step": 9481
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011538015089959373,
+      "loss": 1.0468,
+      "step": 9482
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011537122192955044,
+      "loss": 1.0691,
+      "step": 9483
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011536229295950713,
+      "loss": 1.0165,
+      "step": 9484
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011535336398946382,
+      "loss": 1.128,
+      "step": 9485
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011534443501942051,
+      "loss": 1.0292,
+      "step": 9486
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011533550604937721,
+      "loss": 1.0508,
+      "step": 9487
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001153265770793339,
+      "loss": 1.0743,
+      "step": 9488
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001153176481092906,
+      "loss": 0.9837,
+      "step": 9489
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001153087191392473,
+      "loss": 0.968,
+      "step": 9490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011529979016920398,
+      "loss": 1.0359,
+      "step": 9491
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011529086119916068,
+      "loss": 1.112,
+      "step": 9492
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011528193222911738,
+      "loss": 0.9908,
+      "step": 9493
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011527300325907406,
+      "loss": 0.9293,
+      "step": 9494
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011526407428903076,
+      "loss": 1.0904,
+      "step": 9495
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011525514531898747,
+      "loss": 0.9501,
+      "step": 9496
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011524621634894415,
+      "loss": 1.0504,
+      "step": 9497
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011523728737890085,
+      "loss": 0.9462,
+      "step": 9498
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011522835840885755,
+      "loss": 1.032,
+      "step": 9499
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011521942943881423,
+      "loss": 0.9946,
+      "step": 9500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011521050046877093,
+      "loss": 0.9995,
+      "step": 9501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011520157149872764,
+      "loss": 0.9799,
+      "step": 9502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011519264252868432,
+      "loss": 1.1611,
+      "step": 9503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011518371355864102,
+      "loss": 0.8501,
+      "step": 9504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011517478458859772,
+      "loss": 0.9929,
+      "step": 9505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001151658556185544,
+      "loss": 1.0573,
+      "step": 9506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001151569266485111,
+      "loss": 1.0131,
+      "step": 9507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001151479976784678,
+      "loss": 1.0314,
+      "step": 9508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011513906870842449,
+      "loss": 1.0933,
+      "step": 9509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011513013973838118,
+      "loss": 1.1658,
+      "step": 9510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011512121076833788,
+      "loss": 1.0604,
+      "step": 9511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011511228179829456,
+      "loss": 0.9556,
+      "step": 9512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011510335282825126,
+      "loss": 0.9786,
+      "step": 9513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011509442385820796,
+      "loss": 1.1207,
+      "step": 9514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011508549488816464,
+      "loss": 0.9456,
+      "step": 9515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011507656591812135,
+      "loss": 0.9318,
+      "step": 9516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011506763694807805,
+      "loss": 1.0714,
+      "step": 9517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011505870797803473,
+      "loss": 1.1672,
+      "step": 9518
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011504977900799143,
+      "loss": 0.9404,
+      "step": 9519
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011504085003794813,
+      "loss": 1.0954,
+      "step": 9520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011503192106790481,
+      "loss": 1.0362,
+      "step": 9521
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00011502299209786152,
+      "loss": 1.0648,
+      "step": 9522
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011501406312781822,
+      "loss": 1.026,
+      "step": 9523
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001150051341577749,
+      "loss": 1.0806,
+      "step": 9524
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001149962051877316,
+      "loss": 1.0949,
+      "step": 9525
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001149872762176883,
+      "loss": 0.9782,
+      "step": 9526
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011497834724764498,
+      "loss": 0.9108,
+      "step": 9527
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011496941827760169,
+      "loss": 0.9834,
+      "step": 9528
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011496048930755839,
+      "loss": 1.052,
+      "step": 9529
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011495156033751507,
+      "loss": 0.9402,
+      "step": 9530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011494263136747177,
+      "loss": 1.2085,
+      "step": 9531
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011493370239742846,
+      "loss": 0.933,
+      "step": 9532
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011492477342738515,
+      "loss": 1.0957,
+      "step": 9533
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011491584445734184,
+      "loss": 1.1266,
+      "step": 9534
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011490691548729855,
+      "loss": 0.9477,
+      "step": 9535
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011489798651725522,
+      "loss": 0.9649,
+      "step": 9536
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011488905754721193,
+      "loss": 1.1015,
+      "step": 9537
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011488012857716863,
+      "loss": 1.1349,
+      "step": 9538
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011487119960712531,
+      "loss": 1.0444,
+      "step": 9539
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011486227063708201,
+      "loss": 1.0079,
+      "step": 9540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011485334166703872,
+      "loss": 0.9266,
+      "step": 9541
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011484441269699542,
+      "loss": 0.9244,
+      "step": 9542
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001148354837269521,
+      "loss": 1.3133,
+      "step": 9543
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001148265547569088,
+      "loss": 0.9852,
+      "step": 9544
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001148176257868655,
+      "loss": 1.0184,
+      "step": 9545
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011480869681682218,
+      "loss": 1.1799,
+      "step": 9546
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011479976784677889,
+      "loss": 0.9733,
+      "step": 9547
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011479083887673559,
+      "loss": 1.0199,
+      "step": 9548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011478190990669227,
+      "loss": 1.0334,
+      "step": 9549
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011477298093664897,
+      "loss": 1.0383,
+      "step": 9550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011476405196660566,
+      "loss": 1.1176,
+      "step": 9551
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011475512299656235,
+      "loss": 1.0081,
+      "step": 9552
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011474619402651906,
+      "loss": 1.0272,
+      "step": 9553
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011473726505647575,
+      "loss": 0.947,
+      "step": 9554
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011472833608643244,
+      "loss": 1.0132,
+      "step": 9555
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011471940711638913,
+      "loss": 1.0227,
+      "step": 9556
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011471047814634583,
+      "loss": 0.9696,
+      "step": 9557
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011470154917630251,
+      "loss": 1.1638,
+      "step": 9558
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011469262020625921,
+      "loss": 0.9343,
+      "step": 9559
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011468369123621592,
+      "loss": 1.0533,
+      "step": 9560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001146747622661726,
+      "loss": 0.9989,
+      "step": 9561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001146658332961293,
+      "loss": 1.0516,
+      "step": 9562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000114656904326086,
+      "loss": 1.092,
+      "step": 9563
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011464797535604268,
+      "loss": 1.1095,
+      "step": 9564
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011463904638599938,
+      "loss": 0.8254,
+      "step": 9565
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011463011741595609,
+      "loss": 1.0848,
+      "step": 9566
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011462118844591276,
+      "loss": 1.0818,
+      "step": 9567
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011461225947586947,
+      "loss": 1.0726,
+      "step": 9568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011460333050582617,
+      "loss": 1.0824,
+      "step": 9569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011459440153578285,
+      "loss": 1.072,
+      "step": 9570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011458547256573955,
+      "loss": 1.0623,
+      "step": 9571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011457654359569626,
+      "loss": 1.0306,
+      "step": 9572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011456761462565293,
+      "loss": 0.9658,
+      "step": 9573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011455868565560964,
+      "loss": 0.9819,
+      "step": 9574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011454975668556633,
+      "loss": 0.9635,
+      "step": 9575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011454082771552302,
+      "loss": 1.0606,
+      "step": 9576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011453189874547971,
+      "loss": 1.0881,
+      "step": 9577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011452296977543641,
+      "loss": 0.9754,
+      "step": 9578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001145140408053931,
+      "loss": 0.9466,
+      "step": 9579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001145051118353498,
+      "loss": 1.023,
+      "step": 9580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001144961828653065,
+      "loss": 1.0157,
+      "step": 9581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011448725389526317,
+      "loss": 1.018,
+      "step": 9582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011447832492521988,
+      "loss": 1.169,
+      "step": 9583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011446939595517658,
+      "loss": 0.9852,
+      "step": 9584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011446046698513326,
+      "loss": 0.9628,
+      "step": 9585
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011445153801508996,
+      "loss": 1.0172,
+      "step": 9586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011444260904504667,
+      "loss": 1.0001,
+      "step": 9587
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011443368007500334,
+      "loss": 0.9412,
+      "step": 9588
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011442475110496005,
+      "loss": 1.198,
+      "step": 9589
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011441582213491675,
+      "loss": 1.0551,
+      "step": 9590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011440689316487343,
+      "loss": 1.0217,
+      "step": 9591
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011439796419483013,
+      "loss": 1.1286,
+      "step": 9592
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011438903522478684,
+      "loss": 1.041,
+      "step": 9593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011438010625474351,
+      "loss": 1.0669,
+      "step": 9594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011437117728470022,
+      "loss": 0.9837,
+      "step": 9595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00011436224831465692,
+      "loss": 1.148,
+      "step": 9596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001143533193446136,
+      "loss": 1.2087,
+      "step": 9597
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001143443903745703,
+      "loss": 1.0389,
+      "step": 9598
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000114335461404527,
+      "loss": 1.2033,
+      "step": 9599
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011432653243448368,
+      "loss": 0.9836,
+      "step": 9600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011431760346444038,
+      "loss": 1.0157,
+      "step": 9601
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011430867449439708,
+      "loss": 1.0651,
+      "step": 9602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011429974552435376,
+      "loss": 1.031,
+      "step": 9603
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011429081655431046,
+      "loss": 1.0097,
+      "step": 9604
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011428188758426716,
+      "loss": 1.0523,
+      "step": 9605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011427295861422384,
+      "loss": 1.0174,
+      "step": 9606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011426402964418055,
+      "loss": 1.1461,
+      "step": 9607
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011425510067413725,
+      "loss": 0.9809,
+      "step": 9608
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011424617170409393,
+      "loss": 0.9308,
+      "step": 9609
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011423724273405063,
+      "loss": 0.9204,
+      "step": 9610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011422831376400733,
+      "loss": 1.1297,
+      "step": 9611
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011421938479396401,
+      "loss": 1.1214,
+      "step": 9612
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011421045582392072,
+      "loss": 0.9755,
+      "step": 9613
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011420152685387742,
+      "loss": 0.9426,
+      "step": 9614
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001141925978838341,
+      "loss": 1.1207,
+      "step": 9615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001141836689137908,
+      "loss": 0.9366,
+      "step": 9616
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001141747399437475,
+      "loss": 1.1519,
+      "step": 9617
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011416581097370418,
+      "loss": 1.1398,
+      "step": 9618
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011415688200366089,
+      "loss": 1.066,
+      "step": 9619
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011414795303361759,
+      "loss": 1.1348,
+      "step": 9620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011413902406357427,
+      "loss": 1.0557,
+      "step": 9621
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011413009509353097,
+      "loss": 0.9919,
+      "step": 9622
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011412116612348766,
+      "loss": 1.0139,
+      "step": 9623
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011411223715344435,
+      "loss": 1.1549,
+      "step": 9624
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011410330818340104,
+      "loss": 1.124,
+      "step": 9625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011409437921335775,
+      "loss": 1.0113,
+      "step": 9626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011408545024331442,
+      "loss": 1.069,
+      "step": 9627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011407652127327113,
+      "loss": 1.0675,
+      "step": 9628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011406759230322783,
+      "loss": 1.0603,
+      "step": 9629
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011405866333318451,
+      "loss": 1.0032,
+      "step": 9630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011404973436314121,
+      "loss": 1.0449,
+      "step": 9631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011404080539309792,
+      "loss": 1.0305,
+      "step": 9632
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011403187642305459,
+      "loss": 0.9338,
+      "step": 9633
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001140229474530113,
+      "loss": 1.017,
+      "step": 9634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000114014018482968,
+      "loss": 0.9796,
+      "step": 9635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011400508951292468,
+      "loss": 1.0494,
+      "step": 9636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011399616054288138,
+      "loss": 1.1483,
+      "step": 9637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011398723157283809,
+      "loss": 0.8984,
+      "step": 9638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011397830260279476,
+      "loss": 1.1652,
+      "step": 9639
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011396937363275147,
+      "loss": 1.0224,
+      "step": 9640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011396044466270817,
+      "loss": 1.121,
+      "step": 9641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011395151569266486,
+      "loss": 1.0789,
+      "step": 9642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011394258672262155,
+      "loss": 0.9496,
+      "step": 9643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011393365775257824,
+      "loss": 1.1595,
+      "step": 9644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011392472878253495,
+      "loss": 1.0531,
+      "step": 9645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011391579981249164,
+      "loss": 0.9871,
+      "step": 9646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011390687084244833,
+      "loss": 1.0036,
+      "step": 9647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011389794187240503,
+      "loss": 0.9827,
+      "step": 9648
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011388901290236171,
+      "loss": 1.2767,
+      "step": 9649
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011388008393231841,
+      "loss": 0.9758,
+      "step": 9650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011387115496227512,
+      "loss": 0.9892,
+      "step": 9651
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011386222599223179,
+      "loss": 1.0228,
+      "step": 9652
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001138532970221885,
+      "loss": 1.0635,
+      "step": 9653
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001138443680521452,
+      "loss": 1.1065,
+      "step": 9654
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011383543908210188,
+      "loss": 1.0662,
+      "step": 9655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011382651011205858,
+      "loss": 1.2338,
+      "step": 9656
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011381758114201529,
+      "loss": 1.0205,
+      "step": 9657
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011380865217197196,
+      "loss": 1.2113,
+      "step": 9658
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011379972320192867,
+      "loss": 1.1028,
+      "step": 9659
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011379079423188537,
+      "loss": 1.0712,
+      "step": 9660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011378186526184205,
+      "loss": 1.0892,
+      "step": 9661
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011377293629179875,
+      "loss": 1.0976,
+      "step": 9662
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011376400732175546,
+      "loss": 0.982,
+      "step": 9663
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011375507835171213,
+      "loss": 1.0678,
+      "step": 9664
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011374614938166884,
+      "loss": 1.1167,
+      "step": 9665
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011373722041162553,
+      "loss": 0.9699,
+      "step": 9666
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011372829144158222,
+      "loss": 1.0123,
+      "step": 9667
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011371936247153891,
+      "loss": 1.032,
+      "step": 9668
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011371043350149561,
+      "loss": 1.0174,
+      "step": 9669
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001137015045314523,
+      "loss": 1.1506,
+      "step": 9670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000113692575561409,
+      "loss": 1.0574,
+      "step": 9671
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001136836465913657,
+      "loss": 0.9645,
+      "step": 9672
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011367471762132237,
+      "loss": 1.0344,
+      "step": 9673
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011366578865127908,
+      "loss": 0.9332,
+      "step": 9674
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011365685968123578,
+      "loss": 1.1181,
+      "step": 9675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011364793071119246,
+      "loss": 1.0853,
+      "step": 9676
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011363900174114916,
+      "loss": 1.0059,
+      "step": 9677
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011363007277110587,
+      "loss": 0.9306,
+      "step": 9678
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011362114380106254,
+      "loss": 1.0282,
+      "step": 9679
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011361221483101925,
+      "loss": 1.2814,
+      "step": 9680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011360328586097595,
+      "loss": 1.0765,
+      "step": 9681
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011359435689093263,
+      "loss": 1.0518,
+      "step": 9682
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011358542792088933,
+      "loss": 1.0717,
+      "step": 9683
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011357649895084604,
+      "loss": 1.1012,
+      "step": 9684
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011356756998080271,
+      "loss": 0.9512,
+      "step": 9685
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011355864101075942,
+      "loss": 1.021,
+      "step": 9686
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011354971204071612,
+      "loss": 1.0278,
+      "step": 9687
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001135407830706728,
+      "loss": 0.9282,
+      "step": 9688
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001135318541006295,
+      "loss": 1.0637,
+      "step": 9689
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001135229251305862,
+      "loss": 1.1191,
+      "step": 9690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011351399616054288,
+      "loss": 1.0737,
+      "step": 9691
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011350506719049957,
+      "loss": 0.9855,
+      "step": 9692
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011349613822045628,
+      "loss": 1.1437,
+      "step": 9693
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011348720925041296,
+      "loss": 0.9958,
+      "step": 9694
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011347828028036966,
+      "loss": 0.966,
+      "step": 9695
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011346935131032636,
+      "loss": 0.9647,
+      "step": 9696
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011346042234028304,
+      "loss": 0.9352,
+      "step": 9697
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011345149337023974,
+      "loss": 1.1078,
+      "step": 9698
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011344256440019645,
+      "loss": 1.1657,
+      "step": 9699
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011343363543015313,
+      "loss": 1.1385,
+      "step": 9700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011342470646010983,
+      "loss": 0.9421,
+      "step": 9701
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011341577749006653,
+      "loss": 1.241,
+      "step": 9702
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011340684852002321,
+      "loss": 0.9803,
+      "step": 9703
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011339791954997991,
+      "loss": 1.1998,
+      "step": 9704
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011338899057993662,
+      "loss": 0.9946,
+      "step": 9705
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001133800616098933,
+      "loss": 1.0586,
+      "step": 9706
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011337113263985,
+      "loss": 0.9115,
+      "step": 9707
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001133622036698067,
+      "loss": 1.0547,
+      "step": 9708
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011335327469976338,
+      "loss": 0.9592,
+      "step": 9709
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011334434572972008,
+      "loss": 1.0178,
+      "step": 9710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011333541675967679,
+      "loss": 0.9903,
+      "step": 9711
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011332648778963347,
+      "loss": 1.0967,
+      "step": 9712
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011331755881959017,
+      "loss": 1.14,
+      "step": 9713
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011330862984954686,
+      "loss": 1.0002,
+      "step": 9714
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011329970087950355,
+      "loss": 1.0476,
+      "step": 9715
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011329077190946024,
+      "loss": 1.0884,
+      "step": 9716
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011328184293941695,
+      "loss": 0.9845,
+      "step": 9717
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011327291396937362,
+      "loss": 1.0435,
+      "step": 9718
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011326398499933033,
+      "loss": 1.0151,
+      "step": 9719
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011325505602928703,
+      "loss": 0.8843,
+      "step": 9720
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011324612705924371,
+      "loss": 1.0023,
+      "step": 9721
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011323719808920041,
+      "loss": 1.1074,
+      "step": 9722
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011322826911915712,
+      "loss": 0.9362,
+      "step": 9723
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011321934014911379,
+      "loss": 0.9343,
+      "step": 9724
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001132104111790705,
+      "loss": 1.0233,
+      "step": 9725
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001132014822090272,
+      "loss": 1.0792,
+      "step": 9726
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011319255323898388,
+      "loss": 1.0582,
+      "step": 9727
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011318362426894058,
+      "loss": 1.0588,
+      "step": 9728
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011317469529889729,
+      "loss": 1.0421,
+      "step": 9729
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011316576632885396,
+      "loss": 1.0619,
+      "step": 9730
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011315683735881067,
+      "loss": 0.9881,
+      "step": 9731
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011314790838876737,
+      "loss": 0.9967,
+      "step": 9732
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011313897941872405,
+      "loss": 1.0546,
+      "step": 9733
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011313005044868075,
+      "loss": 1.1078,
+      "step": 9734
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011312112147863744,
+      "loss": 1.1477,
+      "step": 9735
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011311219250859413,
+      "loss": 0.8561,
+      "step": 9736
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011310326353855084,
+      "loss": 0.9637,
+      "step": 9737
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011309433456850753,
+      "loss": 1.0466,
+      "step": 9738
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011308540559846423,
+      "loss": 1.0681,
+      "step": 9739
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011307647662842091,
+      "loss": 1.0527,
+      "step": 9740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011306754765837761,
+      "loss": 1.0912,
+      "step": 9741
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011305861868833432,
+      "loss": 0.9933,
+      "step": 9742
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011304968971829099,
+      "loss": 1.0128,
+      "step": 9743
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001130407607482477,
+      "loss": 1.0312,
+      "step": 9744
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001130318317782044,
+      "loss": 0.9888,
+      "step": 9745
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011302290280816108,
+      "loss": 0.9491,
+      "step": 9746
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011301397383811778,
+      "loss": 0.951,
+      "step": 9747
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011300504486807449,
+      "loss": 1.1105,
+      "step": 9748
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011299611589803116,
+      "loss": 0.9581,
+      "step": 9749
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011298718692798787,
+      "loss": 1.1097,
+      "step": 9750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011297825795794457,
+      "loss": 1.0104,
+      "step": 9751
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011296932898790125,
+      "loss": 0.9592,
+      "step": 9752
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011296040001785795,
+      "loss": 1.0136,
+      "step": 9753
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011295147104781466,
+      "loss": 0.9685,
+      "step": 9754
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011294254207777133,
+      "loss": 1.0485,
+      "step": 9755
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011293361310772804,
+      "loss": 1.0505,
+      "step": 9756
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011292468413768473,
+      "loss": 0.8846,
+      "step": 9757
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011291575516764142,
+      "loss": 1.0602,
+      "step": 9758
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011290682619759811,
+      "loss": 1.1246,
+      "step": 9759
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011289789722755481,
+      "loss": 1.1642,
+      "step": 9760
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001128889682575115,
+      "loss": 1.0584,
+      "step": 9761
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011288003928746819,
+      "loss": 1.0002,
+      "step": 9762
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001128711103174249,
+      "loss": 1.0112,
+      "step": 9763
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011286218134738157,
+      "loss": 0.9254,
+      "step": 9764
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011285325237733828,
+      "loss": 0.9317,
+      "step": 9765
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011284432340729498,
+      "loss": 0.9897,
+      "step": 9766
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011283539443725166,
+      "loss": 0.976,
+      "step": 9767
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011282646546720836,
+      "loss": 0.993,
+      "step": 9768
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011281753649716507,
+      "loss": 0.9888,
+      "step": 9769
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011280860752712174,
+      "loss": 0.8953,
+      "step": 9770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011279967855707845,
+      "loss": 0.9527,
+      "step": 9771
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011279074958703515,
+      "loss": 1.0007,
+      "step": 9772
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011278182061699183,
+      "loss": 1.0561,
+      "step": 9773
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011277289164694853,
+      "loss": 1.0988,
+      "step": 9774
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011276396267690524,
+      "loss": 0.9945,
+      "step": 9775
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011275503370686191,
+      "loss": 1.2118,
+      "step": 9776
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011274610473681862,
+      "loss": 1.0011,
+      "step": 9777
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011273717576677532,
+      "loss": 1.0279,
+      "step": 9778
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000112728246796732,
+      "loss": 1.0315,
+      "step": 9779
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001127193178266887,
+      "loss": 1.0817,
+      "step": 9780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001127103888566454,
+      "loss": 1.0719,
+      "step": 9781
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011270145988660208,
+      "loss": 1.085,
+      "step": 9782
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011269253091655877,
+      "loss": 0.9406,
+      "step": 9783
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011268360194651548,
+      "loss": 1.0417,
+      "step": 9784
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011267467297647216,
+      "loss": 0.9865,
+      "step": 9785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011266574400642886,
+      "loss": 0.9701,
+      "step": 9786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011265681503638556,
+      "loss": 0.9732,
+      "step": 9787
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011264788606634224,
+      "loss": 1.0398,
+      "step": 9788
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011263895709629894,
+      "loss": 0.9974,
+      "step": 9789
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011263002812625565,
+      "loss": 1.1669,
+      "step": 9790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011262109915621233,
+      "loss": 1.0342,
+      "step": 9791
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011261217018616903,
+      "loss": 1.1714,
+      "step": 9792
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011260324121612573,
+      "loss": 1.0632,
+      "step": 9793
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011259431224608241,
+      "loss": 1.0002,
+      "step": 9794
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011258538327603911,
+      "loss": 1.1016,
+      "step": 9795
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011257645430599582,
+      "loss": 1.0492,
+      "step": 9796
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001125675253359525,
+      "loss": 1.0906,
+      "step": 9797
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001125585963659092,
+      "loss": 0.9995,
+      "step": 9798
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001125496673958659,
+      "loss": 1.0272,
+      "step": 9799
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011254073842582258,
+      "loss": 1.171,
+      "step": 9800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011253180945577928,
+      "loss": 1.0682,
+      "step": 9801
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011252288048573599,
+      "loss": 0.9761,
+      "step": 9802
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011251395151569267,
+      "loss": 1.1031,
+      "step": 9803
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011250502254564937,
+      "loss": 1.0226,
+      "step": 9804
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011249609357560606,
+      "loss": 1.0793,
+      "step": 9805
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011248716460556275,
+      "loss": 1.0106,
+      "step": 9806
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011247823563551944,
+      "loss": 1.0389,
+      "step": 9807
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011246930666547615,
+      "loss": 0.9398,
+      "step": 9808
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011246037769543282,
+      "loss": 1.0598,
+      "step": 9809
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011245144872538953,
+      "loss": 1.0042,
+      "step": 9810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011244251975534623,
+      "loss": 1.0798,
+      "step": 9811
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011243359078530291,
+      "loss": 0.9341,
+      "step": 9812
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011242466181525961,
+      "loss": 1.1095,
+      "step": 9813
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011241573284521631,
+      "loss": 0.9712,
+      "step": 9814
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011240680387517299,
+      "loss": 1.0521,
+      "step": 9815
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001123978749051297,
+      "loss": 1.0045,
+      "step": 9816
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001123889459350864,
+      "loss": 1.0609,
+      "step": 9817
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011238001696504308,
+      "loss": 0.9831,
+      "step": 9818
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011237108799499978,
+      "loss": 1.0148,
+      "step": 9819
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011236215902495648,
+      "loss": 0.9455,
+      "step": 9820
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011235323005491316,
+      "loss": 1.1665,
+      "step": 9821
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011234430108486987,
+      "loss": 0.9788,
+      "step": 9822
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011233537211482657,
+      "loss": 1.0939,
+      "step": 9823
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011232644314478325,
+      "loss": 1.0663,
+      "step": 9824
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011231751417473995,
+      "loss": 1.1283,
+      "step": 9825
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011230858520469664,
+      "loss": 0.9416,
+      "step": 9826
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011229965623465333,
+      "loss": 1.0211,
+      "step": 9827
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011229072726461004,
+      "loss": 1.0819,
+      "step": 9828
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011228179829456673,
+      "loss": 1.0598,
+      "step": 9829
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011227286932452342,
+      "loss": 1.0752,
+      "step": 9830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011226394035448011,
+      "loss": 1.0994,
+      "step": 9831
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011225501138443681,
+      "loss": 1.0722,
+      "step": 9832
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011224608241439349,
+      "loss": 0.9884,
+      "step": 9833
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011223715344435019,
+      "loss": 1.0803,
+      "step": 9834
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001122282244743069,
+      "loss": 1.069,
+      "step": 9835
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001122192955042636,
+      "loss": 0.9483,
+      "step": 9836
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011221036653422028,
+      "loss": 1.0338,
+      "step": 9837
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011220143756417698,
+      "loss": 0.9574,
+      "step": 9838
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011219250859413369,
+      "loss": 0.9825,
+      "step": 9839
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011218357962409036,
+      "loss": 0.989,
+      "step": 9840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011217465065404707,
+      "loss": 0.9927,
+      "step": 9841
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011216572168400377,
+      "loss": 0.9996,
+      "step": 9842
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011215679271396045,
+      "loss": 1.0234,
+      "step": 9843
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011214786374391715,
+      "loss": 1.021,
+      "step": 9844
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011213893477387386,
+      "loss": 0.9772,
+      "step": 9845
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011213000580383053,
+      "loss": 0.9818,
+      "step": 9846
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011212107683378724,
+      "loss": 1.0326,
+      "step": 9847
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011211214786374393,
+      "loss": 1.1885,
+      "step": 9848
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011210321889370062,
+      "loss": 0.9452,
+      "step": 9849
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011209428992365731,
+      "loss": 1.0738,
+      "step": 9850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011208536095361401,
+      "loss": 1.0166,
+      "step": 9851
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011207643198357069,
+      "loss": 0.9829,
+      "step": 9852
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011206750301352739,
+      "loss": 1.0341,
+      "step": 9853
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001120585740434841,
+      "loss": 0.9761,
+      "step": 9854
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011204964507344077,
+      "loss": 1.0951,
+      "step": 9855
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011204071610339748,
+      "loss": 1.0539,
+      "step": 9856
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011203178713335418,
+      "loss": 1.0156,
+      "step": 9857
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011202285816331086,
+      "loss": 1.0331,
+      "step": 9858
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011201392919326756,
+      "loss": 0.905,
+      "step": 9859
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011200500022322427,
+      "loss": 1.144,
+      "step": 9860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011199607125318094,
+      "loss": 1.0209,
+      "step": 9861
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011198714228313765,
+      "loss": 1.1347,
+      "step": 9862
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011197821331309435,
+      "loss": 1.0628,
+      "step": 9863
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011196928434305103,
+      "loss": 0.9924,
+      "step": 9864
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011196035537300773,
+      "loss": 0.9389,
+      "step": 9865
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011195142640296444,
+      "loss": 1.0286,
+      "step": 9866
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011194249743292111,
+      "loss": 0.9856,
+      "step": 9867
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011193356846287782,
+      "loss": 0.9092,
+      "step": 9868
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011192463949283452,
+      "loss": 1.2358,
+      "step": 9869
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001119157105227912,
+      "loss": 1.1875,
+      "step": 9870
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001119067815527479,
+      "loss": 1.2147,
+      "step": 9871
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001118978525827046,
+      "loss": 0.9931,
+      "step": 9872
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011188892361266128,
+      "loss": 0.8968,
+      "step": 9873
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011187999464261797,
+      "loss": 1.1328,
+      "step": 9874
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011187106567257468,
+      "loss": 1.0439,
+      "step": 9875
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011186213670253136,
+      "loss": 0.9693,
+      "step": 9876
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011185320773248806,
+      "loss": 1.0064,
+      "step": 9877
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011184427876244476,
+      "loss": 1.055,
+      "step": 9878
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011183534979240144,
+      "loss": 1.1105,
+      "step": 9879
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011182642082235814,
+      "loss": 1.135,
+      "step": 9880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011181749185231485,
+      "loss": 1.0298,
+      "step": 9881
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011180856288227153,
+      "loss": 0.9941,
+      "step": 9882
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011179963391222823,
+      "loss": 0.9967,
+      "step": 9883
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011179070494218493,
+      "loss": 0.8781,
+      "step": 9884
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011178177597214161,
+      "loss": 1.0515,
+      "step": 9885
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011177284700209831,
+      "loss": 0.9727,
+      "step": 9886
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011176391803205502,
+      "loss": 0.9814,
+      "step": 9887
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001117549890620117,
+      "loss": 1.0143,
+      "step": 9888
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001117460600919684,
+      "loss": 1.0686,
+      "step": 9889
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001117371311219251,
+      "loss": 0.9925,
+      "step": 9890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011172820215188178,
+      "loss": 1.3718,
+      "step": 9891
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011171927318183848,
+      "loss": 0.9936,
+      "step": 9892
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011171034421179519,
+      "loss": 1.0496,
+      "step": 9893
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011170141524175187,
+      "loss": 0.9834,
+      "step": 9894
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011169248627170857,
+      "loss": 0.9758,
+      "step": 9895
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011168355730166526,
+      "loss": 1.0017,
+      "step": 9896
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011167462833162195,
+      "loss": 0.9286,
+      "step": 9897
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011166569936157864,
+      "loss": 1.0565,
+      "step": 9898
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011165677039153534,
+      "loss": 1.0853,
+      "step": 9899
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011164784142149202,
+      "loss": 1.0358,
+      "step": 9900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011163891245144873,
+      "loss": 1.0379,
+      "step": 9901
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011162998348140543,
+      "loss": 0.9426,
+      "step": 9902
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001116210545113621,
+      "loss": 1.0342,
+      "step": 9903
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011161212554131881,
+      "loss": 0.9056,
+      "step": 9904
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011160319657127551,
+      "loss": 0.9882,
+      "step": 9905
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011159426760123219,
+      "loss": 0.9924,
+      "step": 9906
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001115853386311889,
+      "loss": 1.0309,
+      "step": 9907
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001115764096611456,
+      "loss": 1.131,
+      "step": 9908
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011156748069110228,
+      "loss": 1.0014,
+      "step": 9909
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011155855172105898,
+      "loss": 0.9449,
+      "step": 9910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011154962275101568,
+      "loss": 0.9717,
+      "step": 9911
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011154069378097236,
+      "loss": 1.0767,
+      "step": 9912
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011153176481092907,
+      "loss": 1.1101,
+      "step": 9913
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011152283584088577,
+      "loss": 1.0172,
+      "step": 9914
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011151390687084245,
+      "loss": 1.0344,
+      "step": 9915
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011150497790079915,
+      "loss": 0.9284,
+      "step": 9916
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011149604893075584,
+      "loss": 1.0545,
+      "step": 9917
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011148711996071253,
+      "loss": 1.0142,
+      "step": 9918
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011147819099066924,
+      "loss": 1.0219,
+      "step": 9919
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011146926202062593,
+      "loss": 1.1141,
+      "step": 9920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011146033305058262,
+      "loss": 0.9394,
+      "step": 9921
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011145140408053931,
+      "loss": 0.9996,
+      "step": 9922
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011144247511049601,
+      "loss": 1.03,
+      "step": 9923
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011143354614045269,
+      "loss": 1.0234,
+      "step": 9924
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011142461717040939,
+      "loss": 0.9333,
+      "step": 9925
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001114156882003661,
+      "loss": 1.0789,
+      "step": 9926
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011140675923032277,
+      "loss": 0.9921,
+      "step": 9927
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011139783026027948,
+      "loss": 1.0224,
+      "step": 9928
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011138890129023618,
+      "loss": 1.0037,
+      "step": 9929
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011137997232019286,
+      "loss": 1.024,
+      "step": 9930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011137104335014956,
+      "loss": 0.9563,
+      "step": 9931
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011136211438010627,
+      "loss": 1.0386,
+      "step": 9932
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011135318541006297,
+      "loss": 1.1691,
+      "step": 9933
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011134425644001965,
+      "loss": 1.2683,
+      "step": 9934
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011133532746997635,
+      "loss": 1.059,
+      "step": 9935
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011132639849993306,
+      "loss": 1.0204,
+      "step": 9936
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011131746952988973,
+      "loss": 1.0472,
+      "step": 9937
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011130854055984644,
+      "loss": 1.1192,
+      "step": 9938
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011129961158980313,
+      "loss": 1.033,
+      "step": 9939
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011129068261975982,
+      "loss": 1.2323,
+      "step": 9940
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011128175364971651,
+      "loss": 1.0221,
+      "step": 9941
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011127282467967321,
+      "loss": 0.9549,
+      "step": 9942
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011126389570962989,
+      "loss": 1.0136,
+      "step": 9943
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011125496673958659,
+      "loss": 1.0002,
+      "step": 9944
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001112460377695433,
+      "loss": 0.9341,
+      "step": 9945
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011123710879949997,
+      "loss": 1.0326,
+      "step": 9946
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011122817982945668,
+      "loss": 0.9233,
+      "step": 9947
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011121925085941338,
+      "loss": 0.8805,
+      "step": 9948
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011121032188937006,
+      "loss": 1.0029,
+      "step": 9949
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011120139291932676,
+      "loss": 1.0814,
+      "step": 9950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011119246394928347,
+      "loss": 1.098,
+      "step": 9951
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011118353497924014,
+      "loss": 1.0797,
+      "step": 9952
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011117460600919685,
+      "loss": 0.9075,
+      "step": 9953
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011116567703915355,
+      "loss": 1.0084,
+      "step": 9954
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011115674806911023,
+      "loss": 1.1225,
+      "step": 9955
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011114781909906693,
+      "loss": 1.1534,
+      "step": 9956
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011113889012902364,
+      "loss": 1.128,
+      "step": 9957
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011112996115898031,
+      "loss": 0.9111,
+      "step": 9958
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011112103218893702,
+      "loss": 1.0366,
+      "step": 9959
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011111210321889372,
+      "loss": 1.0368,
+      "step": 9960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001111031742488504,
+      "loss": 1.0624,
+      "step": 9961
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001110942452788071,
+      "loss": 1.1966,
+      "step": 9962
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011108531630876379,
+      "loss": 1.0501,
+      "step": 9963
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011107638733872048,
+      "loss": 0.8251,
+      "step": 9964
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011106745836867717,
+      "loss": 1.0373,
+      "step": 9965
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011105852939863388,
+      "loss": 0.9276,
+      "step": 9966
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011104960042859055,
+      "loss": 1.0696,
+      "step": 9967
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011104067145854726,
+      "loss": 1.0127,
+      "step": 9968
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011103174248850396,
+      "loss": 1.0004,
+      "step": 9969
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011102281351846064,
+      "loss": 0.9747,
+      "step": 9970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011101388454841734,
+      "loss": 1.1191,
+      "step": 9971
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011100495557837405,
+      "loss": 1.09,
+      "step": 9972
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011099602660833072,
+      "loss": 1.0461,
+      "step": 9973
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011098709763828743,
+      "loss": 1.0548,
+      "step": 9974
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011097816866824413,
+      "loss": 0.9864,
+      "step": 9975
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011096923969820081,
+      "loss": 1.2817,
+      "step": 9976
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011096031072815751,
+      "loss": 0.967,
+      "step": 9977
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011095138175811422,
+      "loss": 0.9828,
+      "step": 9978
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001109424527880709,
+      "loss": 0.9446,
+      "step": 9979
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001109335238180276,
+      "loss": 1.1263,
+      "step": 9980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001109245948479843,
+      "loss": 1.0154,
+      "step": 9981
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011091566587794098,
+      "loss": 0.9204,
+      "step": 9982
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011090673690789768,
+      "loss": 1.0107,
+      "step": 9983
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011089780793785437,
+      "loss": 1.2645,
+      "step": 9984
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011088887896781106,
+      "loss": 1.1648,
+      "step": 9985
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011087994999776777,
+      "loss": 1.0382,
+      "step": 9986
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011087102102772446,
+      "loss": 0.9951,
+      "step": 9987
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011086209205768115,
+      "loss": 1.0024,
+      "step": 9988
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011085316308763784,
+      "loss": 1.1383,
+      "step": 9989
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011084423411759454,
+      "loss": 0.9139,
+      "step": 9990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011083530514755122,
+      "loss": 1.1104,
+      "step": 9991
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011082637617750793,
+      "loss": 0.9571,
+      "step": 9992
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011081744720746463,
+      "loss": 0.9541,
+      "step": 9993
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001108085182374213,
+      "loss": 1.0439,
+      "step": 9994
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011079958926737801,
+      "loss": 1.0749,
+      "step": 9995
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011079066029733471,
+      "loss": 1.0359,
+      "step": 9996
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011078173132729139,
+      "loss": 1.1121,
+      "step": 9997
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001107728023572481,
+      "loss": 1.0268,
+      "step": 9998
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001107638733872048,
+      "loss": 1.0371,
+      "step": 9999
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011075494441716148,
+      "loss": 1.0015,
+      "step": 10000
+    },
+    {
+      "epoch": 1.34,
+      "eval_loss": 1.0268536806106567,
+      "eval_runtime": 3051.4397,
+      "eval_samples_per_second": 6.163,
+      "eval_steps_per_second": 0.77,
+      "step": 10000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011074601544711818,
+      "loss": 0.9281,
+      "step": 10001
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011073708647707488,
+      "loss": 1.1475,
+      "step": 10002
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011072815750703156,
+      "loss": 0.9613,
+      "step": 10003
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011071922853698827,
+      "loss": 1.0496,
+      "step": 10004
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011071029956694497,
+      "loss": 1.0233,
+      "step": 10005
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011070137059690165,
+      "loss": 1.0743,
+      "step": 10006
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011069244162685835,
+      "loss": 0.894,
+      "step": 10007
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011068351265681504,
+      "loss": 1.0965,
+      "step": 10008
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011067458368677173,
+      "loss": 1.0998,
+      "step": 10009
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011066565471672844,
+      "loss": 0.966,
+      "step": 10010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011065672574668513,
+      "loss": 1.2649,
+      "step": 10011
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011064779677664182,
+      "loss": 1.1362,
+      "step": 10012
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001106388678065985,
+      "loss": 1.0741,
+      "step": 10013
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011062993883655521,
+      "loss": 1.098,
+      "step": 10014
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011062100986651189,
+      "loss": 1.0436,
+      "step": 10015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011061208089646859,
+      "loss": 1.1621,
+      "step": 10016
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001106031519264253,
+      "loss": 0.9859,
+      "step": 10017
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011059422295638197,
+      "loss": 1.0337,
+      "step": 10018
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011058529398633868,
+      "loss": 0.9324,
+      "step": 10019
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011057636501629538,
+      "loss": 0.9573,
+      "step": 10020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011056743604625206,
+      "loss": 0.9513,
+      "step": 10021
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011055850707620876,
+      "loss": 0.9716,
+      "step": 10022
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011054957810616547,
+      "loss": 1.1421,
+      "step": 10023
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011054064913612214,
+      "loss": 0.9794,
+      "step": 10024
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011053172016607885,
+      "loss": 0.8112,
+      "step": 10025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011052279119603555,
+      "loss": 1.0637,
+      "step": 10026
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011051386222599223,
+      "loss": 1.0129,
+      "step": 10027
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011050493325594893,
+      "loss": 1.0939,
+      "step": 10028
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011049600428590564,
+      "loss": 1.0356,
+      "step": 10029
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011048707531586231,
+      "loss": 1.0277,
+      "step": 10030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011047814634581902,
+      "loss": 1.0968,
+      "step": 10031
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011046921737577571,
+      "loss": 1.1249,
+      "step": 10032
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011046028840573241,
+      "loss": 1.0912,
+      "step": 10033
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011045135943568909,
+      "loss": 1.0442,
+      "step": 10034
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011044243046564579,
+      "loss": 0.9818,
+      "step": 10035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001104335014956025,
+      "loss": 1.0378,
+      "step": 10036
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011042457252555917,
+      "loss": 1.0907,
+      "step": 10037
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011041564355551588,
+      "loss": 0.9523,
+      "step": 10038
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011040671458547258,
+      "loss": 1.036,
+      "step": 10039
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011039778561542926,
+      "loss": 1.1653,
+      "step": 10040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011038885664538596,
+      "loss": 0.9729,
+      "step": 10041
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011037992767534267,
+      "loss": 1.2002,
+      "step": 10042
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011037099870529934,
+      "loss": 1.0026,
+      "step": 10043
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011036206973525605,
+      "loss": 0.9654,
+      "step": 10044
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011035314076521275,
+      "loss": 1.0401,
+      "step": 10045
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011034421179516943,
+      "loss": 0.8193,
+      "step": 10046
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011033528282512613,
+      "loss": 1.2073,
+      "step": 10047
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011032635385508284,
+      "loss": 1.0897,
+      "step": 10048
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011031742488503951,
+      "loss": 1.0025,
+      "step": 10049
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011030849591499622,
+      "loss": 1.045,
+      "step": 10050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011029956694495292,
+      "loss": 1.0284,
+      "step": 10051
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001102906379749096,
+      "loss": 1.0127,
+      "step": 10052
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001102817090048663,
+      "loss": 1.0595,
+      "step": 10053
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011027278003482299,
+      "loss": 0.9829,
+      "step": 10054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011026385106477968,
+      "loss": 1.0125,
+      "step": 10055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011025492209473637,
+      "loss": 0.9609,
+      "step": 10056
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011024599312469308,
+      "loss": 1.1142,
+      "step": 10057
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011023706415464975,
+      "loss": 1.1007,
+      "step": 10058
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011022813518460646,
+      "loss": 1.0547,
+      "step": 10059
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011021920621456316,
+      "loss": 1.0258,
+      "step": 10060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011021027724451984,
+      "loss": 1.0524,
+      "step": 10061
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011020134827447654,
+      "loss": 0.9699,
+      "step": 10062
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011019241930443325,
+      "loss": 1.049,
+      "step": 10063
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011018349033438992,
+      "loss": 1.0255,
+      "step": 10064
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011017456136434663,
+      "loss": 1.0039,
+      "step": 10065
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011016563239430333,
+      "loss": 0.9511,
+      "step": 10066
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011015670342426001,
+      "loss": 1.1578,
+      "step": 10067
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011014777445421671,
+      "loss": 1.0119,
+      "step": 10068
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011013884548417342,
+      "loss": 0.9483,
+      "step": 10069
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001101299165141301,
+      "loss": 1.0548,
+      "step": 10070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001101209875440868,
+      "loss": 0.9706,
+      "step": 10071
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001101120585740435,
+      "loss": 1.1933,
+      "step": 10072
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011010312960400018,
+      "loss": 1.0051,
+      "step": 10073
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011009420063395688,
+      "loss": 0.9249,
+      "step": 10074
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011008527166391357,
+      "loss": 1.1654,
+      "step": 10075
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011007634269387026,
+      "loss": 0.9604,
+      "step": 10076
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011006741372382697,
+      "loss": 1.0846,
+      "step": 10077
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011005848475378366,
+      "loss": 0.9617,
+      "step": 10078
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011004955578374035,
+      "loss": 1.0898,
+      "step": 10079
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011004062681369704,
+      "loss": 1.0441,
+      "step": 10080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011003169784365374,
+      "loss": 1.0755,
+      "step": 10081
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011002276887361042,
+      "loss": 1.1305,
+      "step": 10082
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011001383990356712,
+      "loss": 1.1605,
+      "step": 10083
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011000491093352383,
+      "loss": 1.0669,
+      "step": 10084
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001099959819634805,
+      "loss": 1.044,
+      "step": 10085
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010998705299343721,
+      "loss": 0.9461,
+      "step": 10086
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010997812402339391,
+      "loss": 1.0395,
+      "step": 10087
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010996919505335059,
+      "loss": 1.0669,
+      "step": 10088
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001099602660833073,
+      "loss": 1.1038,
+      "step": 10089
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000109951337113264,
+      "loss": 1.0173,
+      "step": 10090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010994240814322068,
+      "loss": 0.9027,
+      "step": 10091
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010993347917317738,
+      "loss": 1.0289,
+      "step": 10092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010992455020313408,
+      "loss": 1.0421,
+      "step": 10093
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010991562123309076,
+      "loss": 0.9974,
+      "step": 10094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010990669226304746,
+      "loss": 1.0679,
+      "step": 10095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010989776329300417,
+      "loss": 0.9606,
+      "step": 10096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010988883432296085,
+      "loss": 0.9166,
+      "step": 10097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010987990535291755,
+      "loss": 1.0391,
+      "step": 10098
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010987097638287424,
+      "loss": 0.9436,
+      "step": 10099
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010986204741283093,
+      "loss": 0.9946,
+      "step": 10100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010985311844278763,
+      "loss": 0.9597,
+      "step": 10101
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010984418947274433,
+      "loss": 1.0116,
+      "step": 10102
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010983526050270102,
+      "loss": 0.981,
+      "step": 10103
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001098263315326577,
+      "loss": 0.9886,
+      "step": 10104
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010981740256261441,
+      "loss": 1.0957,
+      "step": 10105
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010980847359257109,
+      "loss": 1.1866,
+      "step": 10106
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010979954462252779,
+      "loss": 1.0009,
+      "step": 10107
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001097906156524845,
+      "loss": 1.0607,
+      "step": 10108
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010978168668244117,
+      "loss": 1.0262,
+      "step": 10109
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010977275771239788,
+      "loss": 1.1078,
+      "step": 10110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010976382874235458,
+      "loss": 1.1171,
+      "step": 10111
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010975489977231126,
+      "loss": 1.1137,
+      "step": 10112
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010974597080226796,
+      "loss": 0.9258,
+      "step": 10113
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010973704183222467,
+      "loss": 1.0175,
+      "step": 10114
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010972811286218134,
+      "loss": 1.1109,
+      "step": 10115
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010971918389213805,
+      "loss": 0.9307,
+      "step": 10116
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010971025492209475,
+      "loss": 1.0162,
+      "step": 10117
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010970132595205143,
+      "loss": 1.0351,
+      "step": 10118
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00010969239698200813,
+      "loss": 1.14,
+      "step": 10119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010968346801196484,
+      "loss": 1.0598,
+      "step": 10120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010967453904192151,
+      "loss": 1.0165,
+      "step": 10121
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010966561007187822,
+      "loss": 0.971,
+      "step": 10122
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001096566811018349,
+      "loss": 1.1029,
+      "step": 10123
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001096477521317916,
+      "loss": 1.1526,
+      "step": 10124
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010963882316174829,
+      "loss": 1.0167,
+      "step": 10125
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010962989419170499,
+      "loss": 1.1376,
+      "step": 10126
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010962096522166168,
+      "loss": 1.1651,
+      "step": 10127
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010961203625161837,
+      "loss": 1.2586,
+      "step": 10128
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010960310728157508,
+      "loss": 0.9899,
+      "step": 10129
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010959417831153178,
+      "loss": 1.1373,
+      "step": 10130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010958524934148846,
+      "loss": 0.9668,
+      "step": 10131
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010957632037144516,
+      "loss": 0.9632,
+      "step": 10132
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010956739140140187,
+      "loss": 1.0868,
+      "step": 10133
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010955846243135854,
+      "loss": 0.8809,
+      "step": 10134
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010954953346131525,
+      "loss": 1.0344,
+      "step": 10135
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010954060449127195,
+      "loss": 0.9932,
+      "step": 10136
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010953167552122863,
+      "loss": 1.0458,
+      "step": 10137
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010952274655118533,
+      "loss": 0.9556,
+      "step": 10138
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010951381758114204,
+      "loss": 0.9421,
+      "step": 10139
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010950488861109871,
+      "loss": 1.0177,
+      "step": 10140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010949595964105542,
+      "loss": 1.0295,
+      "step": 10141
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010948703067101212,
+      "loss": 0.9708,
+      "step": 10142
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001094781017009688,
+      "loss": 1.0901,
+      "step": 10143
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001094691727309255,
+      "loss": 1.116,
+      "step": 10144
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010946024376088219,
+      "loss": 1.1065,
+      "step": 10145
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010945131479083888,
+      "loss": 0.9539,
+      "step": 10146
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010944238582079557,
+      "loss": 0.9601,
+      "step": 10147
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010943345685075228,
+      "loss": 0.8714,
+      "step": 10148
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010942452788070895,
+      "loss": 0.9954,
+      "step": 10149
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010941559891066566,
+      "loss": 1.0079,
+      "step": 10150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010940666994062236,
+      "loss": 1.0549,
+      "step": 10151
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010939774097057904,
+      "loss": 1.013,
+      "step": 10152
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010938881200053574,
+      "loss": 1.0902,
+      "step": 10153
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010937988303049245,
+      "loss": 1.1225,
+      "step": 10154
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010937095406044912,
+      "loss": 1.0246,
+      "step": 10155
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010936202509040583,
+      "loss": 1.0526,
+      "step": 10156
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010935309612036253,
+      "loss": 1.0197,
+      "step": 10157
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010934416715031921,
+      "loss": 0.9852,
+      "step": 10158
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010933523818027591,
+      "loss": 0.9484,
+      "step": 10159
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010932630921023262,
+      "loss": 1.0271,
+      "step": 10160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001093173802401893,
+      "loss": 0.976,
+      "step": 10161
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000109308451270146,
+      "loss": 0.8752,
+      "step": 10162
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001092995223001027,
+      "loss": 1.1548,
+      "step": 10163
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010929059333005938,
+      "loss": 0.9216,
+      "step": 10164
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010928166436001608,
+      "loss": 1.1632,
+      "step": 10165
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010927273538997277,
+      "loss": 1.0061,
+      "step": 10166
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010926380641992946,
+      "loss": 1.0599,
+      "step": 10167
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010925487744988617,
+      "loss": 0.9191,
+      "step": 10168
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010924594847984286,
+      "loss": 1.0171,
+      "step": 10169
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010923701950979955,
+      "loss": 0.9933,
+      "step": 10170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010922809053975624,
+      "loss": 1.0044,
+      "step": 10171
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010921916156971294,
+      "loss": 1.1035,
+      "step": 10172
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010921023259966962,
+      "loss": 0.9961,
+      "step": 10173
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010920130362962632,
+      "loss": 0.9766,
+      "step": 10174
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010919237465958303,
+      "loss": 1.0566,
+      "step": 10175
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001091834456895397,
+      "loss": 0.9987,
+      "step": 10176
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010917451671949641,
+      "loss": 0.9324,
+      "step": 10177
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010916558774945311,
+      "loss": 0.9381,
+      "step": 10178
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010915665877940979,
+      "loss": 0.8962,
+      "step": 10179
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001091477298093665,
+      "loss": 1.0686,
+      "step": 10180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001091388008393232,
+      "loss": 1.0753,
+      "step": 10181
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010912987186927988,
+      "loss": 0.987,
+      "step": 10182
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010912094289923658,
+      "loss": 0.9388,
+      "step": 10183
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010911201392919328,
+      "loss": 0.9752,
+      "step": 10184
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010910308495914996,
+      "loss": 1.0272,
+      "step": 10185
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010909415598910666,
+      "loss": 0.9268,
+      "step": 10186
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010908522701906337,
+      "loss": 0.9992,
+      "step": 10187
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010907629804902005,
+      "loss": 0.9586,
+      "step": 10188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010906736907897675,
+      "loss": 1.0979,
+      "step": 10189
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010905844010893344,
+      "loss": 1.0,
+      "step": 10190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010904951113889013,
+      "loss": 0.9926,
+      "step": 10191
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010904058216884682,
+      "loss": 1.0545,
+      "step": 10192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010903165319880352,
+      "loss": 1.0564,
+      "step": 10193
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00010902272422876022,
+      "loss": 0.986,
+      "step": 10194
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001090137952587169,
+      "loss": 1.0689,
+      "step": 10195
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010900486628867361,
+      "loss": 1.0354,
+      "step": 10196
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010899593731863029,
+      "loss": 1.1535,
+      "step": 10197
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010898700834858699,
+      "loss": 0.8887,
+      "step": 10198
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001089780793785437,
+      "loss": 0.994,
+      "step": 10199
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010896915040850037,
+      "loss": 0.9209,
+      "step": 10200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010896022143845708,
+      "loss": 1.1976,
+      "step": 10201
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010895129246841378,
+      "loss": 1.1378,
+      "step": 10202
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010894236349837046,
+      "loss": 1.1,
+      "step": 10203
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010893343452832716,
+      "loss": 0.9254,
+      "step": 10204
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010892450555828386,
+      "loss": 1.1143,
+      "step": 10205
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010891557658824054,
+      "loss": 1.108,
+      "step": 10206
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010890664761819725,
+      "loss": 1.0409,
+      "step": 10207
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010889771864815395,
+      "loss": 1.0627,
+      "step": 10208
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010888878967811063,
+      "loss": 1.0824,
+      "step": 10209
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010887986070806733,
+      "loss": 0.8747,
+      "step": 10210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010887093173802403,
+      "loss": 1.3657,
+      "step": 10211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010886200276798071,
+      "loss": 0.9767,
+      "step": 10212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010885307379793742,
+      "loss": 1.0462,
+      "step": 10213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001088441448278941,
+      "loss": 1.0121,
+      "step": 10214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001088352158578508,
+      "loss": 1.0008,
+      "step": 10215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010882628688780749,
+      "loss": 1.1191,
+      "step": 10216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010881735791776419,
+      "loss": 0.9616,
+      "step": 10217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010880842894772088,
+      "loss": 1.0923,
+      "step": 10218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010879949997767757,
+      "loss": 1.0781,
+      "step": 10219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010879057100763428,
+      "loss": 1.0173,
+      "step": 10220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010878164203759095,
+      "loss": 1.0036,
+      "step": 10221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010877271306754766,
+      "loss": 1.0567,
+      "step": 10222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010876378409750436,
+      "loss": 1.0354,
+      "step": 10223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010875485512746104,
+      "loss": 1.0595,
+      "step": 10224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010874592615741774,
+      "loss": 1.0048,
+      "step": 10225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010873699718737445,
+      "loss": 0.9375,
+      "step": 10226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010872806821733115,
+      "loss": 1.1783,
+      "step": 10227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010871913924728783,
+      "loss": 1.0586,
+      "step": 10228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010871021027724453,
+      "loss": 1.0365,
+      "step": 10229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010870128130720124,
+      "loss": 1.0391,
+      "step": 10230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010869235233715791,
+      "loss": 0.981,
+      "step": 10231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010868342336711462,
+      "loss": 1.0954,
+      "step": 10232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010867449439707132,
+      "loss": 1.0998,
+      "step": 10233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000108665565427028,
+      "loss": 1.0575,
+      "step": 10234
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001086566364569847,
+      "loss": 1.0755,
+      "step": 10235
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010864770748694139,
+      "loss": 0.9744,
+      "step": 10236
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010863877851689808,
+      "loss": 1.0923,
+      "step": 10237
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010862984954685477,
+      "loss": 1.0188,
+      "step": 10238
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010862092057681148,
+      "loss": 1.1013,
+      "step": 10239
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010861199160676815,
+      "loss": 1.0195,
+      "step": 10240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010860306263672486,
+      "loss": 1.0218,
+      "step": 10241
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010859413366668156,
+      "loss": 0.9323,
+      "step": 10242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010858520469663824,
+      "loss": 1.111,
+      "step": 10243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010857627572659494,
+      "loss": 0.9482,
+      "step": 10244
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010856734675655165,
+      "loss": 1.0815,
+      "step": 10245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010855841778650832,
+      "loss": 1.023,
+      "step": 10246
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010854948881646503,
+      "loss": 1.0137,
+      "step": 10247
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010854055984642173,
+      "loss": 1.1287,
+      "step": 10248
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010853163087637841,
+      "loss": 0.949,
+      "step": 10249
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010852270190633511,
+      "loss": 1.0503,
+      "step": 10250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010851377293629182,
+      "loss": 0.9384,
+      "step": 10251
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001085048439662485,
+      "loss": 0.9789,
+      "step": 10252
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001084959149962052,
+      "loss": 1.0064,
+      "step": 10253
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001084869860261619,
+      "loss": 1.0345,
+      "step": 10254
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010847805705611858,
+      "loss": 1.0755,
+      "step": 10255
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010846912808607528,
+      "loss": 1.15,
+      "step": 10256
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010846019911603197,
+      "loss": 1.1149,
+      "step": 10257
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010845127014598866,
+      "loss": 0.9774,
+      "step": 10258
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010844234117594537,
+      "loss": 1.0169,
+      "step": 10259
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010843341220590206,
+      "loss": 1.1019,
+      "step": 10260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010842448323585875,
+      "loss": 0.9933,
+      "step": 10261
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010841555426581544,
+      "loss": 0.9601,
+      "step": 10262
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010840662529577214,
+      "loss": 1.0934,
+      "step": 10263
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010839769632572882,
+      "loss": 1.0478,
+      "step": 10264
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010838876735568552,
+      "loss": 1.0441,
+      "step": 10265
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010837983838564223,
+      "loss": 0.888,
+      "step": 10266
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001083709094155989,
+      "loss": 1.0897,
+      "step": 10267
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010836198044555561,
+      "loss": 0.8933,
+      "step": 10268
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00010835305147551231,
+      "loss": 0.9865,
+      "step": 10269
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010834412250546899,
+      "loss": 1.002,
+      "step": 10270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001083351935354257,
+      "loss": 0.9611,
+      "step": 10271
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001083262645653824,
+      "loss": 1.1043,
+      "step": 10272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010831733559533908,
+      "loss": 0.9773,
+      "step": 10273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010830840662529578,
+      "loss": 1.078,
+      "step": 10274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010829947765525248,
+      "loss": 1.147,
+      "step": 10275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010829054868520916,
+      "loss": 1.0911,
+      "step": 10276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010828161971516586,
+      "loss": 0.9815,
+      "step": 10277
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010827269074512257,
+      "loss": 0.9713,
+      "step": 10278
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010826376177507925,
+      "loss": 1.1362,
+      "step": 10279
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010825483280503595,
+      "loss": 1.0681,
+      "step": 10280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010824590383499264,
+      "loss": 1.0351,
+      "step": 10281
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010823697486494933,
+      "loss": 0.9897,
+      "step": 10282
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010822804589490602,
+      "loss": 1.1254,
+      "step": 10283
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010821911692486272,
+      "loss": 1.1249,
+      "step": 10284
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010821018795481942,
+      "loss": 0.8914,
+      "step": 10285
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001082012589847761,
+      "loss": 0.9387,
+      "step": 10286
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010819233001473281,
+      "loss": 0.9545,
+      "step": 10287
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010818340104468949,
+      "loss": 1.0449,
+      "step": 10288
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010817447207464619,
+      "loss": 1.1084,
+      "step": 10289
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001081655431046029,
+      "loss": 0.9061,
+      "step": 10290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010815661413455957,
+      "loss": 1.0757,
+      "step": 10291
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010814768516451628,
+      "loss": 1.0119,
+      "step": 10292
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010813875619447298,
+      "loss": 1.0727,
+      "step": 10293
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010812982722442966,
+      "loss": 1.0432,
+      "step": 10294
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010812089825438636,
+      "loss": 0.9882,
+      "step": 10295
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010811196928434306,
+      "loss": 1.1144,
+      "step": 10296
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010810304031429974,
+      "loss": 0.9194,
+      "step": 10297
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010809411134425645,
+      "loss": 1.0044,
+      "step": 10298
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010808518237421315,
+      "loss": 1.0901,
+      "step": 10299
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010807625340416983,
+      "loss": 1.0265,
+      "step": 10300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010806732443412653,
+      "loss": 1.0099,
+      "step": 10301
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010805839546408323,
+      "loss": 1.0405,
+      "step": 10302
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010804946649403991,
+      "loss": 0.991,
+      "step": 10303
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010804053752399662,
+      "loss": 0.9934,
+      "step": 10304
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001080316085539533,
+      "loss": 1.0234,
+      "step": 10305
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010802267958391,
+      "loss": 1.0649,
+      "step": 10306
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010801375061386669,
+      "loss": 1.2253,
+      "step": 10307
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010800482164382339,
+      "loss": 0.8819,
+      "step": 10308
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010799589267378008,
+      "loss": 0.9369,
+      "step": 10309
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010798696370373677,
+      "loss": 1.0831,
+      "step": 10310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010797803473369348,
+      "loss": 1.0834,
+      "step": 10311
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010796910576365015,
+      "loss": 1.0392,
+      "step": 10312
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010796017679360686,
+      "loss": 1.0618,
+      "step": 10313
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010795124782356356,
+      "loss": 0.9887,
+      "step": 10314
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010794231885352024,
+      "loss": 1.0414,
+      "step": 10315
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010793338988347694,
+      "loss": 1.0099,
+      "step": 10316
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010792446091343365,
+      "loss": 1.014,
+      "step": 10317
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010791553194339032,
+      "loss": 1.961,
+      "step": 10318
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010790660297334703,
+      "loss": 0.9498,
+      "step": 10319
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010789767400330373,
+      "loss": 0.9886,
+      "step": 10320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010788874503326041,
+      "loss": 1.0543,
+      "step": 10321
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010787981606321711,
+      "loss": 1.061,
+      "step": 10322
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010787088709317382,
+      "loss": 0.9567,
+      "step": 10323
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010786195812313049,
+      "loss": 0.9828,
+      "step": 10324
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001078530291530872,
+      "loss": 1.0059,
+      "step": 10325
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001078441001830439,
+      "loss": 1.1415,
+      "step": 10326
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010783517121300059,
+      "loss": 1.1606,
+      "step": 10327
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010782624224295728,
+      "loss": 1.0416,
+      "step": 10328
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010781731327291397,
+      "loss": 0.9282,
+      "step": 10329
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010780838430287068,
+      "loss": 1.1405,
+      "step": 10330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010779945533282735,
+      "loss": 1.0635,
+      "step": 10331
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010779052636278406,
+      "loss": 1.0321,
+      "step": 10332
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010778159739274076,
+      "loss": 0.9726,
+      "step": 10333
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010777266842269744,
+      "loss": 1.0508,
+      "step": 10334
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010776373945265414,
+      "loss": 1.0849,
+      "step": 10335
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010775481048261085,
+      "loss": 0.9378,
+      "step": 10336
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010774588151256752,
+      "loss": 1.0241,
+      "step": 10337
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010773695254252423,
+      "loss": 1.0429,
+      "step": 10338
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010772802357248093,
+      "loss": 1.0038,
+      "step": 10339
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010771909460243761,
+      "loss": 0.9949,
+      "step": 10340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010771016563239431,
+      "loss": 0.9807,
+      "step": 10341
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010770123666235102,
+      "loss": 0.9336,
+      "step": 10342
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 1.0119,
+      "step": 10343
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001076833787222644,
+      "loss": 1.0617,
+      "step": 10344
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001076744497522211,
+      "loss": 1.0437,
+      "step": 10345
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010766552078217778,
+      "loss": 1.0655,
+      "step": 10346
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010765659181213448,
+      "loss": 1.1744,
+      "step": 10347
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010764766284209117,
+      "loss": 1.0471,
+      "step": 10348
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010763873387204786,
+      "loss": 0.8671,
+      "step": 10349
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010762980490200457,
+      "loss": 1.0054,
+      "step": 10350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010762087593196126,
+      "loss": 0.9226,
+      "step": 10351
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010761194696191795,
+      "loss": 1.0075,
+      "step": 10352
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010760301799187464,
+      "loss": 1.1217,
+      "step": 10353
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010759408902183134,
+      "loss": 1.1243,
+      "step": 10354
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010758516005178802,
+      "loss": 1.0207,
+      "step": 10355
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010757623108174472,
+      "loss": 0.9937,
+      "step": 10356
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010756730211170143,
+      "loss": 1.0023,
+      "step": 10357
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001075583731416581,
+      "loss": 1.0911,
+      "step": 10358
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010754944417161481,
+      "loss": 0.9899,
+      "step": 10359
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010754051520157151,
+      "loss": 1.1024,
+      "step": 10360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010753158623152819,
+      "loss": 1.1077,
+      "step": 10361
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001075226572614849,
+      "loss": 1.1441,
+      "step": 10362
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001075137282914416,
+      "loss": 1.0023,
+      "step": 10363
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010750479932139827,
+      "loss": 1.0554,
+      "step": 10364
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010749587035135498,
+      "loss": 0.9767,
+      "step": 10365
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010748694138131168,
+      "loss": 1.1074,
+      "step": 10366
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010747801241126836,
+      "loss": 1.1479,
+      "step": 10367
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010746908344122506,
+      "loss": 1.0671,
+      "step": 10368
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010746015447118177,
+      "loss": 0.9345,
+      "step": 10369
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010745122550113844,
+      "loss": 1.0075,
+      "step": 10370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010744229653109515,
+      "loss": 1.0879,
+      "step": 10371
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010743336756105184,
+      "loss": 1.0659,
+      "step": 10372
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010742443859100853,
+      "loss": 1.1082,
+      "step": 10373
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010741550962096522,
+      "loss": 1.138,
+      "step": 10374
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010740658065092192,
+      "loss": 1.0818,
+      "step": 10375
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010739765168087861,
+      "loss": 1.0963,
+      "step": 10376
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001073887227108353,
+      "loss": 0.918,
+      "step": 10377
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010737979374079201,
+      "loss": 0.9492,
+      "step": 10378
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010737086477074869,
+      "loss": 1.049,
+      "step": 10379
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010736193580070539,
+      "loss": 1.1615,
+      "step": 10380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001073530068306621,
+      "loss": 1.0284,
+      "step": 10381
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010734407786061877,
+      "loss": 0.9589,
+      "step": 10382
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010733514889057548,
+      "loss": 1.0868,
+      "step": 10383
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010732621992053218,
+      "loss": 1.1627,
+      "step": 10384
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010731729095048886,
+      "loss": 1.0847,
+      "step": 10385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010730836198044556,
+      "loss": 1.0055,
+      "step": 10386
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010729943301040226,
+      "loss": 1.1075,
+      "step": 10387
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010729050404035894,
+      "loss": 0.9918,
+      "step": 10388
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010728157507031565,
+      "loss": 1.0324,
+      "step": 10389
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010727264610027235,
+      "loss": 0.9826,
+      "step": 10390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010726371713022903,
+      "loss": 1.0015,
+      "step": 10391
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010725478816018573,
+      "loss": 0.9691,
+      "step": 10392
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010724585919014243,
+      "loss": 1.0236,
+      "step": 10393
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010723693022009911,
+      "loss": 1.1083,
+      "step": 10394
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010722800125005582,
+      "loss": 1.0594,
+      "step": 10395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001072190722800125,
+      "loss": 0.9927,
+      "step": 10396
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001072101433099692,
+      "loss": 0.9659,
+      "step": 10397
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010720121433992589,
+      "loss": 1.0242,
+      "step": 10398
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010719228536988259,
+      "loss": 0.9621,
+      "step": 10399
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010718335639983927,
+      "loss": 1.0636,
+      "step": 10400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010717442742979597,
+      "loss": 1.0088,
+      "step": 10401
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010716549845975268,
+      "loss": 1.1067,
+      "step": 10402
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010715656948970935,
+      "loss": 1.0829,
+      "step": 10403
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010714764051966606,
+      "loss": 1.0615,
+      "step": 10404
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010713871154962276,
+      "loss": 1.0563,
+      "step": 10405
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010712978257957944,
+      "loss": 1.0263,
+      "step": 10406
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010712085360953614,
+      "loss": 1.0402,
+      "step": 10407
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010711192463949285,
+      "loss": 1.0665,
+      "step": 10408
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010710299566944952,
+      "loss": 1.1541,
+      "step": 10409
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010709406669940623,
+      "loss": 0.9427,
+      "step": 10410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010708513772936293,
+      "loss": 1.1455,
+      "step": 10411
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010707620875931961,
+      "loss": 1.1006,
+      "step": 10412
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010706727978927631,
+      "loss": 0.9735,
+      "step": 10413
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010705835081923302,
+      "loss": 1.0054,
+      "step": 10414
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010704942184918969,
+      "loss": 1.0092,
+      "step": 10415
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001070404928791464,
+      "loss": 1.0426,
+      "step": 10416
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001070315639091031,
+      "loss": 1.1396,
+      "step": 10417
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010702263493905978,
+      "loss": 1.1368,
+      "step": 10418
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010701370596901648,
+      "loss": 1.0405,
+      "step": 10419
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010700477699897317,
+      "loss": 0.9352,
+      "step": 10420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010699584802892986,
+      "loss": 1.0457,
+      "step": 10421
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010698691905888655,
+      "loss": 0.964,
+      "step": 10422
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010697799008884326,
+      "loss": 1.0717,
+      "step": 10423
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010696906111879996,
+      "loss": 1.1101,
+      "step": 10424
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010696013214875664,
+      "loss": 1.0323,
+      "step": 10425
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010695120317871334,
+      "loss": 1.0463,
+      "step": 10426
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010694227420867005,
+      "loss": 0.9034,
+      "step": 10427
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010693334523862672,
+      "loss": 1.0673,
+      "step": 10428
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010692441626858343,
+      "loss": 1.0042,
+      "step": 10429
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010691548729854013,
+      "loss": 1.1029,
+      "step": 10430
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010690655832849681,
+      "loss": 1.0161,
+      "step": 10431
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010689762935845351,
+      "loss": 1.0478,
+      "step": 10432
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010688870038841022,
+      "loss": 1.0141,
+      "step": 10433
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010687977141836689,
+      "loss": 0.9066,
+      "step": 10434
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001068708424483236,
+      "loss": 1.0113,
+      "step": 10435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001068619134782803,
+      "loss": 0.9963,
+      "step": 10436
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010685298450823698,
+      "loss": 1.0065,
+      "step": 10437
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010684405553819368,
+      "loss": 0.9386,
+      "step": 10438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010683512656815037,
+      "loss": 1.0352,
+      "step": 10439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010682619759810706,
+      "loss": 1.0557,
+      "step": 10440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010681726862806377,
+      "loss": 0.997,
+      "step": 10441
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010680833965802046,
+      "loss": 0.9776,
+      "step": 10442
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010679941068797715,
+      "loss": 1.0431,
+      "step": 10443
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010679048171793384,
+      "loss": 0.9915,
+      "step": 10444
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010678155274789054,
+      "loss": 0.9887,
+      "step": 10445
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010677262377784722,
+      "loss": 1.0884,
+      "step": 10446
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010676369480780392,
+      "loss": 1.1195,
+      "step": 10447
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010675476583776063,
+      "loss": 1.1666,
+      "step": 10448
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001067458368677173,
+      "loss": 1.0037,
+      "step": 10449
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010673690789767401,
+      "loss": 1.0687,
+      "step": 10450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010672797892763071,
+      "loss": 0.9716,
+      "step": 10451
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010671904995758739,
+      "loss": 0.9833,
+      "step": 10452
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001067101209875441,
+      "loss": 0.9854,
+      "step": 10453
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001067011920175008,
+      "loss": 1.0145,
+      "step": 10454
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010669226304745747,
+      "loss": 1.1264,
+      "step": 10455
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010668333407741418,
+      "loss": 1.1146,
+      "step": 10456
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010667440510737088,
+      "loss": 1.0365,
+      "step": 10457
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010666547613732756,
+      "loss": 0.9866,
+      "step": 10458
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010665654716728426,
+      "loss": 1.0122,
+      "step": 10459
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010664761819724097,
+      "loss": 1.0852,
+      "step": 10460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010663868922719764,
+      "loss": 1.0882,
+      "step": 10461
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010662976025715435,
+      "loss": 1.0634,
+      "step": 10462
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010662083128711104,
+      "loss": 1.0802,
+      "step": 10463
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010661190231706773,
+      "loss": 1.1485,
+      "step": 10464
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010660297334702442,
+      "loss": 1.0462,
+      "step": 10465
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010659404437698112,
+      "loss": 0.9771,
+      "step": 10466
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010658511540693781,
+      "loss": 1.1203,
+      "step": 10467
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001065761864368945,
+      "loss": 0.9233,
+      "step": 10468
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010656725746685121,
+      "loss": 0.9106,
+      "step": 10469
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010655832849680789,
+      "loss": 1.1492,
+      "step": 10470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010654939952676459,
+      "loss": 0.9519,
+      "step": 10471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001065404705567213,
+      "loss": 0.9844,
+      "step": 10472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010653154158667797,
+      "loss": 0.9845,
+      "step": 10473
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010652261261663467,
+      "loss": 1.0043,
+      "step": 10474
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010651368364659138,
+      "loss": 1.1769,
+      "step": 10475
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010650475467654806,
+      "loss": 1.0263,
+      "step": 10476
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010649582570650476,
+      "loss": 1.0705,
+      "step": 10477
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010648689673646146,
+      "loss": 0.9546,
+      "step": 10478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010647796776641814,
+      "loss": 1.0585,
+      "step": 10479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010646903879637484,
+      "loss": 0.943,
+      "step": 10480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010646010982633155,
+      "loss": 0.8203,
+      "step": 10481
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010645118085628823,
+      "loss": 1.0718,
+      "step": 10482
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010644225188624493,
+      "loss": 1.1198,
+      "step": 10483
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010643332291620163,
+      "loss": 0.9885,
+      "step": 10484
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010642439394615831,
+      "loss": 0.9602,
+      "step": 10485
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010641546497611501,
+      "loss": 1.1374,
+      "step": 10486
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001064065360060717,
+      "loss": 1.048,
+      "step": 10487
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001063976070360284,
+      "loss": 0.9228,
+      "step": 10488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010638867806598509,
+      "loss": 1.0038,
+      "step": 10489
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010637974909594179,
+      "loss": 1.0937,
+      "step": 10490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010637082012589847,
+      "loss": 1.1115,
+      "step": 10491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010636189115585517,
+      "loss": 1.0503,
+      "step": 10492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010635296218581188,
+      "loss": 1.117,
+      "step": 10493
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010634403321576855,
+      "loss": 1.344,
+      "step": 10494
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010633510424572526,
+      "loss": 0.9874,
+      "step": 10495
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010632617527568196,
+      "loss": 0.944,
+      "step": 10496
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010631724630563864,
+      "loss": 0.972,
+      "step": 10497
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010630831733559534,
+      "loss": 1.0773,
+      "step": 10498
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010629938836555205,
+      "loss": 0.9606,
+      "step": 10499
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010629045939550872,
+      "loss": 0.9506,
+      "step": 10500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010628153042546543,
+      "loss": 0.9673,
+      "step": 10501
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010627260145542213,
+      "loss": 0.947,
+      "step": 10502
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010626367248537881,
+      "loss": 0.9862,
+      "step": 10503
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010625474351533551,
+      "loss": 1.0779,
+      "step": 10504
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010624581454529222,
+      "loss": 1.108,
+      "step": 10505
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010623688557524889,
+      "loss": 1.0258,
+      "step": 10506
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001062279566052056,
+      "loss": 1.1428,
+      "step": 10507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001062190276351623,
+      "loss": 0.9915,
+      "step": 10508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010621009866511898,
+      "loss": 0.9081,
+      "step": 10509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010620116969507568,
+      "loss": 1.0894,
+      "step": 10510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010619224072503237,
+      "loss": 1.0628,
+      "step": 10511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010618331175498906,
+      "loss": 0.9662,
+      "step": 10512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010617438278494575,
+      "loss": 1.0297,
+      "step": 10513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010616545381490246,
+      "loss": 1.0532,
+      "step": 10514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010615652484485913,
+      "loss": 1.2587,
+      "step": 10515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010614759587481584,
+      "loss": 1.0173,
+      "step": 10516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010613866690477254,
+      "loss": 1.0177,
+      "step": 10517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010612973793472922,
+      "loss": 1.0142,
+      "step": 10518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010612080896468592,
+      "loss": 0.9462,
+      "step": 10519
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010611187999464263,
+      "loss": 1.0608,
+      "step": 10520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010610295102459933,
+      "loss": 1.0588,
+      "step": 10521
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010609402205455601,
+      "loss": 1.0884,
+      "step": 10522
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010608509308451271,
+      "loss": 0.9776,
+      "step": 10523
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010607616411446942,
+      "loss": 1.0001,
+      "step": 10524
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010606723514442609,
+      "loss": 1.141,
+      "step": 10525
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001060583061743828,
+      "loss": 1.0185,
+      "step": 10526
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001060493772043395,
+      "loss": 1.0135,
+      "step": 10527
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010604044823429618,
+      "loss": 1.0073,
+      "step": 10528
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010603151926425288,
+      "loss": 0.9888,
+      "step": 10529
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010602259029420957,
+      "loss": 1.0713,
+      "step": 10530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010601366132416626,
+      "loss": 1.053,
+      "step": 10531
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010600473235412295,
+      "loss": 1.1144,
+      "step": 10532
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010599580338407966,
+      "loss": 1.0714,
+      "step": 10533
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010598687441403635,
+      "loss": 1.0044,
+      "step": 10534
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010597794544399304,
+      "loss": 0.9956,
+      "step": 10535
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010596901647394974,
+      "loss": 1.053,
+      "step": 10536
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010596008750390642,
+      "loss": 1.1711,
+      "step": 10537
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010595115853386312,
+      "loss": 1.0804,
+      "step": 10538
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010594222956381983,
+      "loss": 1.0138,
+      "step": 10539
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001059333005937765,
+      "loss": 0.9211,
+      "step": 10540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010592437162373321,
+      "loss": 0.9787,
+      "step": 10541
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010591544265368991,
+      "loss": 0.9613,
+      "step": 10542
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010590651368364659,
+      "loss": 0.9683,
+      "step": 10543
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010589758471360329,
+      "loss": 1.0802,
+      "step": 10544
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010588865574356,
+      "loss": 0.9993,
+      "step": 10545
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010587972677351667,
+      "loss": 0.9232,
+      "step": 10546
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010587079780347338,
+      "loss": 0.7785,
+      "step": 10547
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010586186883343008,
+      "loss": 0.9872,
+      "step": 10548
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010585293986338676,
+      "loss": 1.0418,
+      "step": 10549
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010584401089334346,
+      "loss": 0.9843,
+      "step": 10550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010583508192330017,
+      "loss": 0.9341,
+      "step": 10551
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010582615295325684,
+      "loss": 1.0489,
+      "step": 10552
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010581722398321355,
+      "loss": 0.9607,
+      "step": 10553
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010580829501317024,
+      "loss": 1.0358,
+      "step": 10554
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010579936604312693,
+      "loss": 1.1708,
+      "step": 10555
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010579043707308362,
+      "loss": 0.9918,
+      "step": 10556
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010578150810304032,
+      "loss": 1.0554,
+      "step": 10557
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010577257913299701,
+      "loss": 1.1328,
+      "step": 10558
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001057636501629537,
+      "loss": 1.1466,
+      "step": 10559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010575472119291041,
+      "loss": 1.1306,
+      "step": 10560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010574579222286709,
+      "loss": 1.0809,
+      "step": 10561
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010573686325282379,
+      "loss": 1.0972,
+      "step": 10562
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001057279342827805,
+      "loss": 0.9871,
+      "step": 10563
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010571900531273717,
+      "loss": 0.9948,
+      "step": 10564
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010571007634269387,
+      "loss": 1.0293,
+      "step": 10565
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010570114737265058,
+      "loss": 1.0768,
+      "step": 10566
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010569221840260726,
+      "loss": 0.9096,
+      "step": 10567
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010568328943256396,
+      "loss": 0.9886,
+      "step": 10568
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010567436046252066,
+      "loss": 1.1447,
+      "step": 10569
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010566543149247734,
+      "loss": 1.0186,
+      "step": 10570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010565650252243404,
+      "loss": 1.081,
+      "step": 10571
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010564757355239075,
+      "loss": 0.9767,
+      "step": 10572
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010563864458234743,
+      "loss": 0.998,
+      "step": 10573
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010562971561230413,
+      "loss": 1.0058,
+      "step": 10574
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010562078664226083,
+      "loss": 0.9107,
+      "step": 10575
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010561185767221751,
+      "loss": 0.9853,
+      "step": 10576
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010560292870217421,
+      "loss": 1.0388,
+      "step": 10577
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001055939997321309,
+      "loss": 1.0352,
+      "step": 10578
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001055850707620876,
+      "loss": 0.9493,
+      "step": 10579
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010557614179204429,
+      "loss": 1.1816,
+      "step": 10580
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010556721282200099,
+      "loss": 1.1304,
+      "step": 10581
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010555828385195767,
+      "loss": 1.0876,
+      "step": 10582
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010554935488191437,
+      "loss": 1.0093,
+      "step": 10583
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010554042591187107,
+      "loss": 1.2915,
+      "step": 10584
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010553149694182775,
+      "loss": 0.923,
+      "step": 10585
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010552256797178446,
+      "loss": 1.1407,
+      "step": 10586
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010551363900174116,
+      "loss": 1.1996,
+      "step": 10587
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010550471003169784,
+      "loss": 1.1268,
+      "step": 10588
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010549578106165454,
+      "loss": 1.0851,
+      "step": 10589
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010548685209161124,
+      "loss": 1.1208,
+      "step": 10590
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010547792312156792,
+      "loss": 1.0202,
+      "step": 10591
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010546899415152463,
+      "loss": 1.0564,
+      "step": 10592
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010546006518148133,
+      "loss": 1.1905,
+      "step": 10593
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000105451136211438,
+      "loss": 0.9144,
+      "step": 10594
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010544220724139471,
+      "loss": 1.1572,
+      "step": 10595
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010543327827135141,
+      "loss": 0.9645,
+      "step": 10596
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010542434930130809,
+      "loss": 1.052,
+      "step": 10597
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001054154203312648,
+      "loss": 1.0177,
+      "step": 10598
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001054064913612215,
+      "loss": 1.0629,
+      "step": 10599
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010539756239117818,
+      "loss": 1.0974,
+      "step": 10600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010538863342113488,
+      "loss": 0.991,
+      "step": 10601
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010537970445109157,
+      "loss": 1.0593,
+      "step": 10602
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010537077548104826,
+      "loss": 1.0681,
+      "step": 10603
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010536184651100495,
+      "loss": 1.1262,
+      "step": 10604
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010535291754096166,
+      "loss": 1.2016,
+      "step": 10605
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010534398857091833,
+      "loss": 1.059,
+      "step": 10606
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010533505960087504,
+      "loss": 1.0123,
+      "step": 10607
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010532613063083174,
+      "loss": 1.0973,
+      "step": 10608
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010531720166078842,
+      "loss": 1.0871,
+      "step": 10609
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010530827269074512,
+      "loss": 1.0638,
+      "step": 10610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010529934372070183,
+      "loss": 0.8582,
+      "step": 10611
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001052904147506585,
+      "loss": 1.0754,
+      "step": 10612
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010528148578061521,
+      "loss": 0.9864,
+      "step": 10613
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010527255681057191,
+      "loss": 1.1686,
+      "step": 10614
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010526362784052859,
+      "loss": 1.1471,
+      "step": 10615
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010525469887048529,
+      "loss": 1.0552,
+      "step": 10616
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000105245769900442,
+      "loss": 0.9963,
+      "step": 10617
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001052368409303987,
+      "loss": 0.9861,
+      "step": 10618
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010522791196035538,
+      "loss": 1.0279,
+      "step": 10619
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010521898299031208,
+      "loss": 0.9136,
+      "step": 10620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010521005402026877,
+      "loss": 1.0358,
+      "step": 10621
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010520112505022546,
+      "loss": 1.1268,
+      "step": 10622
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010519219608018215,
+      "loss": 1.1763,
+      "step": 10623
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010518326711013886,
+      "loss": 1.0656,
+      "step": 10624
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010517433814009555,
+      "loss": 0.98,
+      "step": 10625
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010516540917005224,
+      "loss": 0.9638,
+      "step": 10626
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010515648020000894,
+      "loss": 0.9943,
+      "step": 10627
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010514755122996562,
+      "loss": 1.0702,
+      "step": 10628
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010513862225992232,
+      "loss": 0.9123,
+      "step": 10629
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010512969328987903,
+      "loss": 0.9596,
+      "step": 10630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001051207643198357,
+      "loss": 1.0651,
+      "step": 10631
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010511183534979241,
+      "loss": 1.0554,
+      "step": 10632
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010510290637974911,
+      "loss": 1.0009,
+      "step": 10633
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010509397740970579,
+      "loss": 0.9637,
+      "step": 10634
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010508504843966249,
+      "loss": 0.928,
+      "step": 10635
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001050761194696192,
+      "loss": 1.0348,
+      "step": 10636
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010506719049957587,
+      "loss": 1.0074,
+      "step": 10637
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010505826152953258,
+      "loss": 1.0509,
+      "step": 10638
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010504933255948928,
+      "loss": 0.9997,
+      "step": 10639
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010504040358944596,
+      "loss": 1.0053,
+      "step": 10640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010503147461940266,
+      "loss": 0.9428,
+      "step": 10641
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010502254564935937,
+      "loss": 1.0457,
+      "step": 10642
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010501361667931604,
+      "loss": 0.9965,
+      "step": 10643
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010500468770927275,
+      "loss": 0.9783,
+      "step": 10644
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010499575873922944,
+      "loss": 1.0045,
+      "step": 10645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010498682976918613,
+      "loss": 0.9442,
+      "step": 10646
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010497790079914282,
+      "loss": 0.9759,
+      "step": 10647
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010496897182909952,
+      "loss": 1.0229,
+      "step": 10648
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010496004285905621,
+      "loss": 0.9967,
+      "step": 10649
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001049511138890129,
+      "loss": 1.0022,
+      "step": 10650
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010494218491896961,
+      "loss": 1.0778,
+      "step": 10651
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010493325594892629,
+      "loss": 1.0622,
+      "step": 10652
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010492432697888299,
+      "loss": 1.0104,
+      "step": 10653
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010491539800883969,
+      "loss": 1.0777,
+      "step": 10654
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010490646903879637,
+      "loss": 1.1756,
+      "step": 10655
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010489754006875307,
+      "loss": 1.0556,
+      "step": 10656
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010488861109870978,
+      "loss": 0.9229,
+      "step": 10657
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010487968212866645,
+      "loss": 1.0002,
+      "step": 10658
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010487075315862316,
+      "loss": 1.0801,
+      "step": 10659
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010486182418857986,
+      "loss": 0.969,
+      "step": 10660
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010485289521853654,
+      "loss": 1.0293,
+      "step": 10661
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010484396624849324,
+      "loss": 1.0496,
+      "step": 10662
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010483503727844995,
+      "loss": 1.0406,
+      "step": 10663
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010482610830840662,
+      "loss": 1.1203,
+      "step": 10664
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010481717933836333,
+      "loss": 1.0708,
+      "step": 10665
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010480825036832003,
+      "loss": 0.9143,
+      "step": 10666
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010479932139827671,
+      "loss": 1.0,
+      "step": 10667
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010479039242823341,
+      "loss": 0.9774,
+      "step": 10668
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001047814634581901,
+      "loss": 1.1231,
+      "step": 10669
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001047725344881468,
+      "loss": 1.035,
+      "step": 10670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010476360551810349,
+      "loss": 0.9595,
+      "step": 10671
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010475467654806019,
+      "loss": 0.9558,
+      "step": 10672
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010474574757801687,
+      "loss": 0.9527,
+      "step": 10673
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010473681860797357,
+      "loss": 1.1181,
+      "step": 10674
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010472788963793027,
+      "loss": 1.0456,
+      "step": 10675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010471896066788695,
+      "loss": 1.073,
+      "step": 10676
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010471003169784366,
+      "loss": 1.0293,
+      "step": 10677
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010470110272780036,
+      "loss": 0.9654,
+      "step": 10678
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010469217375775704,
+      "loss": 1.1074,
+      "step": 10679
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010468324478771374,
+      "loss": 1.0185,
+      "step": 10680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010467431581767044,
+      "loss": 0.9978,
+      "step": 10681
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010466538684762712,
+      "loss": 0.8734,
+      "step": 10682
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010465645787758383,
+      "loss": 1.0754,
+      "step": 10683
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010464752890754053,
+      "loss": 0.9888,
+      "step": 10684
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001046385999374972,
+      "loss": 1.0414,
+      "step": 10685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010462967096745391,
+      "loss": 0.9705,
+      "step": 10686
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010462074199741061,
+      "loss": 1.103,
+      "step": 10687
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010461181302736729,
+      "loss": 1.1102,
+      "step": 10688
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000104602884057324,
+      "loss": 0.9348,
+      "step": 10689
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001045939550872807,
+      "loss": 0.9401,
+      "step": 10690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010458502611723738,
+      "loss": 0.9511,
+      "step": 10691
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010457609714719408,
+      "loss": 1.0054,
+      "step": 10692
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010456716817715077,
+      "loss": 1.031,
+      "step": 10693
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010455823920710746,
+      "loss": 0.9567,
+      "step": 10694
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010454931023706415,
+      "loss": 0.8252,
+      "step": 10695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010454038126702086,
+      "loss": 0.9774,
+      "step": 10696
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010453145229697753,
+      "loss": 1.0116,
+      "step": 10697
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010452252332693424,
+      "loss": 1.1104,
+      "step": 10698
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010451359435689094,
+      "loss": 1.019,
+      "step": 10699
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010450466538684762,
+      "loss": 0.9894,
+      "step": 10700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010449573641680432,
+      "loss": 0.9328,
+      "step": 10701
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010448680744676103,
+      "loss": 0.9796,
+      "step": 10702
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001044778784767177,
+      "loss": 1.095,
+      "step": 10703
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010446894950667441,
+      "loss": 1.1049,
+      "step": 10704
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010446002053663111,
+      "loss": 1.0983,
+      "step": 10705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010445109156658779,
+      "loss": 1.1821,
+      "step": 10706
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010444216259654449,
+      "loss": 0.8925,
+      "step": 10707
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001044332336265012,
+      "loss": 0.9301,
+      "step": 10708
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010442430465645787,
+      "loss": 1.0865,
+      "step": 10709
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010441537568641458,
+      "loss": 0.9812,
+      "step": 10710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010440644671637128,
+      "loss": 0.9518,
+      "step": 10711
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010439751774632796,
+      "loss": 0.9819,
+      "step": 10712
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010438858877628466,
+      "loss": 1.0263,
+      "step": 10713
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010437965980624135,
+      "loss": 0.989,
+      "step": 10714
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010437073083619804,
+      "loss": 1.1415,
+      "step": 10715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010436180186615475,
+      "loss": 1.0865,
+      "step": 10716
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010435287289611144,
+      "loss": 0.9936,
+      "step": 10717
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010434394392606814,
+      "loss": 0.9245,
+      "step": 10718
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010433501495602482,
+      "loss": 1.0802,
+      "step": 10719
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010432608598598152,
+      "loss": 0.8724,
+      "step": 10720
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010431715701593823,
+      "loss": 1.0386,
+      "step": 10721
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001043082280458949,
+      "loss": 1.0147,
+      "step": 10722
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010429929907585161,
+      "loss": 0.951,
+      "step": 10723
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010429037010580831,
+      "loss": 1.1092,
+      "step": 10724
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010428144113576499,
+      "loss": 1.0645,
+      "step": 10725
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010427251216572169,
+      "loss": 1.1255,
+      "step": 10726
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001042635831956784,
+      "loss": 1.1092,
+      "step": 10727
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010425465422563507,
+      "loss": 1.0909,
+      "step": 10728
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010424572525559178,
+      "loss": 1.0498,
+      "step": 10729
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010423679628554848,
+      "loss": 1.0028,
+      "step": 10730
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010422786731550516,
+      "loss": 1.0458,
+      "step": 10731
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010421893834546186,
+      "loss": 1.007,
+      "step": 10732
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010421000937541857,
+      "loss": 1.1291,
+      "step": 10733
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010420108040537524,
+      "loss": 1.0859,
+      "step": 10734
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010419215143533195,
+      "loss": 1.0091,
+      "step": 10735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010418322246528864,
+      "loss": 0.9507,
+      "step": 10736
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010417429349524533,
+      "loss": 0.9798,
+      "step": 10737
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010416536452520202,
+      "loss": 1.1773,
+      "step": 10738
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010415643555515872,
+      "loss": 0.9233,
+      "step": 10739
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001041475065851154,
+      "loss": 0.9587,
+      "step": 10740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001041385776150721,
+      "loss": 1.0882,
+      "step": 10741
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010412964864502881,
+      "loss": 1.0332,
+      "step": 10742
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010412071967498548,
+      "loss": 0.9589,
+      "step": 10743
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010411179070494219,
+      "loss": 1.2575,
+      "step": 10744
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010410286173489889,
+      "loss": 1.055,
+      "step": 10745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010409393276485557,
+      "loss": 0.9327,
+      "step": 10746
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010408500379481227,
+      "loss": 0.8912,
+      "step": 10747
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010407607482476898,
+      "loss": 0.9954,
+      "step": 10748
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010406714585472565,
+      "loss": 0.9438,
+      "step": 10749
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010405821688468236,
+      "loss": 0.8739,
+      "step": 10750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010404928791463906,
+      "loss": 1.0056,
+      "step": 10751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010404035894459574,
+      "loss": 0.9562,
+      "step": 10752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010403142997455244,
+      "loss": 1.0056,
+      "step": 10753
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010402250100450915,
+      "loss": 0.9456,
+      "step": 10754
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010401357203446582,
+      "loss": 1.0837,
+      "step": 10755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010400464306442253,
+      "loss": 0.9198,
+      "step": 10756
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010399571409437923,
+      "loss": 1.1213,
+      "step": 10757
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010398678512433591,
+      "loss": 0.9939,
+      "step": 10758
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010397785615429261,
+      "loss": 0.9388,
+      "step": 10759
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001039689271842493,
+      "loss": 1.1739,
+      "step": 10760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000103959998214206,
+      "loss": 1.087,
+      "step": 10761
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010395106924416269,
+      "loss": 0.9961,
+      "step": 10762
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010394214027411939,
+      "loss": 1.1055,
+      "step": 10763
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010393321130407607,
+      "loss": 1.0805,
+      "step": 10764
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010392428233403277,
+      "loss": 0.9658,
+      "step": 10765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010391535336398947,
+      "loss": 0.9097,
+      "step": 10766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010390642439394615,
+      "loss": 0.9591,
+      "step": 10767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010389749542390286,
+      "loss": 1.0041,
+      "step": 10768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010388856645385956,
+      "loss": 1.0344,
+      "step": 10769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010387963748381624,
+      "loss": 1.0513,
+      "step": 10770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010387070851377294,
+      "loss": 0.984,
+      "step": 10771
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010386177954372964,
+      "loss": 1.0818,
+      "step": 10772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010385285057368632,
+      "loss": 1.0982,
+      "step": 10773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010384392160364303,
+      "loss": 0.935,
+      "step": 10774
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010383499263359973,
+      "loss": 1.0968,
+      "step": 10775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001038260636635564,
+      "loss": 1.0431,
+      "step": 10776
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010381713469351311,
+      "loss": 1.0707,
+      "step": 10777
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010380820572346981,
+      "loss": 0.966,
+      "step": 10778
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010379927675342649,
+      "loss": 1.2837,
+      "step": 10779
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001037903477833832,
+      "loss": 1.1124,
+      "step": 10780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001037814188133399,
+      "loss": 0.865,
+      "step": 10781
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010377248984329658,
+      "loss": 0.9656,
+      "step": 10782
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010376356087325328,
+      "loss": 1.0404,
+      "step": 10783
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010375463190320997,
+      "loss": 0.956,
+      "step": 10784
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010374570293316666,
+      "loss": 0.9911,
+      "step": 10785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010373677396312335,
+      "loss": 1.1152,
+      "step": 10786
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010372784499308006,
+      "loss": 0.9146,
+      "step": 10787
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010371891602303673,
+      "loss": 1.0904,
+      "step": 10788
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010370998705299344,
+      "loss": 1.0781,
+      "step": 10789
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010370105808295014,
+      "loss": 1.052,
+      "step": 10790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010369212911290682,
+      "loss": 1.1039,
+      "step": 10791
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010368320014286352,
+      "loss": 1.054,
+      "step": 10792
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010367427117282023,
+      "loss": 1.0663,
+      "step": 10793
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001036653422027769,
+      "loss": 1.0038,
+      "step": 10794
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001036564132327336,
+      "loss": 1.0122,
+      "step": 10795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010364748426269031,
+      "loss": 0.9875,
+      "step": 10796
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010363855529264699,
+      "loss": 1.1062,
+      "step": 10797
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010362962632260369,
+      "loss": 1.0103,
+      "step": 10798
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001036206973525604,
+      "loss": 1.0415,
+      "step": 10799
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010361176838251707,
+      "loss": 1.0569,
+      "step": 10800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010360283941247378,
+      "loss": 0.9675,
+      "step": 10801
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010359391044243048,
+      "loss": 0.9889,
+      "step": 10802
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010358498147238716,
+      "loss": 1.0458,
+      "step": 10803
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010357605250234386,
+      "loss": 0.9754,
+      "step": 10804
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010356712353230055,
+      "loss": 1.0144,
+      "step": 10805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010355819456225724,
+      "loss": 1.0353,
+      "step": 10806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010354926559221395,
+      "loss": 0.9546,
+      "step": 10807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010354033662217064,
+      "loss": 1.0522,
+      "step": 10808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010353140765212733,
+      "loss": 1.0153,
+      "step": 10809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010352247868208402,
+      "loss": 1.2269,
+      "step": 10810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010351354971204072,
+      "loss": 0.9346,
+      "step": 10811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001035046207419974,
+      "loss": 1.1353,
+      "step": 10812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001034956917719541,
+      "loss": 0.9433,
+      "step": 10813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010348676280191081,
+      "loss": 0.9426,
+      "step": 10814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010347783383186751,
+      "loss": 0.9607,
+      "step": 10815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010346890486182419,
+      "loss": 1.081,
+      "step": 10816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010345997589178089,
+      "loss": 1.0601,
+      "step": 10817
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001034510469217376,
+      "loss": 1.1574,
+      "step": 10818
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010344211795169427,
+      "loss": 0.974,
+      "step": 10819
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010343318898165098,
+      "loss": 1.0671,
+      "step": 10820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010342426001160768,
+      "loss": 1.1709,
+      "step": 10821
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010341533104156436,
+      "loss": 1.0736,
+      "step": 10822
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010340640207152106,
+      "loss": 1.0248,
+      "step": 10823
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010339747310147777,
+      "loss": 1.1764,
+      "step": 10824
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010338854413143444,
+      "loss": 0.9637,
+      "step": 10825
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010337961516139115,
+      "loss": 0.9429,
+      "step": 10826
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010337068619134784,
+      "loss": 0.9804,
+      "step": 10827
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010336175722130453,
+      "loss": 0.9721,
+      "step": 10828
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010335282825126122,
+      "loss": 0.9852,
+      "step": 10829
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010334389928121792,
+      "loss": 1.0083,
+      "step": 10830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001033349703111746,
+      "loss": 0.9596,
+      "step": 10831
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001033260413411313,
+      "loss": 1.0798,
+      "step": 10832
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010331711237108801,
+      "loss": 1.1618,
+      "step": 10833
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010330818340104468,
+      "loss": 1.0453,
+      "step": 10834
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010329925443100139,
+      "loss": 1.0068,
+      "step": 10835
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010329032546095809,
+      "loss": 1.0233,
+      "step": 10836
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010328139649091477,
+      "loss": 1.0597,
+      "step": 10837
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010327246752087147,
+      "loss": 0.9938,
+      "step": 10838
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010326353855082818,
+      "loss": 1.0316,
+      "step": 10839
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010325460958078485,
+      "loss": 1.0306,
+      "step": 10840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010324568061074156,
+      "loss": 1.1483,
+      "step": 10841
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010323675164069826,
+      "loss": 1.1718,
+      "step": 10842
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010322782267065494,
+      "loss": 0.9337,
+      "step": 10843
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010321889370061164,
+      "loss": 1.0337,
+      "step": 10844
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010320996473056835,
+      "loss": 0.9422,
+      "step": 10845
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010320103576052502,
+      "loss": 1.0549,
+      "step": 10846
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010319210679048173,
+      "loss": 1.1445,
+      "step": 10847
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010318317782043843,
+      "loss": 1.0161,
+      "step": 10848
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010317424885039511,
+      "loss": 1.0356,
+      "step": 10849
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010316531988035181,
+      "loss": 1.0555,
+      "step": 10850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001031563909103085,
+      "loss": 1.1387,
+      "step": 10851
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001031474619402652,
+      "loss": 0.9669,
+      "step": 10852
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010313853297022188,
+      "loss": 1.0827,
+      "step": 10853
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010312960400017859,
+      "loss": 0.9539,
+      "step": 10854
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010312067503013527,
+      "loss": 0.9854,
+      "step": 10855
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010311174606009197,
+      "loss": 1.0081,
+      "step": 10856
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010310281709004867,
+      "loss": 0.9587,
+      "step": 10857
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010309388812000535,
+      "loss": 1.0201,
+      "step": 10858
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010308495914996205,
+      "loss": 1.0463,
+      "step": 10859
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010307603017991876,
+      "loss": 1.0194,
+      "step": 10860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010306710120987544,
+      "loss": 1.1286,
+      "step": 10861
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010305817223983214,
+      "loss": 1.0133,
+      "step": 10862
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010304924326978884,
+      "loss": 0.9853,
+      "step": 10863
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010304031429974552,
+      "loss": 1.0248,
+      "step": 10864
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010303138532970222,
+      "loss": 0.926,
+      "step": 10865
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010302245635965893,
+      "loss": 0.9816,
+      "step": 10866
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001030135273896156,
+      "loss": 0.9965,
+      "step": 10867
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010300459841957231,
+      "loss": 1.041,
+      "step": 10868
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010299566944952901,
+      "loss": 0.967,
+      "step": 10869
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010298674047948569,
+      "loss": 1.155,
+      "step": 10870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001029778115094424,
+      "loss": 0.9398,
+      "step": 10871
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010296888253939909,
+      "loss": 1.028,
+      "step": 10872
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010295995356935578,
+      "loss": 1.1191,
+      "step": 10873
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010295102459931248,
+      "loss": 1.2717,
+      "step": 10874
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010294209562926917,
+      "loss": 0.9754,
+      "step": 10875
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010293316665922586,
+      "loss": 0.9911,
+      "step": 10876
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010292423768918255,
+      "loss": 1.054,
+      "step": 10877
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010291530871913926,
+      "loss": 1.1082,
+      "step": 10878
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010290637974909593,
+      "loss": 0.9905,
+      "step": 10879
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010289745077905264,
+      "loss": 1.1233,
+      "step": 10880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010288852180900934,
+      "loss": 1.0376,
+      "step": 10881
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010287959283896602,
+      "loss": 1.1151,
+      "step": 10882
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010287066386892272,
+      "loss": 0.9783,
+      "step": 10883
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010286173489887943,
+      "loss": 1.0796,
+      "step": 10884
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001028528059288361,
+      "loss": 1.0607,
+      "step": 10885
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001028438769587928,
+      "loss": 1.0239,
+      "step": 10886
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010283494798874951,
+      "loss": 1.0215,
+      "step": 10887
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010282601901870619,
+      "loss": 0.9399,
+      "step": 10888
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010281709004866289,
+      "loss": 1.0979,
+      "step": 10889
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001028081610786196,
+      "loss": 1.2611,
+      "step": 10890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010279923210857627,
+      "loss": 1.0272,
+      "step": 10891
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010279030313853298,
+      "loss": 1.1397,
+      "step": 10892
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010278137416848968,
+      "loss": 0.982,
+      "step": 10893
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010277244519844636,
+      "loss": 0.8982,
+      "step": 10894
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010276351622840306,
+      "loss": 1.1119,
+      "step": 10895
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010275458725835975,
+      "loss": 1.1093,
+      "step": 10896
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010274565828831644,
+      "loss": 1.0259,
+      "step": 10897
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010273672931827315,
+      "loss": 0.9555,
+      "step": 10898
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010272780034822984,
+      "loss": 1.0495,
+      "step": 10899
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010271887137818653,
+      "loss": 1.0023,
+      "step": 10900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010270994240814322,
+      "loss": 1.1237,
+      "step": 10901
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010270101343809992,
+      "loss": 0.9762,
+      "step": 10902
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001026920844680566,
+      "loss": 1.0059,
+      "step": 10903
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001026831554980133,
+      "loss": 1.2221,
+      "step": 10904
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010267422652797,
+      "loss": 1.0695,
+      "step": 10905
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010266529755792668,
+      "loss": 1.038,
+      "step": 10906
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010265636858788339,
+      "loss": 0.9359,
+      "step": 10907
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010264743961784009,
+      "loss": 1.0217,
+      "step": 10908
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010263851064779677,
+      "loss": 1.012,
+      "step": 10909
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010262958167775347,
+      "loss": 0.9963,
+      "step": 10910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010262065270771018,
+      "loss": 0.9902,
+      "step": 10911
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010261172373766688,
+      "loss": 0.9481,
+      "step": 10912
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010260279476762356,
+      "loss": 0.8858,
+      "step": 10913
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010259386579758026,
+      "loss": 1.0526,
+      "step": 10914
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010258493682753697,
+      "loss": 0.8557,
+      "step": 10915
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010257600785749364,
+      "loss": 0.9969,
+      "step": 10916
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010256707888745035,
+      "loss": 1.0009,
+      "step": 10917
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010255814991740704,
+      "loss": 1.016,
+      "step": 10918
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010254922094736373,
+      "loss": 1.0215,
+      "step": 10919
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010254029197732042,
+      "loss": 1.1128,
+      "step": 10920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010253136300727712,
+      "loss": 0.9398,
+      "step": 10921
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001025224340372338,
+      "loss": 0.9493,
+      "step": 10922
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001025135050671905,
+      "loss": 0.9612,
+      "step": 10923
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010250457609714721,
+      "loss": 1.0441,
+      "step": 10924
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010249564712710388,
+      "loss": 0.9785,
+      "step": 10925
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010248671815706059,
+      "loss": 0.9679,
+      "step": 10926
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010247778918701729,
+      "loss": 1.0086,
+      "step": 10927
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010246886021697397,
+      "loss": 1.0704,
+      "step": 10928
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010245993124693067,
+      "loss": 0.9944,
+      "step": 10929
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010245100227688738,
+      "loss": 1.0532,
+      "step": 10930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010244207330684405,
+      "loss": 0.9576,
+      "step": 10931
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010243314433680076,
+      "loss": 1.0224,
+      "step": 10932
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010242421536675746,
+      "loss": 1.1302,
+      "step": 10933
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010241528639671414,
+      "loss": 1.1405,
+      "step": 10934
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010240635742667084,
+      "loss": 1.0241,
+      "step": 10935
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010239742845662755,
+      "loss": 1.0339,
+      "step": 10936
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010238849948658422,
+      "loss": 1.0862,
+      "step": 10937
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010237957051654093,
+      "loss": 1.0505,
+      "step": 10938
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010237064154649763,
+      "loss": 1.0236,
+      "step": 10939
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010236171257645431,
+      "loss": 1.0608,
+      "step": 10940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010235278360641101,
+      "loss": 1.0554,
+      "step": 10941
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001023438546363677,
+      "loss": 0.9791,
+      "step": 10942
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001023349256663244,
+      "loss": 1.0097,
+      "step": 10943
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010232599669628108,
+      "loss": 1.0299,
+      "step": 10944
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010231706772623779,
+      "loss": 1.1725,
+      "step": 10945
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010230813875619447,
+      "loss": 1.1164,
+      "step": 10946
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010229920978615117,
+      "loss": 0.9849,
+      "step": 10947
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010229028081610787,
+      "loss": 1.0082,
+      "step": 10948
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010228135184606455,
+      "loss": 0.988,
+      "step": 10949
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010227242287602125,
+      "loss": 1.1311,
+      "step": 10950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010226349390597796,
+      "loss": 1.0083,
+      "step": 10951
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010225456493593464,
+      "loss": 1.0226,
+      "step": 10952
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010224563596589134,
+      "loss": 0.9058,
+      "step": 10953
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010223670699584804,
+      "loss": 1.0129,
+      "step": 10954
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010222777802580472,
+      "loss": 1.0479,
+      "step": 10955
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010221884905576142,
+      "loss": 0.9383,
+      "step": 10956
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010220992008571813,
+      "loss": 1.0471,
+      "step": 10957
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001022009911156748,
+      "loss": 0.9948,
+      "step": 10958
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010219206214563151,
+      "loss": 1.0047,
+      "step": 10959
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010218313317558821,
+      "loss": 1.0535,
+      "step": 10960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010217420420554489,
+      "loss": 1.066,
+      "step": 10961
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001021652752355016,
+      "loss": 0.972,
+      "step": 10962
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010215634626545828,
+      "loss": 1.0168,
+      "step": 10963
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010214741729541498,
+      "loss": 0.9733,
+      "step": 10964
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010213848832537168,
+      "loss": 1.0791,
+      "step": 10965
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010212955935532837,
+      "loss": 1.0974,
+      "step": 10966
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010212063038528506,
+      "loss": 0.9483,
+      "step": 10967
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010211170141524175,
+      "loss": 1.0791,
+      "step": 10968
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010210277244519845,
+      "loss": 1.0372,
+      "step": 10969
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010209384347515513,
+      "loss": 0.9151,
+      "step": 10970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010208491450511184,
+      "loss": 0.9451,
+      "step": 10971
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010207598553506854,
+      "loss": 1.0466,
+      "step": 10972
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010206705656502522,
+      "loss": 1.2093,
+      "step": 10973
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010205812759498192,
+      "loss": 1.0322,
+      "step": 10974
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010204919862493862,
+      "loss": 0.9241,
+      "step": 10975
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001020402696548953,
+      "loss": 0.8921,
+      "step": 10976
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000102031340684852,
+      "loss": 1.0746,
+      "step": 10977
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010202241171480871,
+      "loss": 1.0442,
+      "step": 10978
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010201348274476539,
+      "loss": 0.9748,
+      "step": 10979
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010200455377472209,
+      "loss": 0.8198,
+      "step": 10980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001019956248046788,
+      "loss": 0.9092,
+      "step": 10981
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010198669583463547,
+      "loss": 0.992,
+      "step": 10982
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010197776686459218,
+      "loss": 0.8958,
+      "step": 10983
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010196883789454888,
+      "loss": 1.0254,
+      "step": 10984
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010195990892450556,
+      "loss": 0.9913,
+      "step": 10985
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010195097995446226,
+      "loss": 0.9923,
+      "step": 10986
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010194205098441895,
+      "loss": 1.0722,
+      "step": 10987
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010193312201437564,
+      "loss": 1.012,
+      "step": 10988
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010192419304433235,
+      "loss": 0.945,
+      "step": 10989
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010191526407428904,
+      "loss": 1.1528,
+      "step": 10990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010190633510424573,
+      "loss": 0.9389,
+      "step": 10991
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010189740613420242,
+      "loss": 0.9228,
+      "step": 10992
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010188847716415912,
+      "loss": 1.1376,
+      "step": 10993
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001018795481941158,
+      "loss": 1.1956,
+      "step": 10994
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001018706192240725,
+      "loss": 0.9558,
+      "step": 10995
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001018616902540292,
+      "loss": 1.0434,
+      "step": 10996
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010185276128398588,
+      "loss": 1.0566,
+      "step": 10997
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010184383231394259,
+      "loss": 1.1726,
+      "step": 10998
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010183490334389929,
+      "loss": 1.0257,
+      "step": 10999
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010182597437385597,
+      "loss": 1.0843,
+      "step": 11000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010181704540381267,
+      "loss": 1.0896,
+      "step": 11001
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010180811643376938,
+      "loss": 0.9792,
+      "step": 11002
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010179918746372605,
+      "loss": 1.0454,
+      "step": 11003
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010179025849368276,
+      "loss": 1.0096,
+      "step": 11004
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010178132952363946,
+      "loss": 1.0527,
+      "step": 11005
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010177240055359614,
+      "loss": 1.0003,
+      "step": 11006
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010176347158355284,
+      "loss": 1.0959,
+      "step": 11007
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010175454261350955,
+      "loss": 1.0633,
+      "step": 11008
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010174561364346624,
+      "loss": 1.1172,
+      "step": 11009
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010173668467342293,
+      "loss": 1.1003,
+      "step": 11010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010172775570337962,
+      "loss": 1.006,
+      "step": 11011
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010171882673333632,
+      "loss": 1.1183,
+      "step": 11012
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000101709897763293,
+      "loss": 1.0672,
+      "step": 11013
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001017009687932497,
+      "loss": 1.0878,
+      "step": 11014
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001016920398232064,
+      "loss": 1.0465,
+      "step": 11015
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010168311085316308,
+      "loss": 0.971,
+      "step": 11016
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010167418188311979,
+      "loss": 1.0306,
+      "step": 11017
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010166525291307649,
+      "loss": 1.0728,
+      "step": 11018
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010165632394303317,
+      "loss": 1.0875,
+      "step": 11019
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010164739497298987,
+      "loss": 0.9464,
+      "step": 11020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010163846600294658,
+      "loss": 0.9627,
+      "step": 11021
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010162953703290325,
+      "loss": 1.098,
+      "step": 11022
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010162060806285996,
+      "loss": 1.0063,
+      "step": 11023
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010161167909281666,
+      "loss": 1.0626,
+      "step": 11024
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010160275012277334,
+      "loss": 1.027,
+      "step": 11025
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010159382115273004,
+      "loss": 1.0239,
+      "step": 11026
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010158489218268675,
+      "loss": 0.8533,
+      "step": 11027
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010157596321264342,
+      "loss": 0.9758,
+      "step": 11028
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010156703424260013,
+      "loss": 0.946,
+      "step": 11029
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010155810527255683,
+      "loss": 1.0028,
+      "step": 11030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010154917630251351,
+      "loss": 1.0037,
+      "step": 11031
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010154024733247021,
+      "loss": 1.0726,
+      "step": 11032
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001015313183624269,
+      "loss": 1.0306,
+      "step": 11033
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001015223893923836,
+      "loss": 1.021,
+      "step": 11034
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010151346042234028,
+      "loss": 1.0393,
+      "step": 11035
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010150453145229699,
+      "loss": 1.0506,
+      "step": 11036
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010149560248225366,
+      "loss": 1.126,
+      "step": 11037
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010148667351221037,
+      "loss": 1.0726,
+      "step": 11038
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010147774454216707,
+      "loss": 1.0098,
+      "step": 11039
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010146881557212375,
+      "loss": 1.0053,
+      "step": 11040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010145988660208045,
+      "loss": 1.0206,
+      "step": 11041
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010145095763203716,
+      "loss": 1.0544,
+      "step": 11042
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010144202866199383,
+      "loss": 1.0427,
+      "step": 11043
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010143309969195054,
+      "loss": 1.0541,
+      "step": 11044
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010142417072190724,
+      "loss": 1.1301,
+      "step": 11045
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010141524175186392,
+      "loss": 1.0742,
+      "step": 11046
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010140631278182062,
+      "loss": 1.0443,
+      "step": 11047
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010139738381177733,
+      "loss": 1.1516,
+      "step": 11048
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000101388454841734,
+      "loss": 1.0092,
+      "step": 11049
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010137952587169071,
+      "loss": 1.0695,
+      "step": 11050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010137059690164741,
+      "loss": 1.0652,
+      "step": 11051
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010136166793160409,
+      "loss": 1.0258,
+      "step": 11052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001013527389615608,
+      "loss": 1.0291,
+      "step": 11053
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010134380999151748,
+      "loss": 1.0799,
+      "step": 11054
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010133488102147417,
+      "loss": 0.9495,
+      "step": 11055
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010132595205143088,
+      "loss": 0.9817,
+      "step": 11056
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010131702308138757,
+      "loss": 0.9716,
+      "step": 11057
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010130809411134426,
+      "loss": 1.0684,
+      "step": 11058
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010129916514130095,
+      "loss": 1.0924,
+      "step": 11059
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010129023617125765,
+      "loss": 0.9978,
+      "step": 11060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010128130720121433,
+      "loss": 1.0859,
+      "step": 11061
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010127237823117104,
+      "loss": 0.964,
+      "step": 11062
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010126344926112774,
+      "loss": 0.9967,
+      "step": 11063
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010125452029108442,
+      "loss": 1.0452,
+      "step": 11064
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010124559132104112,
+      "loss": 1.0127,
+      "step": 11065
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010123666235099782,
+      "loss": 0.912,
+      "step": 11066
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001012277333809545,
+      "loss": 1.0766,
+      "step": 11067
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001012188044109112,
+      "loss": 1.0656,
+      "step": 11068
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010120987544086791,
+      "loss": 1.0838,
+      "step": 11069
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010120094647082459,
+      "loss": 1.0563,
+      "step": 11070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010119201750078129,
+      "loss": 1.1186,
+      "step": 11071
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000101183088530738,
+      "loss": 1.0424,
+      "step": 11072
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010117415956069467,
+      "loss": 1.0186,
+      "step": 11073
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010116523059065138,
+      "loss": 0.9835,
+      "step": 11074
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010115630162060808,
+      "loss": 1.0153,
+      "step": 11075
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010114737265056476,
+      "loss": 1.0988,
+      "step": 11076
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010113844368052146,
+      "loss": 0.9915,
+      "step": 11077
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010112951471047815,
+      "loss": 1.0566,
+      "step": 11078
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010112058574043484,
+      "loss": 1.073,
+      "step": 11079
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010111165677039153,
+      "loss": 0.9772,
+      "step": 11080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010110272780034824,
+      "loss": 0.9881,
+      "step": 11081
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010109379883030493,
+      "loss": 1.0801,
+      "step": 11082
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010108486986026162,
+      "loss": 0.9642,
+      "step": 11083
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010107594089021832,
+      "loss": 1.1102,
+      "step": 11084
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000101067011920175,
+      "loss": 1.0904,
+      "step": 11085
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001010580829501317,
+      "loss": 0.9497,
+      "step": 11086
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001010491539800884,
+      "loss": 1.09,
+      "step": 11087
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010104022501004508,
+      "loss": 0.9411,
+      "step": 11088
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010103129604000179,
+      "loss": 1.089,
+      "step": 11089
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010102236706995849,
+      "loss": 0.894,
+      "step": 11090
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010101343809991517,
+      "loss": 1.0535,
+      "step": 11091
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010100450912987187,
+      "loss": 0.9683,
+      "step": 11092
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010099558015982858,
+      "loss": 1.1106,
+      "step": 11093
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010098665118978525,
+      "loss": 1.0525,
+      "step": 11094
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010097772221974196,
+      "loss": 1.0377,
+      "step": 11095
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010096879324969866,
+      "loss": 1.0925,
+      "step": 11096
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010095986427965534,
+      "loss": 0.9475,
+      "step": 11097
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010095093530961204,
+      "loss": 1.0392,
+      "step": 11098
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010094200633956875,
+      "loss": 0.9658,
+      "step": 11099
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010093307736952542,
+      "loss": 1.064,
+      "step": 11100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010092414839948213,
+      "loss": 0.9595,
+      "step": 11101
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010091521942943882,
+      "loss": 0.9894,
+      "step": 11102
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010090629045939551,
+      "loss": 1.0775,
+      "step": 11103
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001008973614893522,
+      "loss": 0.9143,
+      "step": 11104
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001008884325193089,
+      "loss": 0.9941,
+      "step": 11105
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010087950354926559,
+      "loss": 0.9803,
+      "step": 11106
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010087057457922228,
+      "loss": 1.1004,
+      "step": 11107
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010086164560917899,
+      "loss": 1.0009,
+      "step": 11108
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010085271663913569,
+      "loss": 0.9462,
+      "step": 11109
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010084378766909237,
+      "loss": 1.0425,
+      "step": 11110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010083485869904907,
+      "loss": 1.0661,
+      "step": 11111
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010082592972900578,
+      "loss": 0.909,
+      "step": 11112
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010081700075896245,
+      "loss": 1.0719,
+      "step": 11113
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010080807178891916,
+      "loss": 1.0014,
+      "step": 11114
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010079914281887586,
+      "loss": 1.0698,
+      "step": 11115
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010079021384883254,
+      "loss": 1.0733,
+      "step": 11116
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010078128487878924,
+      "loss": 1.2308,
+      "step": 11117
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010077235590874595,
+      "loss": 1.1245,
+      "step": 11118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010076342693870262,
+      "loss": 0.9435,
+      "step": 11119
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010075449796865933,
+      "loss": 1.0103,
+      "step": 11120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010074556899861603,
+      "loss": 1.0137,
+      "step": 11121
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010073664002857271,
+      "loss": 1.047,
+      "step": 11122
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010072771105852941,
+      "loss": 1.2139,
+      "step": 11123
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001007187820884861,
+      "loss": 0.9722,
+      "step": 11124
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010070985311844279,
+      "loss": 1.0065,
+      "step": 11125
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010070092414839948,
+      "loss": 1.049,
+      "step": 11126
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010069199517835619,
+      "loss": 0.9713,
+      "step": 11127
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010068306620831286,
+      "loss": 1.1186,
+      "step": 11128
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010067413723826957,
+      "loss": 1.0069,
+      "step": 11129
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010066520826822627,
+      "loss": 0.9443,
+      "step": 11130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010065627929818295,
+      "loss": 1.0656,
+      "step": 11131
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010064735032813965,
+      "loss": 1.0174,
+      "step": 11132
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010063842135809636,
+      "loss": 0.9411,
+      "step": 11133
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010062949238805303,
+      "loss": 1.0468,
+      "step": 11134
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010062056341800974,
+      "loss": 1.508,
+      "step": 11135
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010061163444796644,
+      "loss": 0.9987,
+      "step": 11136
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010060270547792312,
+      "loss": 1.0778,
+      "step": 11137
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010059377650787982,
+      "loss": 1.1227,
+      "step": 11138
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010058484753783653,
+      "loss": 1.1003,
+      "step": 11139
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001005759185677932,
+      "loss": 0.9751,
+      "step": 11140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010056698959774991,
+      "loss": 1.078,
+      "step": 11141
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010055806062770661,
+      "loss": 0.9679,
+      "step": 11142
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010054913165766329,
+      "loss": 1.0836,
+      "step": 11143
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010054020268762,
+      "loss": 1.0672,
+      "step": 11144
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010053127371757668,
+      "loss": 1.0087,
+      "step": 11145
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010052234474753337,
+      "loss": 1.0747,
+      "step": 11146
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010051341577749008,
+      "loss": 0.9185,
+      "step": 11147
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010050448680744677,
+      "loss": 1.0543,
+      "step": 11148
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010049555783740346,
+      "loss": 1.033,
+      "step": 11149
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010048662886736015,
+      "loss": 1.1363,
+      "step": 11150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010047769989731685,
+      "loss": 1.1123,
+      "step": 11151
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010046877092727353,
+      "loss": 1.0452,
+      "step": 11152
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010045984195723023,
+      "loss": 1.046,
+      "step": 11153
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010045091298718694,
+      "loss": 0.911,
+      "step": 11154
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010044198401714362,
+      "loss": 1.1,
+      "step": 11155
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010043305504710032,
+      "loss": 1.0005,
+      "step": 11156
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010042412607705702,
+      "loss": 1.0538,
+      "step": 11157
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001004151971070137,
+      "loss": 1.0072,
+      "step": 11158
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001004062681369704,
+      "loss": 0.915,
+      "step": 11159
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010039733916692711,
+      "loss": 0.863,
+      "step": 11160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010038841019688379,
+      "loss": 1.0235,
+      "step": 11161
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010037948122684049,
+      "loss": 0.9701,
+      "step": 11162
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001003705522567972,
+      "loss": 1.0083,
+      "step": 11163
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010036162328675387,
+      "loss": 1.0859,
+      "step": 11164
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010035269431671057,
+      "loss": 1.0877,
+      "step": 11165
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010034376534666728,
+      "loss": 1.0545,
+      "step": 11166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010033483637662396,
+      "loss": 1.2203,
+      "step": 11167
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010032590740658066,
+      "loss": 0.927,
+      "step": 11168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010031697843653735,
+      "loss": 1.1326,
+      "step": 11169
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010030804946649404,
+      "loss": 1.0349,
+      "step": 11170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010029912049645073,
+      "loss": 1.1131,
+      "step": 11171
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010029019152640744,
+      "loss": 0.9556,
+      "step": 11172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010028126255636413,
+      "loss": 1.0367,
+      "step": 11173
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010027233358632082,
+      "loss": 0.9836,
+      "step": 11174
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010026340461627752,
+      "loss": 1.091,
+      "step": 11175
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001002544756462342,
+      "loss": 1.0198,
+      "step": 11176
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001002455466761909,
+      "loss": 0.9145,
+      "step": 11177
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001002366177061476,
+      "loss": 0.9313,
+      "step": 11178
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010022768873610428,
+      "loss": 0.9458,
+      "step": 11179
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010021875976606099,
+      "loss": 0.912,
+      "step": 11180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010020983079601769,
+      "loss": 1.2548,
+      "step": 11181
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010020090182597437,
+      "loss": 1.001,
+      "step": 11182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010019197285593107,
+      "loss": 0.9218,
+      "step": 11183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010018304388588778,
+      "loss": 0.9948,
+      "step": 11184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010017411491584445,
+      "loss": 1.0971,
+      "step": 11185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010016518594580116,
+      "loss": 1.1525,
+      "step": 11186
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010015625697575786,
+      "loss": 0.9536,
+      "step": 11187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010014732800571454,
+      "loss": 0.9837,
+      "step": 11188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010013839903567124,
+      "loss": 1.0338,
+      "step": 11189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010012947006562795,
+      "loss": 1.0146,
+      "step": 11190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010012054109558462,
+      "loss": 0.9793,
+      "step": 11191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010011161212554133,
+      "loss": 0.9727,
+      "step": 11192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010010268315549802,
+      "loss": 1.0435,
+      "step": 11193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010009375418545471,
+      "loss": 0.8937,
+      "step": 11194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001000848252154114,
+      "loss": 0.9071,
+      "step": 11195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001000758962453681,
+      "loss": 1.1093,
+      "step": 11196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010006696727532479,
+      "loss": 0.9959,
+      "step": 11197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010005803830528148,
+      "loss": 0.9792,
+      "step": 11198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010004910933523819,
+      "loss": 0.9364,
+      "step": 11199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010004018036519486,
+      "loss": 1.0028,
+      "step": 11200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010003125139515157,
+      "loss": 1.1358,
+      "step": 11201
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010002232242510827,
+      "loss": 1.0564,
+      "step": 11202
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010001339345506495,
+      "loss": 1.029,
+      "step": 11203
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010000446448502165,
+      "loss": 0.9869,
+      "step": 11204
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.999553551497836e-05,
+      "loss": 1.0316,
+      "step": 11205
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.998660654493505e-05,
+      "loss": 0.988,
+      "step": 11206
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.997767757489174e-05,
+      "loss": 0.9757,
+      "step": 11207
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.996874860484844e-05,
+      "loss": 0.9518,
+      "step": 11208
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.995981963480513e-05,
+      "loss": 1.1073,
+      "step": 11209
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.995089066476182e-05,
+      "loss": 0.9954,
+      "step": 11210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.994196169471853e-05,
+      "loss": 0.9604,
+      "step": 11211
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.993303272467522e-05,
+      "loss": 1.0241,
+      "step": 11212
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.992410375463191e-05,
+      "loss": 1.0631,
+      "step": 11213
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.991517478458861e-05,
+      "loss": 1.0557,
+      "step": 11214
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.99062458145453e-05,
+      "loss": 1.007,
+      "step": 11215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989731684450199e-05,
+      "loss": 1.0809,
+      "step": 11216
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.988838787445868e-05,
+      "loss": 0.9978,
+      "step": 11217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.987945890441537e-05,
+      "loss": 1.0381,
+      "step": 11218
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.987052993437206e-05,
+      "loss": 1.0972,
+      "step": 11219
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.986160096432877e-05,
+      "loss": 1.0153,
+      "step": 11220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.985267199428546e-05,
+      "loss": 1.0956,
+      "step": 11221
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.984374302424215e-05,
+      "loss": 1.0307,
+      "step": 11222
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.983481405419885e-05,
+      "loss": 1.1258,
+      "step": 11223
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.982588508415554e-05,
+      "loss": 1.0174,
+      "step": 11224
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.981695611411223e-05,
+      "loss": 1.0619,
+      "step": 11225
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.980802714406894e-05,
+      "loss": 1.0898,
+      "step": 11226
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.979909817402563e-05,
+      "loss": 0.9996,
+      "step": 11227
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.979016920398232e-05,
+      "loss": 1.0396,
+      "step": 11228
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.978124023393902e-05,
+      "loss": 1.1566,
+      "step": 11229
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.977231126389571e-05,
+      "loss": 0.911,
+      "step": 11230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.976338229385242e-05,
+      "loss": 0.897,
+      "step": 11231
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.975445332380911e-05,
+      "loss": 0.9782,
+      "step": 11232
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.97455243537658e-05,
+      "loss": 1.019,
+      "step": 11233
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.97365953837225e-05,
+      "loss": 1.1027,
+      "step": 11234
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.972766641367919e-05,
+      "loss": 1.175,
+      "step": 11235
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.971873744363588e-05,
+      "loss": 1.0179,
+      "step": 11236
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.970980847359257e-05,
+      "loss": 1.1867,
+      "step": 11237
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.970087950354928e-05,
+      "loss": 1.0934,
+      "step": 11238
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.969195053350597e-05,
+      "loss": 0.8985,
+      "step": 11239
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.968302156346266e-05,
+      "loss": 1.0572,
+      "step": 11240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.967409259341935e-05,
+      "loss": 1.0013,
+      "step": 11241
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.966516362337604e-05,
+      "loss": 0.9523,
+      "step": 11242
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.965623465333274e-05,
+      "loss": 1.052,
+      "step": 11243
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.964730568328943e-05,
+      "loss": 1.1121,
+      "step": 11244
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.963837671324613e-05,
+      "loss": 1.1147,
+      "step": 11245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.962944774320283e-05,
+      "loss": 1.1158,
+      "step": 11246
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.962051877315952e-05,
+      "loss": 1.0644,
+      "step": 11247
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.961158980311621e-05,
+      "loss": 1.0562,
+      "step": 11248
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.960266083307291e-05,
+      "loss": 1.0763,
+      "step": 11249
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.95937318630296e-05,
+      "loss": 1.0515,
+      "step": 11250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.95848028929863e-05,
+      "loss": 1.1732,
+      "step": 11251
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.9575873922943e-05,
+      "loss": 0.9601,
+      "step": 11252
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.956694495289969e-05,
+      "loss": 1.1035,
+      "step": 11253
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.955801598285638e-05,
+      "loss": 1.1485,
+      "step": 11254
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.954908701281308e-05,
+      "loss": 0.9096,
+      "step": 11255
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.954015804276977e-05,
+      "loss": 1.1484,
+      "step": 11256
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.953122907272647e-05,
+      "loss": 1.0212,
+      "step": 11257
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.952230010268317e-05,
+      "loss": 1.0449,
+      "step": 11258
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.951337113263986e-05,
+      "loss": 0.9226,
+      "step": 11259
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.950444216259655e-05,
+      "loss": 0.9268,
+      "step": 11260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.949551319255324e-05,
+      "loss": 1.0921,
+      "step": 11261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.948658422250993e-05,
+      "loss": 1.0083,
+      "step": 11262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.947765525246664e-05,
+      "loss": 1.0681,
+      "step": 11263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.946872628242333e-05,
+      "loss": 1.0394,
+      "step": 11264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.945979731238002e-05,
+      "loss": 0.9878,
+      "step": 11265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.94508683423367e-05,
+      "loss": 1.027,
+      "step": 11266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.944193937229341e-05,
+      "loss": 1.1057,
+      "step": 11267
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.94330104022501e-05,
+      "loss": 1.0253,
+      "step": 11268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.942408143220679e-05,
+      "loss": 0.9241,
+      "step": 11269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.94151524621635e-05,
+      "loss": 1.1081,
+      "step": 11270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.940622349212019e-05,
+      "loss": 1.0715,
+      "step": 11271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.939729452207688e-05,
+      "loss": 1.2086,
+      "step": 11272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.938836555203358e-05,
+      "loss": 1.0079,
+      "step": 11273
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.937943658199027e-05,
+      "loss": 1.3258,
+      "step": 11274
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.937050761194696e-05,
+      "loss": 1.0637,
+      "step": 11275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.936157864190367e-05,
+      "loss": 1.053,
+      "step": 11276
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.935264967186036e-05,
+      "loss": 1.02,
+      "step": 11277
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.934372070181706e-05,
+      "loss": 0.9597,
+      "step": 11278
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.933479173177375e-05,
+      "loss": 0.9861,
+      "step": 11279
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.932586276173044e-05,
+      "loss": 1.0296,
+      "step": 11280
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.931693379168715e-05,
+      "loss": 1.146,
+      "step": 11281
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.930800482164384e-05,
+      "loss": 0.9955,
+      "step": 11282
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.929907585160053e-05,
+      "loss": 0.9971,
+      "step": 11283
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.929014688155722e-05,
+      "loss": 1.0264,
+      "step": 11284
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.928121791151391e-05,
+      "loss": 1.0743,
+      "step": 11285
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.92722889414706e-05,
+      "loss": 1.0804,
+      "step": 11286
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.92633599714273e-05,
+      "loss": 1.1051,
+      "step": 11287
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.925443100138399e-05,
+      "loss": 0.9489,
+      "step": 11288
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.924550203134068e-05,
+      "loss": 0.9701,
+      "step": 11289
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.923657306129739e-05,
+      "loss": 0.8958,
+      "step": 11290
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.922764409125408e-05,
+      "loss": 0.9916,
+      "step": 11291
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.921871512121077e-05,
+      "loss": 1.0096,
+      "step": 11292
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.920978615116747e-05,
+      "loss": 1.1445,
+      "step": 11293
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.920085718112416e-05,
+      "loss": 0.8862,
+      "step": 11294
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.919192821108085e-05,
+      "loss": 0.9866,
+      "step": 11295
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.918299924103756e-05,
+      "loss": 1.1461,
+      "step": 11296
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.917407027099425e-05,
+      "loss": 1.0858,
+      "step": 11297
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.916514130095094e-05,
+      "loss": 1.01,
+      "step": 11298
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.915621233090764e-05,
+      "loss": 1.0185,
+      "step": 11299
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.914728336086433e-05,
+      "loss": 1.0653,
+      "step": 11300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.913835439082102e-05,
+      "loss": 1.0324,
+      "step": 11301
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.912942542077773e-05,
+      "loss": 0.9347,
+      "step": 11302
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.912049645073442e-05,
+      "loss": 0.9046,
+      "step": 11303
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.911156748069111e-05,
+      "loss": 0.9457,
+      "step": 11304
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.910263851064781e-05,
+      "loss": 0.8508,
+      "step": 11305
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.90937095406045e-05,
+      "loss": 1.0961,
+      "step": 11306
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.908478057056119e-05,
+      "loss": 0.9293,
+      "step": 11307
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.907585160051788e-05,
+      "loss": 1.0442,
+      "step": 11308
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.906692263047457e-05,
+      "loss": 0.9662,
+      "step": 11309
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.905799366043126e-05,
+      "loss": 0.9947,
+      "step": 11310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.904906469038797e-05,
+      "loss": 1.0853,
+      "step": 11311
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.904013572034466e-05,
+      "loss": 1.0735,
+      "step": 11312
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.903120675030135e-05,
+      "loss": 0.9734,
+      "step": 11313
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.902227778025805e-05,
+      "loss": 1.0113,
+      "step": 11314
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.901334881021474e-05,
+      "loss": 1.055,
+      "step": 11315
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.900441984017143e-05,
+      "loss": 0.9467,
+      "step": 11316
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.899549087012814e-05,
+      "loss": 1.0218,
+      "step": 11317
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.898656190008483e-05,
+      "loss": 0.9491,
+      "step": 11318
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.897763293004152e-05,
+      "loss": 0.9839,
+      "step": 11319
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.896870395999822e-05,
+      "loss": 0.9977,
+      "step": 11320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.895977498995491e-05,
+      "loss": 0.9645,
+      "step": 11321
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.89508460199116e-05,
+      "loss": 1.0317,
+      "step": 11322
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.894191704986831e-05,
+      "loss": 1.0416,
+      "step": 11323
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.8932988079825e-05,
+      "loss": 1.0734,
+      "step": 11324
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.892405910978169e-05,
+      "loss": 0.865,
+      "step": 11325
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.891513013973839e-05,
+      "loss": 0.9258,
+      "step": 11326
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.890620116969508e-05,
+      "loss": 1.0371,
+      "step": 11327
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.889727219965177e-05,
+      "loss": 1.0917,
+      "step": 11328
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.888834322960848e-05,
+      "loss": 1.1354,
+      "step": 11329
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.887941425956517e-05,
+      "loss": 1.0301,
+      "step": 11330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.887048528952186e-05,
+      "loss": 0.9924,
+      "step": 11331
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.886155631947855e-05,
+      "loss": 0.956,
+      "step": 11332
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.885262734943524e-05,
+      "loss": 0.9077,
+      "step": 11333
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.884369837939194e-05,
+      "loss": 0.968,
+      "step": 11334
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.883476940934863e-05,
+      "loss": 1.1434,
+      "step": 11335
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.882584043930532e-05,
+      "loss": 0.9714,
+      "step": 11336
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.881691146926203e-05,
+      "loss": 1.008,
+      "step": 11337
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.880798249921872e-05,
+      "loss": 1.0208,
+      "step": 11338
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.879905352917541e-05,
+      "loss": 0.9015,
+      "step": 11339
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.879012455913211e-05,
+      "loss": 0.9871,
+      "step": 11340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.87811955890888e-05,
+      "loss": 1.0571,
+      "step": 11341
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.87722666190455e-05,
+      "loss": 1.0119,
+      "step": 11342
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.87633376490022e-05,
+      "loss": 1.077,
+      "step": 11343
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.875440867895889e-05,
+      "loss": 1.0429,
+      "step": 11344
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.874547970891558e-05,
+      "loss": 0.947,
+      "step": 11345
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.873655073887228e-05,
+      "loss": 1.0259,
+      "step": 11346
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.872762176882897e-05,
+      "loss": 1.2215,
+      "step": 11347
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.871869279878566e-05,
+      "loss": 0.9594,
+      "step": 11348
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.870976382874237e-05,
+      "loss": 1.0922,
+      "step": 11349
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.870083485869906e-05,
+      "loss": 0.9763,
+      "step": 11350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.869190588865575e-05,
+      "loss": 0.9966,
+      "step": 11351
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.868297691861244e-05,
+      "loss": 1.055,
+      "step": 11352
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.867404794856913e-05,
+      "loss": 0.9112,
+      "step": 11353
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.866511897852582e-05,
+      "loss": 0.9857,
+      "step": 11354
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.865619000848253e-05,
+      "loss": 1.1452,
+      "step": 11355
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.864726103843922e-05,
+      "loss": 1.0484,
+      "step": 11356
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.86383320683959e-05,
+      "loss": 0.9703,
+      "step": 11357
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.862940309835261e-05,
+      "loss": 1.0639,
+      "step": 11358
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.86204741283093e-05,
+      "loss": 0.9148,
+      "step": 11359
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.861154515826599e-05,
+      "loss": 1.0766,
+      "step": 11360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.86026161882227e-05,
+      "loss": 1.0526,
+      "step": 11361
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.859368721817939e-05,
+      "loss": 0.985,
+      "step": 11362
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.858475824813608e-05,
+      "loss": 1.0612,
+      "step": 11363
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.857582927809278e-05,
+      "loss": 0.9278,
+      "step": 11364
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.856690030804947e-05,
+      "loss": 0.9502,
+      "step": 11365
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.855797133800616e-05,
+      "loss": 1.0022,
+      "step": 11366
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.854904236796287e-05,
+      "loss": 1.1022,
+      "step": 11367
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.854011339791956e-05,
+      "loss": 1.0231,
+      "step": 11368
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.853118442787625e-05,
+      "loss": 1.3572,
+      "step": 11369
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.852225545783295e-05,
+      "loss": 1.0075,
+      "step": 11370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.851332648778964e-05,
+      "loss": 1.023,
+      "step": 11371
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.850439751774633e-05,
+      "loss": 1.0019,
+      "step": 11372
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.849546854770304e-05,
+      "loss": 0.9664,
+      "step": 11373
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.848653957765973e-05,
+      "loss": 1.0634,
+      "step": 11374
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.847761060761642e-05,
+      "loss": 1.2122,
+      "step": 11375
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.84686816375731e-05,
+      "loss": 0.8891,
+      "step": 11376
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.84597526675298e-05,
+      "loss": 1.0335,
+      "step": 11377
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.84508236974865e-05,
+      "loss": 1.0362,
+      "step": 11378
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.844189472744319e-05,
+      "loss": 1.1043,
+      "step": 11379
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.843296575739988e-05,
+      "loss": 1.002,
+      "step": 11380
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.842403678735659e-05,
+      "loss": 1.041,
+      "step": 11381
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.841510781731328e-05,
+      "loss": 1.0388,
+      "step": 11382
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.840617884726997e-05,
+      "loss": 1.1722,
+      "step": 11383
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.839724987722667e-05,
+      "loss": 1.0786,
+      "step": 11384
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.838832090718336e-05,
+      "loss": 0.9322,
+      "step": 11385
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.837939193714005e-05,
+      "loss": 1.0293,
+      "step": 11386
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.837046296709676e-05,
+      "loss": 1.0611,
+      "step": 11387
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.836153399705345e-05,
+      "loss": 1.0145,
+      "step": 11388
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.835260502701014e-05,
+      "loss": 0.9755,
+      "step": 11389
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.834367605696684e-05,
+      "loss": 0.9201,
+      "step": 11390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.833474708692353e-05,
+      "loss": 1.0947,
+      "step": 11391
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.832581811688022e-05,
+      "loss": 1.0653,
+      "step": 11392
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.831688914683693e-05,
+      "loss": 1.1161,
+      "step": 11393
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.830796017679362e-05,
+      "loss": 1.0165,
+      "step": 11394
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.829903120675031e-05,
+      "loss": 1.0358,
+      "step": 11395
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.829010223670701e-05,
+      "loss": 0.9575,
+      "step": 11396
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.82811732666637e-05,
+      "loss": 0.9896,
+      "step": 11397
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.827224429662039e-05,
+      "loss": 1.1972,
+      "step": 11398
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.826331532657708e-05,
+      "loss": 1.1675,
+      "step": 11399
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.825438635653377e-05,
+      "loss": 1.0541,
+      "step": 11400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.824545738649046e-05,
+      "loss": 1.0739,
+      "step": 11401
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.823652841644717e-05,
+      "loss": 1.0434,
+      "step": 11402
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.822759944640386e-05,
+      "loss": 0.9257,
+      "step": 11403
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.821867047636055e-05,
+      "loss": 1.0158,
+      "step": 11404
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.820974150631725e-05,
+      "loss": 1.2202,
+      "step": 11405
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.820081253627394e-05,
+      "loss": 1.0405,
+      "step": 11406
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.819188356623063e-05,
+      "loss": 0.9665,
+      "step": 11407
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.818295459618734e-05,
+      "loss": 1.0298,
+      "step": 11408
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.817402562614403e-05,
+      "loss": 0.9344,
+      "step": 11409
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.816509665610072e-05,
+      "loss": 1.1428,
+      "step": 11410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.815616768605742e-05,
+      "loss": 1.0939,
+      "step": 11411
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.814723871601411e-05,
+      "loss": 0.9476,
+      "step": 11412
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.81383097459708e-05,
+      "loss": 1.0619,
+      "step": 11413
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.812938077592751e-05,
+      "loss": 0.93,
+      "step": 11414
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.81204518058842e-05,
+      "loss": 0.969,
+      "step": 11415
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.811152283584089e-05,
+      "loss": 1.0352,
+      "step": 11416
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.810259386579759e-05,
+      "loss": 0.9768,
+      "step": 11417
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.809366489575428e-05,
+      "loss": 1.0929,
+      "step": 11418
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.808473592571097e-05,
+      "loss": 0.9437,
+      "step": 11419
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.807580695566766e-05,
+      "loss": 1.0215,
+      "step": 11420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.806687798562437e-05,
+      "loss": 1.1938,
+      "step": 11421
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.805794901558106e-05,
+      "loss": 1.0745,
+      "step": 11422
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.804902004553775e-05,
+      "loss": 1.1501,
+      "step": 11423
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.804009107549444e-05,
+      "loss": 1.0284,
+      "step": 11424
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.803116210545114e-05,
+      "loss": 1.072,
+      "step": 11425
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.802223313540783e-05,
+      "loss": 1.0526,
+      "step": 11426
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.801330416536452e-05,
+      "loss": 1.1874,
+      "step": 11427
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.800437519532123e-05,
+      "loss": 1.073,
+      "step": 11428
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.799544622527792e-05,
+      "loss": 1.1969,
+      "step": 11429
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.798651725523461e-05,
+      "loss": 1.0051,
+      "step": 11430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.797758828519131e-05,
+      "loss": 1.0411,
+      "step": 11431
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.7968659315148e-05,
+      "loss": 0.9377,
+      "step": 11432
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.79597303451047e-05,
+      "loss": 1.0634,
+      "step": 11433
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.79508013750614e-05,
+      "loss": 0.992,
+      "step": 11434
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.794187240501809e-05,
+      "loss": 1.0267,
+      "step": 11435
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.793294343497478e-05,
+      "loss": 1.0672,
+      "step": 11436
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.792401446493148e-05,
+      "loss": 1.0182,
+      "step": 11437
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.791508549488817e-05,
+      "loss": 0.9928,
+      "step": 11438
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.790615652484486e-05,
+      "loss": 1.0952,
+      "step": 11439
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.789722755480157e-05,
+      "loss": 1.0191,
+      "step": 11440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.788829858475826e-05,
+      "loss": 1.0023,
+      "step": 11441
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.787936961471495e-05,
+      "loss": 1.0137,
+      "step": 11442
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.787044064467164e-05,
+      "loss": 1.0527,
+      "step": 11443
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.786151167462833e-05,
+      "loss": 1.0842,
+      "step": 11444
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.785258270458502e-05,
+      "loss": 0.9855,
+      "step": 11445
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.784365373454172e-05,
+      "loss": 1.1662,
+      "step": 11446
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.783472476449842e-05,
+      "loss": 1.0746,
+      "step": 11447
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.78257957944551e-05,
+      "loss": 0.9236,
+      "step": 11448
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.781686682441181e-05,
+      "loss": 1.0377,
+      "step": 11449
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.78079378543685e-05,
+      "loss": 1.0689,
+      "step": 11450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.779900888432519e-05,
+      "loss": 0.9727,
+      "step": 11451
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.77900799142819e-05,
+      "loss": 0.9451,
+      "step": 11452
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.778115094423859e-05,
+      "loss": 1.0984,
+      "step": 11453
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.777222197419528e-05,
+      "loss": 0.9102,
+      "step": 11454
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.776329300415198e-05,
+      "loss": 1.022,
+      "step": 11455
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.775436403410867e-05,
+      "loss": 1.0531,
+      "step": 11456
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.774543506406536e-05,
+      "loss": 1.1161,
+      "step": 11457
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.773650609402206e-05,
+      "loss": 1.021,
+      "step": 11458
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.772757712397876e-05,
+      "loss": 1.0336,
+      "step": 11459
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.771864815393545e-05,
+      "loss": 1.1638,
+      "step": 11460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.770971918389215e-05,
+      "loss": 1.0585,
+      "step": 11461
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.770079021384884e-05,
+      "loss": 1.0164,
+      "step": 11462
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.769186124380553e-05,
+      "loss": 0.9843,
+      "step": 11463
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.768293227376223e-05,
+      "loss": 1.0087,
+      "step": 11464
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.767400330371893e-05,
+      "loss": 0.9388,
+      "step": 11465
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.766507433367562e-05,
+      "loss": 1.2422,
+      "step": 11466
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.76561453636323e-05,
+      "loss": 1.0478,
+      "step": 11467
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.7647216393589e-05,
+      "loss": 1.2146,
+      "step": 11468
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.763828742354569e-05,
+      "loss": 1.0625,
+      "step": 11469
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.762935845350239e-05,
+      "loss": 0.8681,
+      "step": 11470
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.762042948345908e-05,
+      "loss": 1.0527,
+      "step": 11471
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.761150051341577e-05,
+      "loss": 1.0982,
+      "step": 11472
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.760257154337248e-05,
+      "loss": 1.0629,
+      "step": 11473
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.759364257332917e-05,
+      "loss": 1.1024,
+      "step": 11474
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.758471360328587e-05,
+      "loss": 0.9651,
+      "step": 11475
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.757578463324256e-05,
+      "loss": 1.1782,
+      "step": 11476
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.756685566319925e-05,
+      "loss": 1.1745,
+      "step": 11477
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.755792669315596e-05,
+      "loss": 1.0744,
+      "step": 11478
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.754899772311265e-05,
+      "loss": 1.0185,
+      "step": 11479
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.754006875306934e-05,
+      "loss": 1.0107,
+      "step": 11480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.753113978302604e-05,
+      "loss": 1.1366,
+      "step": 11481
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.752221081298273e-05,
+      "loss": 1.0172,
+      "step": 11482
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.751328184293942e-05,
+      "loss": 1.138,
+      "step": 11483
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.750435287289613e-05,
+      "loss": 1.161,
+      "step": 11484
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.749542390285282e-05,
+      "loss": 1.0102,
+      "step": 11485
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.74864949328095e-05,
+      "loss": 1.0849,
+      "step": 11486
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.747756596276621e-05,
+      "loss": 0.981,
+      "step": 11487
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.74686369927229e-05,
+      "loss": 1.0681,
+      "step": 11488
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.745970802267959e-05,
+      "loss": 1.0161,
+      "step": 11489
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.745077905263628e-05,
+      "loss": 1.0575,
+      "step": 11490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.744185008259297e-05,
+      "loss": 1.2003,
+      "step": 11491
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.743292111254966e-05,
+      "loss": 0.9849,
+      "step": 11492
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.742399214250637e-05,
+      "loss": 0.9625,
+      "step": 11493
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.741506317246306e-05,
+      "loss": 1.0151,
+      "step": 11494
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.740613420241975e-05,
+      "loss": 1.1549,
+      "step": 11495
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.739720523237645e-05,
+      "loss": 0.9705,
+      "step": 11496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.738827626233314e-05,
+      "loss": 1.0368,
+      "step": 11497
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.737934729228983e-05,
+      "loss": 1.4051,
+      "step": 11498
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.737041832224654e-05,
+      "loss": 0.8825,
+      "step": 11499
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.736148935220323e-05,
+      "loss": 1.0289,
+      "step": 11500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.735256038215992e-05,
+      "loss": 0.9462,
+      "step": 11501
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.734363141211662e-05,
+      "loss": 1.1212,
+      "step": 11502
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.733470244207331e-05,
+      "loss": 1.0363,
+      "step": 11503
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.732577347203e-05,
+      "loss": 1.075,
+      "step": 11504
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.731684450198671e-05,
+      "loss": 1.1226,
+      "step": 11505
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.73079155319434e-05,
+      "loss": 1.1123,
+      "step": 11506
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.729898656190009e-05,
+      "loss": 1.0809,
+      "step": 11507
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.729005759185679e-05,
+      "loss": 0.8525,
+      "step": 11508
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.728112862181348e-05,
+      "loss": 1.0498,
+      "step": 11509
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.727219965177017e-05,
+      "loss": 1.0829,
+      "step": 11510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.726327068172686e-05,
+      "loss": 1.0872,
+      "step": 11511
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.725434171168357e-05,
+      "loss": 1.0063,
+      "step": 11512
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.724541274164026e-05,
+      "loss": 0.9473,
+      "step": 11513
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.723648377159695e-05,
+      "loss": 1.1797,
+      "step": 11514
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.722755480155364e-05,
+      "loss": 1.1011,
+      "step": 11515
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.721862583151033e-05,
+      "loss": 1.0355,
+      "step": 11516
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.720969686146703e-05,
+      "loss": 0.987,
+      "step": 11517
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.720076789142372e-05,
+      "loss": 0.9796,
+      "step": 11518
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.719183892138041e-05,
+      "loss": 1.0362,
+      "step": 11519
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.718290995133712e-05,
+      "loss": 1.1018,
+      "step": 11520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.717398098129381e-05,
+      "loss": 1.0423,
+      "step": 11521
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.71650520112505e-05,
+      "loss": 1.0601,
+      "step": 11522
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.71561230412072e-05,
+      "loss": 1.0839,
+      "step": 11523
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.71471940711639e-05,
+      "loss": 1.077,
+      "step": 11524
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.71382651011206e-05,
+      "loss": 1.0516,
+      "step": 11525
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.712933613107729e-05,
+      "loss": 1.2129,
+      "step": 11526
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.712040716103398e-05,
+      "loss": 1.0473,
+      "step": 11527
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.711147819099068e-05,
+      "loss": 1.0097,
+      "step": 11528
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.710254922094737e-05,
+      "loss": 1.1003,
+      "step": 11529
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.709362025090406e-05,
+      "loss": 0.9012,
+      "step": 11530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.708469128086077e-05,
+      "loss": 1.0817,
+      "step": 11531
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.707576231081746e-05,
+      "loss": 1.091,
+      "step": 11532
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.706683334077415e-05,
+      "loss": 1.0429,
+      "step": 11533
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.705790437073084e-05,
+      "loss": 1.0503,
+      "step": 11534
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.704897540068753e-05,
+      "loss": 1.1069,
+      "step": 11535
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.704004643064422e-05,
+      "loss": 0.9948,
+      "step": 11536
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.703111746060092e-05,
+      "loss": 0.939,
+      "step": 11537
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.702218849055761e-05,
+      "loss": 1.0009,
+      "step": 11538
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.70132595205143e-05,
+      "loss": 0.9901,
+      "step": 11539
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.700433055047101e-05,
+      "loss": 1.0432,
+      "step": 11540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.69954015804277e-05,
+      "loss": 1.0998,
+      "step": 11541
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.698647261038439e-05,
+      "loss": 1.0535,
+      "step": 11542
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.69775436403411e-05,
+      "loss": 1.06,
+      "step": 11543
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.696861467029778e-05,
+      "loss": 1.0499,
+      "step": 11544
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.695968570025448e-05,
+      "loss": 0.9887,
+      "step": 11545
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.695075673021118e-05,
+      "loss": 1.144,
+      "step": 11546
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.694182776016787e-05,
+      "loss": 1.0185,
+      "step": 11547
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.693289879012456e-05,
+      "loss": 0.9827,
+      "step": 11548
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.692396982008126e-05,
+      "loss": 1.0782,
+      "step": 11549
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.691504085003795e-05,
+      "loss": 1.053,
+      "step": 11550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690611187999465e-05,
+      "loss": 0.8819,
+      "step": 11551
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689718290995135e-05,
+      "loss": 1.0495,
+      "step": 11552
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.688825393990804e-05,
+      "loss": 0.9492,
+      "step": 11553
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.687932496986473e-05,
+      "loss": 1.0723,
+      "step": 11554
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.687039599982143e-05,
+      "loss": 1.0203,
+      "step": 11555
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.686146702977812e-05,
+      "loss": 0.8368,
+      "step": 11556
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.685253805973482e-05,
+      "loss": 0.9743,
+      "step": 11557
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.68436090896915e-05,
+      "loss": 1.0042,
+      "step": 11558
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.68346801196482e-05,
+      "loss": 1.0326,
+      "step": 11559
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.682575114960489e-05,
+      "loss": 1.2748,
+      "step": 11560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.681682217956159e-05,
+      "loss": 0.9728,
+      "step": 11561
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.680789320951828e-05,
+      "loss": 1.0238,
+      "step": 11562
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.679896423947497e-05,
+      "loss": 0.9346,
+      "step": 11563
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.679003526943168e-05,
+      "loss": 0.9488,
+      "step": 11564
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.678110629938837e-05,
+      "loss": 0.9848,
+      "step": 11565
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.677217732934506e-05,
+      "loss": 0.9698,
+      "step": 11566
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.676324835930176e-05,
+      "loss": 1.0841,
+      "step": 11567
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.675431938925845e-05,
+      "loss": 1.1051,
+      "step": 11568
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.674539041921514e-05,
+      "loss": 1.0077,
+      "step": 11569
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.673646144917185e-05,
+      "loss": 1.0783,
+      "step": 11570
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.672753247912854e-05,
+      "loss": 0.9328,
+      "step": 11571
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.671860350908524e-05,
+      "loss": 0.9987,
+      "step": 11572
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.670967453904193e-05,
+      "loss": 0.9545,
+      "step": 11573
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.670074556899862e-05,
+      "loss": 1.005,
+      "step": 11574
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.669181659895533e-05,
+      "loss": 0.922,
+      "step": 11575
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.668288762891202e-05,
+      "loss": 0.9986,
+      "step": 11576
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.66739586588687e-05,
+      "loss": 1.0198,
+      "step": 11577
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.666502968882541e-05,
+      "loss": 1.0493,
+      "step": 11578
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.66561007187821e-05,
+      "loss": 1.0705,
+      "step": 11579
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.664717174873879e-05,
+      "loss": 0.9354,
+      "step": 11580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.663824277869548e-05,
+      "loss": 0.9346,
+      "step": 11581
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.662931380865217e-05,
+      "loss": 0.9514,
+      "step": 11582
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.662038483860886e-05,
+      "loss": 1.0755,
+      "step": 11583
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.661145586856557e-05,
+      "loss": 1.0411,
+      "step": 11584
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.660252689852226e-05,
+      "loss": 0.9824,
+      "step": 11585
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.659359792847895e-05,
+      "loss": 1.099,
+      "step": 11586
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.658466895843565e-05,
+      "loss": 1.0993,
+      "step": 11587
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.657573998839234e-05,
+      "loss": 1.1369,
+      "step": 11588
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.656681101834903e-05,
+      "loss": 0.9697,
+      "step": 11589
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.655788204830574e-05,
+      "loss": 1.0374,
+      "step": 11590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.654895307826243e-05,
+      "loss": 1.016,
+      "step": 11591
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.654002410821912e-05,
+      "loss": 1.1248,
+      "step": 11592
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.653109513817582e-05,
+      "loss": 1.0676,
+      "step": 11593
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.652216616813251e-05,
+      "loss": 1.1234,
+      "step": 11594
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.65132371980892e-05,
+      "loss": 1.0511,
+      "step": 11595
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.65043082280459e-05,
+      "loss": 1.0763,
+      "step": 11596
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.64953792580026e-05,
+      "loss": 0.973,
+      "step": 11597
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.648645028795929e-05,
+      "loss": 1.0283,
+      "step": 11598
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.647752131791599e-05,
+      "loss": 0.8408,
+      "step": 11599
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.646859234787268e-05,
+      "loss": 0.9745,
+      "step": 11600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.645966337782937e-05,
+      "loss": 0.9559,
+      "step": 11601
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.645073440778606e-05,
+      "loss": 1.0347,
+      "step": 11602
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.644180543774277e-05,
+      "loss": 0.7872,
+      "step": 11603
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.643287646769946e-05,
+      "loss": 0.9134,
+      "step": 11604
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.642394749765615e-05,
+      "loss": 1.0939,
+      "step": 11605
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.641501852761284e-05,
+      "loss": 1.0068,
+      "step": 11606
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.640608955756953e-05,
+      "loss": 1.0656,
+      "step": 11607
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.639716058752623e-05,
+      "loss": 0.9676,
+      "step": 11608
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.638823161748292e-05,
+      "loss": 1.0238,
+      "step": 11609
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.637930264743961e-05,
+      "loss": 1.0912,
+      "step": 11610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.637037367739632e-05,
+      "loss": 1.0155,
+      "step": 11611
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.636144470735301e-05,
+      "loss": 0.9146,
+      "step": 11612
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.63525157373097e-05,
+      "loss": 1.109,
+      "step": 11613
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.63435867672664e-05,
+      "loss": 1.013,
+      "step": 11614
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.63346577972231e-05,
+      "loss": 0.9765,
+      "step": 11615
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.632572882717978e-05,
+      "loss": 1.1666,
+      "step": 11616
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.631679985713649e-05,
+      "loss": 1.2183,
+      "step": 11617
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.630787088709318e-05,
+      "loss": 0.9204,
+      "step": 11618
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.629894191704987e-05,
+      "loss": 1.0052,
+      "step": 11619
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.629001294700657e-05,
+      "loss": 0.9952,
+      "step": 11620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.628108397696326e-05,
+      "loss": 1.0442,
+      "step": 11621
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.627215500691997e-05,
+      "loss": 0.9542,
+      "step": 11622
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.626322603687666e-05,
+      "loss": 0.9623,
+      "step": 11623
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.625429706683335e-05,
+      "loss": 1.2278,
+      "step": 11624
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.624536809679004e-05,
+      "loss": 0.9429,
+      "step": 11625
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.623643912674673e-05,
+      "loss": 1.0717,
+      "step": 11626
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.622751015670342e-05,
+      "loss": 1.1018,
+      "step": 11627
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.621858118666012e-05,
+      "loss": 1.2085,
+      "step": 11628
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.620965221661681e-05,
+      "loss": 0.9921,
+      "step": 11629
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.62007232465735e-05,
+      "loss": 0.8688,
+      "step": 11630
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.619179427653021e-05,
+      "loss": 0.9288,
+      "step": 11631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.61828653064869e-05,
+      "loss": 0.9907,
+      "step": 11632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.617393633644359e-05,
+      "loss": 1.0076,
+      "step": 11633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.61650073664003e-05,
+      "loss": 0.9694,
+      "step": 11634
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.615607839635698e-05,
+      "loss": 1.0069,
+      "step": 11635
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.614714942631368e-05,
+      "loss": 1.024,
+      "step": 11636
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.613822045627038e-05,
+      "loss": 1.084,
+      "step": 11637
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.612929148622707e-05,
+      "loss": 0.9106,
+      "step": 11638
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.612036251618376e-05,
+      "loss": 1.0319,
+      "step": 11639
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.611143354614046e-05,
+      "loss": 0.9994,
+      "step": 11640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.610250457609715e-05,
+      "loss": 1.0213,
+      "step": 11641
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.609357560605384e-05,
+      "loss": 1.0185,
+      "step": 11642
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.608464663601055e-05,
+      "loss": 1.2203,
+      "step": 11643
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.607571766596724e-05,
+      "loss": 0.9801,
+      "step": 11644
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.606678869592393e-05,
+      "loss": 1.0261,
+      "step": 11645
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.605785972588063e-05,
+      "loss": 1.2679,
+      "step": 11646
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.604893075583732e-05,
+      "loss": 1.0901,
+      "step": 11647
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.604000178579401e-05,
+      "loss": 1.0018,
+      "step": 11648
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.60310728157507e-05,
+      "loss": 0.985,
+      "step": 11649
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.60221438457074e-05,
+      "loss": 0.9806,
+      "step": 11650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.601321487566409e-05,
+      "loss": 0.9968,
+      "step": 11651
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.600428590562079e-05,
+      "loss": 1.0022,
+      "step": 11652
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.599535693557748e-05,
+      "loss": 1.1568,
+      "step": 11653
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.598642796553417e-05,
+      "loss": 1.0091,
+      "step": 11654
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.597749899549088e-05,
+      "loss": 1.0183,
+      "step": 11655
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.596857002544757e-05,
+      "loss": 1.0232,
+      "step": 11656
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.595964105540426e-05,
+      "loss": 0.9779,
+      "step": 11657
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.595071208536096e-05,
+      "loss": 1.1029,
+      "step": 11658
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.594178311531765e-05,
+      "loss": 0.9917,
+      "step": 11659
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.593285414527434e-05,
+      "loss": 1.0208,
+      "step": 11660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.592392517523105e-05,
+      "loss": 1.0135,
+      "step": 11661
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.591499620518774e-05,
+      "loss": 1.0751,
+      "step": 11662
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.590606723514443e-05,
+      "loss": 1.0253,
+      "step": 11663
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.589713826510113e-05,
+      "loss": 0.9816,
+      "step": 11664
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.588820929505782e-05,
+      "loss": 0.9856,
+      "step": 11665
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.587928032501451e-05,
+      "loss": 1.0437,
+      "step": 11666
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.587035135497122e-05,
+      "loss": 0.9846,
+      "step": 11667
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.58614223849279e-05,
+      "loss": 1.1031,
+      "step": 11668
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.58524934148846e-05,
+      "loss": 1.1806,
+      "step": 11669
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.58435644448413e-05,
+      "loss": 1.1179,
+      "step": 11670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.583463547479799e-05,
+      "loss": 1.04,
+      "step": 11671
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.582570650475468e-05,
+      "loss": 0.9503,
+      "step": 11672
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.581677753471137e-05,
+      "loss": 0.9509,
+      "step": 11673
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.580784856466806e-05,
+      "loss": 1.0174,
+      "step": 11674
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.579891959462477e-05,
+      "loss": 0.9634,
+      "step": 11675
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.578999062458146e-05,
+      "loss": 0.8844,
+      "step": 11676
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.578106165453815e-05,
+      "loss": 1.1033,
+      "step": 11677
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.577213268449485e-05,
+      "loss": 1.0091,
+      "step": 11678
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.576320371445154e-05,
+      "loss": 0.9981,
+      "step": 11679
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.575427474440823e-05,
+      "loss": 1.0406,
+      "step": 11680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.574534577436494e-05,
+      "loss": 1.0485,
+      "step": 11681
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.573641680432163e-05,
+      "loss": 1.0514,
+      "step": 11682
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.572748783427832e-05,
+      "loss": 0.9755,
+      "step": 11683
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.571855886423502e-05,
+      "loss": 1.0945,
+      "step": 11684
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.570962989419171e-05,
+      "loss": 0.9883,
+      "step": 11685
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.57007009241484e-05,
+      "loss": 1.139,
+      "step": 11686
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.56917719541051e-05,
+      "loss": 1.0454,
+      "step": 11687
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.56828429840618e-05,
+      "loss": 0.8908,
+      "step": 11688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.567391401401849e-05,
+      "loss": 1.0949,
+      "step": 11689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.566498504397519e-05,
+      "loss": 1.0625,
+      "step": 11690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.565605607393188e-05,
+      "loss": 1.036,
+      "step": 11691
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.564712710388857e-05,
+      "loss": 1.1024,
+      "step": 11692
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.563819813384526e-05,
+      "loss": 0.9715,
+      "step": 11693
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.562926916380195e-05,
+      "loss": 0.9899,
+      "step": 11694
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.562034019375866e-05,
+      "loss": 0.87,
+      "step": 11695
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.561141122371535e-05,
+      "loss": 0.9984,
+      "step": 11696
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.560248225367204e-05,
+      "loss": 1.0681,
+      "step": 11697
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.559355328362873e-05,
+      "loss": 1.0602,
+      "step": 11698
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.558462431358543e-05,
+      "loss": 1.1299,
+      "step": 11699
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.557569534354212e-05,
+      "loss": 1.087,
+      "step": 11700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.556676637349881e-05,
+      "loss": 1.051,
+      "step": 11701
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.555783740345552e-05,
+      "loss": 1.0207,
+      "step": 11702
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.554890843341221e-05,
+      "loss": 1.0617,
+      "step": 11703
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.55399794633689e-05,
+      "loss": 1.0194,
+      "step": 11704
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.55310504933256e-05,
+      "loss": 1.0296,
+      "step": 11705
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.55221215232823e-05,
+      "loss": 0.9293,
+      "step": 11706
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.551319255323898e-05,
+      "loss": 1.0254,
+      "step": 11707
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.550426358319569e-05,
+      "loss": 1.0967,
+      "step": 11708
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.549533461315238e-05,
+      "loss": 1.015,
+      "step": 11709
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.548640564310907e-05,
+      "loss": 0.911,
+      "step": 11710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.547747667306577e-05,
+      "loss": 0.9902,
+      "step": 11711
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.546854770302246e-05,
+      "loss": 1.0377,
+      "step": 11712
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.545961873297915e-05,
+      "loss": 1.1838,
+      "step": 11713
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.545068976293586e-05,
+      "loss": 1.2099,
+      "step": 11714
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.544176079289255e-05,
+      "loss": 1.0013,
+      "step": 11715
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.543283182284924e-05,
+      "loss": 0.9787,
+      "step": 11716
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.542390285280593e-05,
+      "loss": 0.9988,
+      "step": 11717
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.541497388276262e-05,
+      "loss": 1.0785,
+      "step": 11718
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.540604491271932e-05,
+      "loss": 1.0755,
+      "step": 11719
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.539711594267601e-05,
+      "loss": 0.9534,
+      "step": 11720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.53881869726327e-05,
+      "loss": 1.0478,
+      "step": 11721
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.537925800258941e-05,
+      "loss": 1.0716,
+      "step": 11722
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.53703290325461e-05,
+      "loss": 1.1012,
+      "step": 11723
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.536140006250279e-05,
+      "loss": 1.0757,
+      "step": 11724
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.53524710924595e-05,
+      "loss": 1.0327,
+      "step": 11725
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.534354212241618e-05,
+      "loss": 1.1397,
+      "step": 11726
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.533461315237287e-05,
+      "loss": 1.0068,
+      "step": 11727
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.532568418232958e-05,
+      "loss": 1.329,
+      "step": 11728
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.531675521228627e-05,
+      "loss": 1.1406,
+      "step": 11729
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.530782624224296e-05,
+      "loss": 0.9853,
+      "step": 11730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.529889727219966e-05,
+      "loss": 0.9769,
+      "step": 11731
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.528996830215635e-05,
+      "loss": 1.0661,
+      "step": 11732
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.528103933211304e-05,
+      "loss": 0.9367,
+      "step": 11733
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.527211036206975e-05,
+      "loss": 0.9726,
+      "step": 11734
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.526318139202644e-05,
+      "loss": 0.9939,
+      "step": 11735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.525425242198313e-05,
+      "loss": 1.1676,
+      "step": 11736
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.524532345193983e-05,
+      "loss": 0.9803,
+      "step": 11737
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.523639448189652e-05,
+      "loss": 1.0614,
+      "step": 11738
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.522746551185321e-05,
+      "loss": 0.971,
+      "step": 11739
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.52185365418099e-05,
+      "loss": 1.0233,
+      "step": 11740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.52096075717666e-05,
+      "loss": 1.1036,
+      "step": 11741
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.520067860172329e-05,
+      "loss": 1.156,
+      "step": 11742
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.519174963167999e-05,
+      "loss": 1.0109,
+      "step": 11743
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.518282066163668e-05,
+      "loss": 0.937,
+      "step": 11744
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.517389169159337e-05,
+      "loss": 1.0374,
+      "step": 11745
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.516496272155008e-05,
+      "loss": 0.9448,
+      "step": 11746
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.515603375150677e-05,
+      "loss": 1.1635,
+      "step": 11747
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.514710478146346e-05,
+      "loss": 1.0041,
+      "step": 11748
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.513817581142016e-05,
+      "loss": 1.0701,
+      "step": 11749
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.512924684137685e-05,
+      "loss": 1.091,
+      "step": 11750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.512031787133354e-05,
+      "loss": 1.0252,
+      "step": 11751
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.511138890129025e-05,
+      "loss": 1.0667,
+      "step": 11752
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.510245993124694e-05,
+      "loss": 1.0784,
+      "step": 11753
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.509353096120363e-05,
+      "loss": 1.0064,
+      "step": 11754
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.508460199116033e-05,
+      "loss": 1.0814,
+      "step": 11755
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.507567302111702e-05,
+      "loss": 1.1721,
+      "step": 11756
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.506674405107371e-05,
+      "loss": 1.0516,
+      "step": 11757
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.505781508103042e-05,
+      "loss": 1.0733,
+      "step": 11758
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.50488861109871e-05,
+      "loss": 1.0094,
+      "step": 11759
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.50399571409438e-05,
+      "loss": 0.9745,
+      "step": 11760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.50310281709005e-05,
+      "loss": 1.0231,
+      "step": 11761
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.502209920085719e-05,
+      "loss": 1.0929,
+      "step": 11762
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.501317023081388e-05,
+      "loss": 1.0014,
+      "step": 11763
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.500424126077057e-05,
+      "loss": 0.9432,
+      "step": 11764
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.499531229072726e-05,
+      "loss": 0.9522,
+      "step": 11765
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.498638332068395e-05,
+      "loss": 1.0222,
+      "step": 11766
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.497745435064066e-05,
+      "loss": 1.1231,
+      "step": 11767
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.496852538059735e-05,
+      "loss": 1.0043,
+      "step": 11768
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.495959641055405e-05,
+      "loss": 0.9318,
+      "step": 11769
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.495066744051074e-05,
+      "loss": 1.1373,
+      "step": 11770
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.494173847046743e-05,
+      "loss": 1.044,
+      "step": 11771
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.493280950042414e-05,
+      "loss": 1.0046,
+      "step": 11772
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.492388053038083e-05,
+      "loss": 0.9847,
+      "step": 11773
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.491495156033752e-05,
+      "loss": 0.9946,
+      "step": 11774
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.490602259029422e-05,
+      "loss": 1.1948,
+      "step": 11775
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.489709362025091e-05,
+      "loss": 1.0902,
+      "step": 11776
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.48881646502076e-05,
+      "loss": 0.9838,
+      "step": 11777
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.48792356801643e-05,
+      "loss": 1.047,
+      "step": 11778
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.4870306710121e-05,
+      "loss": 0.9564,
+      "step": 11779
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.486137774007769e-05,
+      "loss": 1.0299,
+      "step": 11780
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.485244877003439e-05,
+      "loss": 0.9908,
+      "step": 11781
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.484351979999108e-05,
+      "loss": 1.0773,
+      "step": 11782
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.483459082994777e-05,
+      "loss": 1.1899,
+      "step": 11783
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.482566185990446e-05,
+      "loss": 1.0517,
+      "step": 11784
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.481673288986115e-05,
+      "loss": 1.0346,
+      "step": 11785
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.480780391981786e-05,
+      "loss": 1.0214,
+      "step": 11786
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.479887494977455e-05,
+      "loss": 0.9972,
+      "step": 11787
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.478994597973124e-05,
+      "loss": 1.0747,
+      "step": 11788
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.478101700968793e-05,
+      "loss": 1.0477,
+      "step": 11789
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.477208803964463e-05,
+      "loss": 0.9879,
+      "step": 11790
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.476315906960132e-05,
+      "loss": 0.8872,
+      "step": 11791
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.475423009955801e-05,
+      "loss": 1.0463,
+      "step": 11792
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.474530112951472e-05,
+      "loss": 1.0696,
+      "step": 11793
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.473637215947141e-05,
+      "loss": 0.9982,
+      "step": 11794
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.47274431894281e-05,
+      "loss": 1.094,
+      "step": 11795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.47185142193848e-05,
+      "loss": 1.001,
+      "step": 11796
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.470958524934149e-05,
+      "loss": 1.0139,
+      "step": 11797
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.470065627929818e-05,
+      "loss": 1.0953,
+      "step": 11798
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.469172730925489e-05,
+      "loss": 1.0664,
+      "step": 11799
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.468279833921158e-05,
+      "loss": 1.0624,
+      "step": 11800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.467386936916827e-05,
+      "loss": 0.9663,
+      "step": 11801
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.466494039912497e-05,
+      "loss": 0.9927,
+      "step": 11802
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.465601142908166e-05,
+      "loss": 1.0564,
+      "step": 11803
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.464708245903835e-05,
+      "loss": 1.0128,
+      "step": 11804
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.463815348899506e-05,
+      "loss": 1.0103,
+      "step": 11805
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.462922451895175e-05,
+      "loss": 1.057,
+      "step": 11806
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.462029554890844e-05,
+      "loss": 1.0471,
+      "step": 11807
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.461136657886513e-05,
+      "loss": 1.1301,
+      "step": 11808
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.460243760882182e-05,
+      "loss": 1.1479,
+      "step": 11809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.459350863877851e-05,
+      "loss": 1.0659,
+      "step": 11810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.458457966873521e-05,
+      "loss": 0.9794,
+      "step": 11811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.45756506986919e-05,
+      "loss": 0.9284,
+      "step": 11812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.45667217286486e-05,
+      "loss": 0.8942,
+      "step": 11813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.45577927586053e-05,
+      "loss": 0.9855,
+      "step": 11814
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.454886378856199e-05,
+      "loss": 1.0618,
+      "step": 11815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.45399348185187e-05,
+      "loss": 0.9277,
+      "step": 11816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.453100584847538e-05,
+      "loss": 0.9528,
+      "step": 11817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.452207687843207e-05,
+      "loss": 1.1368,
+      "step": 11818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.451314790838878e-05,
+      "loss": 1.0947,
+      "step": 11819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.450421893834547e-05,
+      "loss": 0.9751,
+      "step": 11820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.449528996830216e-05,
+      "loss": 0.9583,
+      "step": 11821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.448636099825886e-05,
+      "loss": 1.1324,
+      "step": 11822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.447743202821555e-05,
+      "loss": 1.106,
+      "step": 11823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.446850305817224e-05,
+      "loss": 1.0212,
+      "step": 11824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.445957408812895e-05,
+      "loss": 0.9068,
+      "step": 11825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.445064511808564e-05,
+      "loss": 1.0746,
+      "step": 11826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.444171614804233e-05,
+      "loss": 1.0857,
+      "step": 11827
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.443278717799903e-05,
+      "loss": 0.9524,
+      "step": 11828
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.442385820795572e-05,
+      "loss": 0.9915,
+      "step": 11829
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.441492923791241e-05,
+      "loss": 1.046,
+      "step": 11830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.44060002678691e-05,
+      "loss": 1.0101,
+      "step": 11831
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.43970712978258e-05,
+      "loss": 1.172,
+      "step": 11832
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.438814232778249e-05,
+      "loss": 1.0454,
+      "step": 11833
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.437921335773919e-05,
+      "loss": 1.0156,
+      "step": 11834
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.437028438769588e-05,
+      "loss": 0.9554,
+      "step": 11835
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.436135541765257e-05,
+      "loss": 1.0741,
+      "step": 11836
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.435242644760927e-05,
+      "loss": 0.9959,
+      "step": 11837
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.434349747756597e-05,
+      "loss": 1.0755,
+      "step": 11838
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.433456850752266e-05,
+      "loss": 1.0471,
+      "step": 11839
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.432563953747936e-05,
+      "loss": 1.0336,
+      "step": 11840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.431671056743605e-05,
+      "loss": 0.9859,
+      "step": 11841
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.430778159739274e-05,
+      "loss": 0.945,
+      "step": 11842
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.429885262734944e-05,
+      "loss": 1.1045,
+      "step": 11843
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.428992365730614e-05,
+      "loss": 0.9914,
+      "step": 11844
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.428099468726283e-05,
+      "loss": 1.0423,
+      "step": 11845
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.427206571721953e-05,
+      "loss": 0.9548,
+      "step": 11846
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.426313674717622e-05,
+      "loss": 1.0895,
+      "step": 11847
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.425420777713291e-05,
+      "loss": 1.1295,
+      "step": 11848
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.424527880708961e-05,
+      "loss": 1.1224,
+      "step": 11849
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.42363498370463e-05,
+      "loss": 1.0592,
+      "step": 11850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.4227420867003e-05,
+      "loss": 1.2792,
+      "step": 11851
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.42184918969597e-05,
+      "loss": 0.9287,
+      "step": 11852
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.420956292691639e-05,
+      "loss": 0.9566,
+      "step": 11853
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.420063395687308e-05,
+      "loss": 1.0402,
+      "step": 11854
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.419170498682977e-05,
+      "loss": 1.0834,
+      "step": 11855
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.418277601678646e-05,
+      "loss": 1.1044,
+      "step": 11856
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.417384704674315e-05,
+      "loss": 1.0109,
+      "step": 11857
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.416491807669986e-05,
+      "loss": 1.0669,
+      "step": 11858
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.415598910665655e-05,
+      "loss": 0.9569,
+      "step": 11859
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.414706013661324e-05,
+      "loss": 1.1399,
+      "step": 11860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.413813116656994e-05,
+      "loss": 0.9572,
+      "step": 11861
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.412920219652663e-05,
+      "loss": 1.0427,
+      "step": 11862
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.412027322648332e-05,
+      "loss": 1.0201,
+      "step": 11863
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.411134425644003e-05,
+      "loss": 1.0304,
+      "step": 11864
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.410241528639672e-05,
+      "loss": 0.9711,
+      "step": 11865
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.409348631635342e-05,
+      "loss": 0.9167,
+      "step": 11866
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.408455734631011e-05,
+      "loss": 1.134,
+      "step": 11867
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.40756283762668e-05,
+      "loss": 1.0679,
+      "step": 11868
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.40666994062235e-05,
+      "loss": 0.9867,
+      "step": 11869
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.40577704361802e-05,
+      "loss": 1.1508,
+      "step": 11870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.404884146613689e-05,
+      "loss": 1.0396,
+      "step": 11871
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.403991249609359e-05,
+      "loss": 1.0467,
+      "step": 11872
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.403098352605028e-05,
+      "loss": 0.9424,
+      "step": 11873
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.402205455600697e-05,
+      "loss": 0.9325,
+      "step": 11874
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.401312558596366e-05,
+      "loss": 0.9296,
+      "step": 11875
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.400419661592035e-05,
+      "loss": 1.0766,
+      "step": 11876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.399526764587706e-05,
+      "loss": 1.1447,
+      "step": 11877
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.398633867583375e-05,
+      "loss": 1.0187,
+      "step": 11878
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.397740970579044e-05,
+      "loss": 0.959,
+      "step": 11879
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.396848073574713e-05,
+      "loss": 1.0048,
+      "step": 11880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.395955176570383e-05,
+      "loss": 0.9567,
+      "step": 11881
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.395062279566052e-05,
+      "loss": 0.9516,
+      "step": 11882
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.394169382561721e-05,
+      "loss": 0.9722,
+      "step": 11883
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.393276485557392e-05,
+      "loss": 1.0751,
+      "step": 11884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.392383588553061e-05,
+      "loss": 1.069,
+      "step": 11885
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.39149069154873e-05,
+      "loss": 0.9904,
+      "step": 11886
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.3905977945444e-05,
+      "loss": 0.9727,
+      "step": 11887
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.389704897540069e-05,
+      "loss": 1.0516,
+      "step": 11888
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.388812000535738e-05,
+      "loss": 1.0198,
+      "step": 11889
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.387919103531409e-05,
+      "loss": 1.0065,
+      "step": 11890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.387026206527078e-05,
+      "loss": 1.1476,
+      "step": 11891
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.386133309522747e-05,
+      "loss": 1.0131,
+      "step": 11892
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.385240412518417e-05,
+      "loss": 1.0123,
+      "step": 11893
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.384347515514086e-05,
+      "loss": 0.9979,
+      "step": 11894
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.383454618509755e-05,
+      "loss": 1.0002,
+      "step": 11895
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.382561721505426e-05,
+      "loss": 1.0267,
+      "step": 11896
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.381668824501095e-05,
+      "loss": 1.0992,
+      "step": 11897
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.380775927496764e-05,
+      "loss": 1.0182,
+      "step": 11898
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.379883030492433e-05,
+      "loss": 1.0768,
+      "step": 11899
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.378990133488102e-05,
+      "loss": 0.914,
+      "step": 11900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.378097236483771e-05,
+      "loss": 1.034,
+      "step": 11901
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.377204339479441e-05,
+      "loss": 0.9233,
+      "step": 11902
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.37631144247511e-05,
+      "loss": 1.0785,
+      "step": 11903
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.37541854547078e-05,
+      "loss": 0.9679,
+      "step": 11904
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.37452564846645e-05,
+      "loss": 1.1437,
+      "step": 11905
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.373632751462119e-05,
+      "loss": 1.0085,
+      "step": 11906
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.372739854457788e-05,
+      "loss": 1.2668,
+      "step": 11907
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.371846957453458e-05,
+      "loss": 1.0253,
+      "step": 11908
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.370954060449127e-05,
+      "loss": 1.3301,
+      "step": 11909
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.370061163444796e-05,
+      "loss": 1.0283,
+      "step": 11910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.369168266440467e-05,
+      "loss": 1.1267,
+      "step": 11911
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.368275369436136e-05,
+      "loss": 1.0417,
+      "step": 11912
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.367382472431805e-05,
+      "loss": 1.1023,
+      "step": 11913
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366489575427475e-05,
+      "loss": 1.0228,
+      "step": 11914
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.365596678423144e-05,
+      "loss": 1.0621,
+      "step": 11915
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364703781418815e-05,
+      "loss": 1.0046,
+      "step": 11916
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.363810884414484e-05,
+      "loss": 0.9697,
+      "step": 11917
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362917987410153e-05,
+      "loss": 1.0831,
+      "step": 11918
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362025090405823e-05,
+      "loss": 0.9533,
+      "step": 11919
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361132193401492e-05,
+      "loss": 0.9912,
+      "step": 11920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.360239296397161e-05,
+      "loss": 1.0805,
+      "step": 11921
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.35934639939283e-05,
+      "loss": 1.0027,
+      "step": 11922
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3584535023885e-05,
+      "loss": 1.0682,
+      "step": 11923
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357560605384169e-05,
+      "loss": 1.185,
+      "step": 11924
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.356667708379839e-05,
+      "loss": 1.0176,
+      "step": 11925
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.355774811375508e-05,
+      "loss": 1.1471,
+      "step": 11926
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.354881914371177e-05,
+      "loss": 0.9136,
+      "step": 11927
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.353989017366847e-05,
+      "loss": 1.08,
+      "step": 11928
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.353096120362516e-05,
+      "loss": 1.1098,
+      "step": 11929
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.352203223358186e-05,
+      "loss": 1.0192,
+      "step": 11930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.351310326353856e-05,
+      "loss": 0.8873,
+      "step": 11931
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.350417429349525e-05,
+      "loss": 1.1506,
+      "step": 11932
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.349524532345194e-05,
+      "loss": 0.9846,
+      "step": 11933
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.348631635340864e-05,
+      "loss": 1.094,
+      "step": 11934
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.347738738336533e-05,
+      "loss": 0.9694,
+      "step": 11935
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.346845841332203e-05,
+      "loss": 1.1165,
+      "step": 11936
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.345952944327873e-05,
+      "loss": 1.2363,
+      "step": 11937
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.345060047323542e-05,
+      "loss": 1.0648,
+      "step": 11938
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.344167150319211e-05,
+      "loss": 1.1407,
+      "step": 11939
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.343274253314881e-05,
+      "loss": 1.0086,
+      "step": 11940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.34238135631055e-05,
+      "loss": 1.0358,
+      "step": 11941
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.34148845930622e-05,
+      "loss": 0.9258,
+      "step": 11942
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.34059556230189e-05,
+      "loss": 0.9317,
+      "step": 11943
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.339702665297559e-05,
+      "loss": 0.9618,
+      "step": 11944
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.338809768293228e-05,
+      "loss": 1.0083,
+      "step": 11945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.337916871288897e-05,
+      "loss": 1.043,
+      "step": 11946
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.337023974284566e-05,
+      "loss": 1.1069,
+      "step": 11947
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.336131077280235e-05,
+      "loss": 1.0491,
+      "step": 11948
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.335238180275906e-05,
+      "loss": 1.0488,
+      "step": 11949
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.334345283271575e-05,
+      "loss": 0.9694,
+      "step": 11950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.333452386267244e-05,
+      "loss": 0.9634,
+      "step": 11951
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.332559489262914e-05,
+      "loss": 1.1578,
+      "step": 11952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.331666592258583e-05,
+      "loss": 1.0601,
+      "step": 11953
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.330773695254252e-05,
+      "loss": 1.0996,
+      "step": 11954
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.329880798249923e-05,
+      "loss": 1.0669,
+      "step": 11955
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.328987901245592e-05,
+      "loss": 1.0708,
+      "step": 11956
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.32809500424126e-05,
+      "loss": 1.0161,
+      "step": 11957
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.327202107236931e-05,
+      "loss": 0.9272,
+      "step": 11958
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3263092102326e-05,
+      "loss": 0.933,
+      "step": 11959
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.325416313228269e-05,
+      "loss": 1.005,
+      "step": 11960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.32452341622394e-05,
+      "loss": 1.0155,
+      "step": 11961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.323630519219609e-05,
+      "loss": 0.9344,
+      "step": 11962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.322737622215279e-05,
+      "loss": 1.0205,
+      "step": 11963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.321844725210948e-05,
+      "loss": 0.919,
+      "step": 11964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.320951828206617e-05,
+      "loss": 1.1175,
+      "step": 11965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.320058931202286e-05,
+      "loss": 1.2346,
+      "step": 11966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.319166034197955e-05,
+      "loss": 1.2683,
+      "step": 11967
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.318273137193624e-05,
+      "loss": 1.0778,
+      "step": 11968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.317380240189295e-05,
+      "loss": 0.9679,
+      "step": 11969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.316487343184964e-05,
+      "loss": 1.0301,
+      "step": 11970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.315594446180633e-05,
+      "loss": 0.9881,
+      "step": 11971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.314701549176303e-05,
+      "loss": 0.9203,
+      "step": 11972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.313808652171972e-05,
+      "loss": 1.0432,
+      "step": 11973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.312915755167641e-05,
+      "loss": 0.9602,
+      "step": 11974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.312022858163312e-05,
+      "loss": 1.0966,
+      "step": 11975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.311129961158981e-05,
+      "loss": 1.1284,
+      "step": 11976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.31023706415465e-05,
+      "loss": 1.0105,
+      "step": 11977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.30934416715032e-05,
+      "loss": 1.4099,
+      "step": 11978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.308451270145989e-05,
+      "loss": 1.0317,
+      "step": 11979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.307558373141658e-05,
+      "loss": 1.016,
+      "step": 11980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.306665476137329e-05,
+      "loss": 0.9614,
+      "step": 11981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.305772579132998e-05,
+      "loss": 1.1073,
+      "step": 11982
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.304879682128667e-05,
+      "loss": 1.1824,
+      "step": 11983
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.303986785124337e-05,
+      "loss": 1.0133,
+      "step": 11984
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.303093888120006e-05,
+      "loss": 0.9842,
+      "step": 11985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.302200991115675e-05,
+      "loss": 1.1902,
+      "step": 11986
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301308094111346e-05,
+      "loss": 0.9116,
+      "step": 11987
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300415197107015e-05,
+      "loss": 0.9179,
+      "step": 11988
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.299522300102684e-05,
+      "loss": 1.149,
+      "step": 11989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.298629403098353e-05,
+      "loss": 1.0158,
+      "step": 11990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297736506094022e-05,
+      "loss": 0.941,
+      "step": 11991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.296843609089691e-05,
+      "loss": 1.1464,
+      "step": 11992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295950712085361e-05,
+      "loss": 0.9645,
+      "step": 11993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29505781508103e-05,
+      "loss": 1.1731,
+      "step": 11994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.2941649180767e-05,
+      "loss": 0.9844,
+      "step": 11995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29327202107237e-05,
+      "loss": 0.9091,
+      "step": 11996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.292379124068039e-05,
+      "loss": 1.1316,
+      "step": 11997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291486227063708e-05,
+      "loss": 0.9606,
+      "step": 11998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.290593330059378e-05,
+      "loss": 1.065,
+      "step": 11999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.289700433055047e-05,
+      "loss": 1.0657,
+      "step": 12000
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 1.0268734693527222,
+      "eval_runtime": 3050.3129,
+      "eval_samples_per_second": 6.165,
+      "eval_steps_per_second": 0.771,
+      "step": 12000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288807536050716e-05,
+      "loss": 1.0685,
+      "step": 12001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.287914639046387e-05,
+      "loss": 0.9889,
+      "step": 12002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.287021742042056e-05,
+      "loss": 1.0385,
+      "step": 12003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286128845037725e-05,
+      "loss": 1.065,
+      "step": 12004
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285235948033395e-05,
+      "loss": 0.9704,
+      "step": 12005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.284343051029064e-05,
+      "loss": 1.1276,
+      "step": 12006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.283450154024733e-05,
+      "loss": 1.0423,
+      "step": 12007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282557257020404e-05,
+      "loss": 1.0979,
+      "step": 12008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.281664360016073e-05,
+      "loss": 1.069,
+      "step": 12009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280771463011742e-05,
+      "loss": 0.9421,
+      "step": 12010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279878566007412e-05,
+      "loss": 1.0399,
+      "step": 12011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.278985669003081e-05,
+      "loss": 1.1161,
+      "step": 12012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27809277199875e-05,
+      "loss": 1.0322,
+      "step": 12013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27719987499442e-05,
+      "loss": 0.8418,
+      "step": 12014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276306977990088e-05,
+      "loss": 1.0048,
+      "step": 12015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.275414080985759e-05,
+      "loss": 1.1212,
+      "step": 12016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.274521183981428e-05,
+      "loss": 1.0856,
+      "step": 12017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273628286977097e-05,
+      "loss": 1.0291,
+      "step": 12018
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.272735389972767e-05,
+      "loss": 1.05,
+      "step": 12019
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271842492968436e-05,
+      "loss": 0.9524,
+      "step": 12020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270949595964105e-05,
+      "loss": 1.2474,
+      "step": 12021
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270056698959776e-05,
+      "loss": 1.0198,
+      "step": 12022
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269163801955445e-05,
+      "loss": 0.9428,
+      "step": 12023
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268270904951114e-05,
+      "loss": 0.9978,
+      "step": 12024
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.267378007946784e-05,
+      "loss": 1.0038,
+      "step": 12025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266485110942453e-05,
+      "loss": 1.0967,
+      "step": 12026
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.265592213938122e-05,
+      "loss": 1.0523,
+      "step": 12027
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264699316933793e-05,
+      "loss": 1.0629,
+      "step": 12028
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263806419929462e-05,
+      "loss": 1.0609,
+      "step": 12029
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262913522925131e-05,
+      "loss": 1.0745,
+      "step": 12030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262020625920801e-05,
+      "loss": 0.9889,
+      "step": 12031
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26112772891647e-05,
+      "loss": 1.0223,
+      "step": 12032
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26023483191214e-05,
+      "loss": 1.0657,
+      "step": 12033
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259341934907809e-05,
+      "loss": 1.054,
+      "step": 12034
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258449037903479e-05,
+      "loss": 1.0866,
+      "step": 12035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257556140899148e-05,
+      "loss": 0.9473,
+      "step": 12036
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.256663243894817e-05,
+      "loss": 0.9638,
+      "step": 12037
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255770346890486e-05,
+      "loss": 1.0326,
+      "step": 12038
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.254877449886155e-05,
+      "loss": 1.1025,
+      "step": 12039
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253984552881826e-05,
+      "loss": 1.1629,
+      "step": 12040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253091655877495e-05,
+      "loss": 0.9004,
+      "step": 12041
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252198758873164e-05,
+      "loss": 1.0126,
+      "step": 12042
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.251305861868834e-05,
+      "loss": 1.2089,
+      "step": 12043
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.250412964864503e-05,
+      "loss": 1.0955,
+      "step": 12044
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.249520067860172e-05,
+      "loss": 1.019,
+      "step": 12045
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.248627170855843e-05,
+      "loss": 0.9845,
+      "step": 12046
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.247734273851512e-05,
+      "loss": 1.0693,
+      "step": 12047
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24684137684718e-05,
+      "loss": 0.989,
+      "step": 12048
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.245948479842851e-05,
+      "loss": 1.0017,
+      "step": 12049
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24505558283852e-05,
+      "loss": 0.8634,
+      "step": 12050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.244162685834189e-05,
+      "loss": 1.2356,
+      "step": 12051
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24326978882986e-05,
+      "loss": 1.0189,
+      "step": 12052
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.242376891825529e-05,
+      "loss": 1.007,
+      "step": 12053
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.241483994821198e-05,
+      "loss": 0.972,
+      "step": 12054
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.240591097816868e-05,
+      "loss": 1.0949,
+      "step": 12055
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.239698200812537e-05,
+      "loss": 1.0685,
+      "step": 12056
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.238805303808206e-05,
+      "loss": 1.1153,
+      "step": 12057
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.237912406803875e-05,
+      "loss": 1.0567,
+      "step": 12058
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.237019509799544e-05,
+      "loss": 0.978,
+      "step": 12059
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.236126612795215e-05,
+      "loss": 0.9134,
+      "step": 12060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.235233715790884e-05,
+      "loss": 1.029,
+      "step": 12061
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.234340818786553e-05,
+      "loss": 1.1538,
+      "step": 12062
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233447921782223e-05,
+      "loss": 1.1706,
+      "step": 12063
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.232555024777892e-05,
+      "loss": 1.0655,
+      "step": 12064
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231662127773561e-05,
+      "loss": 1.0429,
+      "step": 12065
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 1.1175,
+      "step": 12066
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.2298763337649e-05,
+      "loss": 0.9939,
+      "step": 12067
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22898343676057e-05,
+      "loss": 1.0312,
+      "step": 12068
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22809053975624e-05,
+      "loss": 1.1622,
+      "step": 12069
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.227197642751909e-05,
+      "loss": 0.9346,
+      "step": 12070
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226304745747578e-05,
+      "loss": 1.0405,
+      "step": 12071
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225411848743249e-05,
+      "loss": 0.9921,
+      "step": 12072
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.224518951738918e-05,
+      "loss": 1.0658,
+      "step": 12073
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223626054734587e-05,
+      "loss": 0.9741,
+      "step": 12074
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.222733157730257e-05,
+      "loss": 1.0996,
+      "step": 12075
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221840260725926e-05,
+      "loss": 1.1257,
+      "step": 12076
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.220947363721595e-05,
+      "loss": 1.0219,
+      "step": 12077
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.220054466717266e-05,
+      "loss": 1.0179,
+      "step": 12078
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.219161569712935e-05,
+      "loss": 1.2048,
+      "step": 12079
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.218268672708604e-05,
+      "loss": 0.9083,
+      "step": 12080
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.217375775704273e-05,
+      "loss": 1.0413,
+      "step": 12081
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216482878699942e-05,
+      "loss": 0.9693,
+      "step": 12082
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215589981695611e-05,
+      "loss": 1.0253,
+      "step": 12083
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214697084691281e-05,
+      "loss": 0.9194,
+      "step": 12084
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21380418768695e-05,
+      "loss": 0.9832,
+      "step": 12085
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21291129068262e-05,
+      "loss": 1.0223,
+      "step": 12086
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21201839367829e-05,
+      "loss": 1.0637,
+      "step": 12087
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.211125496673959e-05,
+      "loss": 1.0738,
+      "step": 12088
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210232599669628e-05,
+      "loss": 0.9717,
+      "step": 12089
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209339702665298e-05,
+      "loss": 1.063,
+      "step": 12090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208446805660967e-05,
+      "loss": 1.149,
+      "step": 12091
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207553908656636e-05,
+      "loss": 1.2354,
+      "step": 12092
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206661011652307e-05,
+      "loss": 1.1265,
+      "step": 12093
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.205768114647976e-05,
+      "loss": 1.2038,
+      "step": 12094
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204875217643645e-05,
+      "loss": 1.0296,
+      "step": 12095
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203982320639315e-05,
+      "loss": 0.9908,
+      "step": 12096
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203089423634984e-05,
+      "loss": 1.0687,
+      "step": 12097
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.202196526630653e-05,
+      "loss": 1.0129,
+      "step": 12098
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.201303629626324e-05,
+      "loss": 1.074,
+      "step": 12099
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200410732621993e-05,
+      "loss": 0.9935,
+      "step": 12100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.199517835617662e-05,
+      "loss": 1.0045,
+      "step": 12101
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198624938613332e-05,
+      "loss": 1.0492,
+      "step": 12102
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.197732041609001e-05,
+      "loss": 1.0675,
+      "step": 12103
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19683914460467e-05,
+      "loss": 1.0792,
+      "step": 12104
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19594624760034e-05,
+      "loss": 1.0241,
+      "step": 12105
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.195053350596008e-05,
+      "loss": 0.976,
+      "step": 12106
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194160453591678e-05,
+      "loss": 0.9369,
+      "step": 12107
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.193267556587348e-05,
+      "loss": 0.9613,
+      "step": 12108
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192374659583017e-05,
+      "loss": 1.1623,
+      "step": 12109
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191481762578687e-05,
+      "loss": 1.0195,
+      "step": 12110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190588865574356e-05,
+      "loss": 1.0471,
+      "step": 12111
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189695968570025e-05,
+      "loss": 0.9601,
+      "step": 12112
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188803071565696e-05,
+      "loss": 1.0227,
+      "step": 12113
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.187910174561365e-05,
+      "loss": 1.0697,
+      "step": 12114
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.187017277557034e-05,
+      "loss": 0.9526,
+      "step": 12115
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.186124380552704e-05,
+      "loss": 1.0111,
+      "step": 12116
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.185231483548373e-05,
+      "loss": 1.0757,
+      "step": 12117
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184338586544042e-05,
+      "loss": 0.9814,
+      "step": 12118
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183445689539713e-05,
+      "loss": 0.9512,
+      "step": 12119
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182552792535382e-05,
+      "loss": 1.0193,
+      "step": 12120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.181659895531051e-05,
+      "loss": 0.9729,
+      "step": 12121
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.180766998526721e-05,
+      "loss": 1.015,
+      "step": 12122
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17987410152239e-05,
+      "loss": 1.0156,
+      "step": 12123
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17898120451806e-05,
+      "loss": 1.0532,
+      "step": 12124
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178088307513729e-05,
+      "loss": 0.9922,
+      "step": 12125
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177195410509399e-05,
+      "loss": 1.047,
+      "step": 12126
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176302513505068e-05,
+      "loss": 0.9424,
+      "step": 12127
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175409616500737e-05,
+      "loss": 1.0553,
+      "step": 12128
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174516719496406e-05,
+      "loss": 0.9002,
+      "step": 12129
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.173623822492075e-05,
+      "loss": 0.9632,
+      "step": 12130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.172730925487746e-05,
+      "loss": 1.1326,
+      "step": 12131
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.171838028483415e-05,
+      "loss": 1.1171,
+      "step": 12132
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170945131479084e-05,
+      "loss": 0.9898,
+      "step": 12133
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170052234474754e-05,
+      "loss": 1.0229,
+      "step": 12134
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.169159337470423e-05,
+      "loss": 1.0633,
+      "step": 12135
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.168266440466092e-05,
+      "loss": 1.0847,
+      "step": 12136
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.167373543461762e-05,
+      "loss": 0.9996,
+      "step": 12137
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.166480646457432e-05,
+      "loss": 1.1653,
+      "step": 12138
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.1655877494531e-05,
+      "loss": 1.0375,
+      "step": 12139
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.164694852448771e-05,
+      "loss": 0.9985,
+      "step": 12140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.16380195544444e-05,
+      "loss": 0.928,
+      "step": 12141
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.162909058440109e-05,
+      "loss": 0.9601,
+      "step": 12142
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.16201616143578e-05,
+      "loss": 0.9687,
+      "step": 12143
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.161123264431449e-05,
+      "loss": 1.0022,
+      "step": 12144
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.160230367427118e-05,
+      "loss": 0.957,
+      "step": 12145
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.159337470422788e-05,
+      "loss": 1.0051,
+      "step": 12146
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.158444573418457e-05,
+      "loss": 0.9684,
+      "step": 12147
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.157551676414126e-05,
+      "loss": 0.8771,
+      "step": 12148
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.156658779409795e-05,
+      "loss": 1.1277,
+      "step": 12149
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.155765882405464e-05,
+      "loss": 0.9112,
+      "step": 12150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.154872985401135e-05,
+      "loss": 1.2275,
+      "step": 12151
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.153980088396804e-05,
+      "loss": 1.1636,
+      "step": 12152
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.153087191392473e-05,
+      "loss": 0.9919,
+      "step": 12153
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.152194294388142e-05,
+      "loss": 1.1562,
+      "step": 12154
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.151301397383812e-05,
+      "loss": 1.1499,
+      "step": 12155
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.150408500379481e-05,
+      "loss": 1.0673,
+      "step": 12156
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.14951560337515e-05,
+      "loss": 1.018,
+      "step": 12157
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.14862270637082e-05,
+      "loss": 1.2085,
+      "step": 12158
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.14772980936649e-05,
+      "loss": 0.9633,
+      "step": 12159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.14683691236216e-05,
+      "loss": 0.9719,
+      "step": 12160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.145944015357829e-05,
+      "loss": 1.0373,
+      "step": 12161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.145051118353498e-05,
+      "loss": 0.8769,
+      "step": 12162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.144158221349169e-05,
+      "loss": 1.0493,
+      "step": 12163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.143265324344838e-05,
+      "loss": 0.966,
+      "step": 12164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.142372427340507e-05,
+      "loss": 1.079,
+      "step": 12165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.141479530336177e-05,
+      "loss": 0.9523,
+      "step": 12166
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.140586633331846e-05,
+      "loss": 0.9204,
+      "step": 12167
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.139693736327515e-05,
+      "loss": 1.0442,
+      "step": 12168
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.138800839323186e-05,
+      "loss": 0.9609,
+      "step": 12169
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137907942318855e-05,
+      "loss": 1.134,
+      "step": 12170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137015045314524e-05,
+      "loss": 0.9913,
+      "step": 12171
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.136122148310193e-05,
+      "loss": 1.0428,
+      "step": 12172
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135229251305862e-05,
+      "loss": 0.997,
+      "step": 12173
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.134336354301531e-05,
+      "loss": 0.9397,
+      "step": 12174
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133443457297201e-05,
+      "loss": 1.0122,
+      "step": 12175
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13255056029287e-05,
+      "loss": 0.9923,
+      "step": 12176
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13165766328854e-05,
+      "loss": 0.9409,
+      "step": 12177
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13076476628421e-05,
+      "loss": 1.0689,
+      "step": 12178
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129871869279879e-05,
+      "loss": 1.0486,
+      "step": 12179
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.128978972275548e-05,
+      "loss": 0.9763,
+      "step": 12180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.128086075271218e-05,
+      "loss": 1.0289,
+      "step": 12181
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.127193178266887e-05,
+      "loss": 0.9312,
+      "step": 12182
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.126300281262556e-05,
+      "loss": 0.9641,
+      "step": 12183
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125407384258227e-05,
+      "loss": 0.9348,
+      "step": 12184
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.124514487253896e-05,
+      "loss": 1.0811,
+      "step": 12185
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.123621590249565e-05,
+      "loss": 1.0788,
+      "step": 12186
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122728693245235e-05,
+      "loss": 0.9083,
+      "step": 12187
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.121835796240904e-05,
+      "loss": 0.9601,
+      "step": 12188
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120942899236573e-05,
+      "loss": 0.9735,
+      "step": 12189
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120050002232244e-05,
+      "loss": 0.9466,
+      "step": 12190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.119157105227913e-05,
+      "loss": 0.9769,
+      "step": 12191
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118264208223582e-05,
+      "loss": 1.2111,
+      "step": 12192
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.117371311219252e-05,
+      "loss": 0.8939,
+      "step": 12193
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116478414214921e-05,
+      "loss": 1.0568,
+      "step": 12194
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11558551721059e-05,
+      "loss": 1.1672,
+      "step": 12195
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11469262020626e-05,
+      "loss": 1.0052,
+      "step": 12196
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113799723201928e-05,
+      "loss": 1.1682,
+      "step": 12197
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112906826197597e-05,
+      "loss": 0.9789,
+      "step": 12198
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112013929193268e-05,
+      "loss": 0.9963,
+      "step": 12199
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111121032188937e-05,
+      "loss": 1.0497,
+      "step": 12200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110228135184606e-05,
+      "loss": 0.9539,
+      "step": 12201
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109335238180276e-05,
+      "loss": 1.0659,
+      "step": 12202
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.108442341175945e-05,
+      "loss": 1.0822,
+      "step": 12203
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.107549444171614e-05,
+      "loss": 1.0603,
+      "step": 12204
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106656547167285e-05,
+      "loss": 1.0104,
+      "step": 12205
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105763650162954e-05,
+      "loss": 1.1546,
+      "step": 12206
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104870753158623e-05,
+      "loss": 0.9941,
+      "step": 12207
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103977856154293e-05,
+      "loss": 1.0518,
+      "step": 12208
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103084959149962e-05,
+      "loss": 1.0374,
+      "step": 12209
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102192062145633e-05,
+      "loss": 0.9666,
+      "step": 12210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.101299165141302e-05,
+      "loss": 0.9865,
+      "step": 12211
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.100406268136971e-05,
+      "loss": 0.9405,
+      "step": 12212
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.099513371132641e-05,
+      "loss": 1.1108,
+      "step": 12213
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.09862047412831e-05,
+      "loss": 1.0821,
+      "step": 12214
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.09772757712398e-05,
+      "loss": 1.0897,
+      "step": 12215
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.096834680119648e-05,
+      "loss": 0.9396,
+      "step": 12216
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.095941783115319e-05,
+      "loss": 1.0037,
+      "step": 12217
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.095048886110988e-05,
+      "loss": 1.0858,
+      "step": 12218
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.094155989106657e-05,
+      "loss": 0.9454,
+      "step": 12219
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.093263092102326e-05,
+      "loss": 0.906,
+      "step": 12220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.092370195097995e-05,
+      "loss": 0.9848,
+      "step": 12221
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.091477298093665e-05,
+      "loss": 1.1292,
+      "step": 12222
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.090584401089335e-05,
+      "loss": 0.9802,
+      "step": 12223
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.089691504085004e-05,
+      "loss": 1.0398,
+      "step": 12224
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.088798607080674e-05,
+      "loss": 1.0619,
+      "step": 12225
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.087905710076343e-05,
+      "loss": 0.9885,
+      "step": 12226
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.087012813072012e-05,
+      "loss": 1.0058,
+      "step": 12227
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.086119916067682e-05,
+      "loss": 1.3351,
+      "step": 12228
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.085227019063352e-05,
+      "loss": 1.0564,
+      "step": 12229
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.08433412205902e-05,
+      "loss": 0.9355,
+      "step": 12230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.083441225054691e-05,
+      "loss": 1.024,
+      "step": 12231
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.08254832805036e-05,
+      "loss": 1.0343,
+      "step": 12232
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.081655431046029e-05,
+      "loss": 0.9947,
+      "step": 12233
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.0807625340417e-05,
+      "loss": 1.0722,
+      "step": 12234
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.079869637037369e-05,
+      "loss": 1.0073,
+      "step": 12235
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.078976740033038e-05,
+      "loss": 0.9555,
+      "step": 12236
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.078083843028708e-05,
+      "loss": 0.8982,
+      "step": 12237
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.077190946024377e-05,
+      "loss": 1.1378,
+      "step": 12238
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.076298049020046e-05,
+      "loss": 1.018,
+      "step": 12239
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.075405152015715e-05,
+      "loss": 0.9409,
+      "step": 12240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.074512255011384e-05,
+      "loss": 1.1148,
+      "step": 12241
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.073619358007053e-05,
+      "loss": 0.8862,
+      "step": 12242
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.072726461002724e-05,
+      "loss": 1.0329,
+      "step": 12243
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.071833563998393e-05,
+      "loss": 0.9638,
+      "step": 12244
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.070940666994062e-05,
+      "loss": 1.1076,
+      "step": 12245
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.070047769989732e-05,
+      "loss": 0.9387,
+      "step": 12246
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.069154872985401e-05,
+      "loss": 1.0011,
+      "step": 12247
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.06826197598107e-05,
+      "loss": 0.9463,
+      "step": 12248
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.06736907897674e-05,
+      "loss": 1.0153,
+      "step": 12249
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.06647618197241e-05,
+      "loss": 1.0736,
+      "step": 12250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.065583284968079e-05,
+      "loss": 1.0498,
+      "step": 12251
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.064690387963749e-05,
+      "loss": 0.9767,
+      "step": 12252
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.063797490959418e-05,
+      "loss": 1.03,
+      "step": 12253
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.062904593955087e-05,
+      "loss": 1.0461,
+      "step": 12254
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.062011696950758e-05,
+      "loss": 1.1647,
+      "step": 12255
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.061118799946427e-05,
+      "loss": 1.1752,
+      "step": 12256
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.060225902942097e-05,
+      "loss": 1.0927,
+      "step": 12257
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.059333005937766e-05,
+      "loss": 1.0403,
+      "step": 12258
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.058440108933435e-05,
+      "loss": 0.9893,
+      "step": 12259
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.057547211929106e-05,
+      "loss": 1.0284,
+      "step": 12260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.056654314924775e-05,
+      "loss": 1.1738,
+      "step": 12261
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.055761417920444e-05,
+      "loss": 1.0776,
+      "step": 12262
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.054868520916113e-05,
+      "loss": 0.9866,
+      "step": 12263
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.053975623911782e-05,
+      "loss": 1.0425,
+      "step": 12264
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.053082726907451e-05,
+      "loss": 0.9674,
+      "step": 12265
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.052189829903121e-05,
+      "loss": 0.9568,
+      "step": 12266
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.05129693289879e-05,
+      "loss": 1.0203,
+      "step": 12267
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.050404035894459e-05,
+      "loss": 0.9191,
+      "step": 12268
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.04951113889013e-05,
+      "loss": 0.9727,
+      "step": 12269
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.048618241885799e-05,
+      "loss": 1.0977,
+      "step": 12270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.047725344881468e-05,
+      "loss": 1.3091,
+      "step": 12271
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.046832447877138e-05,
+      "loss": 1.0616,
+      "step": 12272
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.045939550872807e-05,
+      "loss": 0.9142,
+      "step": 12273
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.045046653868476e-05,
+      "loss": 0.9854,
+      "step": 12274
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.044153756864147e-05,
+      "loss": 0.9352,
+      "step": 12275
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.043260859859816e-05,
+      "loss": 1.0483,
+      "step": 12276
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.042367962855485e-05,
+      "loss": 1.0238,
+      "step": 12277
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.041475065851155e-05,
+      "loss": 1.1385,
+      "step": 12278
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.040582168846824e-05,
+      "loss": 0.9076,
+      "step": 12279
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.039689271842493e-05,
+      "loss": 0.893,
+      "step": 12280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.038796374838164e-05,
+      "loss": 1.0896,
+      "step": 12281
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.037903477833833e-05,
+      "loss": 1.0159,
+      "step": 12282
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.037010580829502e-05,
+      "loss": 1.0141,
+      "step": 12283
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.036117683825172e-05,
+      "loss": 1.0317,
+      "step": 12284
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.035224786820841e-05,
+      "loss": 1.0414,
+      "step": 12285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.03433188981651e-05,
+      "loss": 1.0658,
+      "step": 12286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.03343899281218e-05,
+      "loss": 0.9342,
+      "step": 12287
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.032546095807848e-05,
+      "loss": 1.0869,
+      "step": 12288
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.031653198803517e-05,
+      "loss": 1.0736,
+      "step": 12289
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.030760301799188e-05,
+      "loss": 0.983,
+      "step": 12290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.029867404794857e-05,
+      "loss": 1.0595,
+      "step": 12291
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.028974507790526e-05,
+      "loss": 1.0595,
+      "step": 12292
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.028081610786196e-05,
+      "loss": 0.9903,
+      "step": 12293
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.027188713781865e-05,
+      "loss": 0.9811,
+      "step": 12294
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.026295816777534e-05,
+      "loss": 1.0839,
+      "step": 12295
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.025402919773205e-05,
+      "loss": 1.1815,
+      "step": 12296
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.024510022768874e-05,
+      "loss": 1.0958,
+      "step": 12297
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.023617125764543e-05,
+      "loss": 1.0952,
+      "step": 12298
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.022724228760213e-05,
+      "loss": 0.9648,
+      "step": 12299
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.021831331755882e-05,
+      "loss": 1.0317,
+      "step": 12300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.020938434751551e-05,
+      "loss": 0.9129,
+      "step": 12301
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.020045537747222e-05,
+      "loss": 1.0786,
+      "step": 12302
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.019152640742891e-05,
+      "loss": 1.1111,
+      "step": 12303
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.01825974373856e-05,
+      "loss": 1.0496,
+      "step": 12304
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.01736684673423e-05,
+      "loss": 0.9399,
+      "step": 12305
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.0164739497299e-05,
+      "loss": 1.1137,
+      "step": 12306
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.015581052725568e-05,
+      "loss": 0.9185,
+      "step": 12307
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.014688155721237e-05,
+      "loss": 0.9299,
+      "step": 12308
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.013795258716908e-05,
+      "loss": 1.0433,
+      "step": 12309
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.012902361712577e-05,
+      "loss": 0.9029,
+      "step": 12310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.012009464708246e-05,
+      "loss": 1.0049,
+      "step": 12311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.011116567703915e-05,
+      "loss": 1.0925,
+      "step": 12312
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.010223670699585e-05,
+      "loss": 1.018,
+      "step": 12313
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.009330773695254e-05,
+      "loss": 0.9166,
+      "step": 12314
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.008437876690924e-05,
+      "loss": 1.0456,
+      "step": 12315
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.007544979686594e-05,
+      "loss": 1.0149,
+      "step": 12316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.006652082682263e-05,
+      "loss": 1.0843,
+      "step": 12317
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.005759185677932e-05,
+      "loss": 0.8924,
+      "step": 12318
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.004866288673602e-05,
+      "loss": 1.0513,
+      "step": 12319
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.003973391669271e-05,
+      "loss": 1.0688,
+      "step": 12320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.00308049466494e-05,
+      "loss": 0.8923,
+      "step": 12321
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.002187597660611e-05,
+      "loss": 1.1319,
+      "step": 12322
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.00129470065628e-05,
+      "loss": 0.9706,
+      "step": 12323
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.000401803651949e-05,
+      "loss": 1.022,
+      "step": 12324
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.99950890664762e-05,
+      "loss": 1.0506,
+      "step": 12325
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.998616009643288e-05,
+      "loss": 1.0515,
+      "step": 12326
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.997723112638958e-05,
+      "loss": 1.1788,
+      "step": 12327
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.996830215634628e-05,
+      "loss": 1.0615,
+      "step": 12328
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.995937318630297e-05,
+      "loss": 0.9162,
+      "step": 12329
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.995044421625966e-05,
+      "loss": 0.9706,
+      "step": 12330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.994151524621635e-05,
+      "loss": 1.0607,
+      "step": 12331
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.993258627617304e-05,
+      "loss": 0.9703,
+      "step": 12332
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.992365730612973e-05,
+      "loss": 0.9948,
+      "step": 12333
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.991472833608644e-05,
+      "loss": 1.2369,
+      "step": 12334
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.990579936604313e-05,
+      "loss": 1.0803,
+      "step": 12335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.989687039599982e-05,
+      "loss": 0.8794,
+      "step": 12336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.988794142595652e-05,
+      "loss": 1.0736,
+      "step": 12337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.987901245591321e-05,
+      "loss": 1.1147,
+      "step": 12338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.98700834858699e-05,
+      "loss": 0.8594,
+      "step": 12339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.98611545158266e-05,
+      "loss": 1.0479,
+      "step": 12340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.98522255457833e-05,
+      "loss": 0.9935,
+      "step": 12341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.984329657573999e-05,
+      "loss": 1.0421,
+      "step": 12342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.983436760569669e-05,
+      "loss": 1.0072,
+      "step": 12343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.982543863565338e-05,
+      "loss": 1.0292,
+      "step": 12344
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.981650966561007e-05,
+      "loss": 1.1236,
+      "step": 12345
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.980758069556678e-05,
+      "loss": 0.8671,
+      "step": 12346
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.979865172552347e-05,
+      "loss": 1.0806,
+      "step": 12347
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.978972275548016e-05,
+      "loss": 1.0,
+      "step": 12348
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.978079378543686e-05,
+      "loss": 0.9439,
+      "step": 12349
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.977186481539355e-05,
+      "loss": 1.0614,
+      "step": 12350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.976293584535024e-05,
+      "loss": 1.1089,
+      "step": 12351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.975400687530695e-05,
+      "loss": 0.9529,
+      "step": 12352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.974507790526364e-05,
+      "loss": 1.0343,
+      "step": 12353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.973614893522033e-05,
+      "loss": 1.1248,
+      "step": 12354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.972721996517702e-05,
+      "loss": 0.962,
+      "step": 12355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.971829099513371e-05,
+      "loss": 1.0836,
+      "step": 12356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.970936202509041e-05,
+      "loss": 1.0265,
+      "step": 12357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.97004330550471e-05,
+      "loss": 1.0476,
+      "step": 12358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.969150408500379e-05,
+      "loss": 0.9688,
+      "step": 12359
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.96825751149605e-05,
+      "loss": 1.0592,
+      "step": 12360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.967364614491719e-05,
+      "loss": 1.0798,
+      "step": 12361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.966471717487388e-05,
+      "loss": 0.9271,
+      "step": 12362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.965578820483058e-05,
+      "loss": 1.019,
+      "step": 12363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.964685923478727e-05,
+      "loss": 1.002,
+      "step": 12364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.963793026474396e-05,
+      "loss": 1.122,
+      "step": 12365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.962900129470067e-05,
+      "loss": 1.1727,
+      "step": 12366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.962007232465736e-05,
+      "loss": 1.0639,
+      "step": 12367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.961114335461405e-05,
+      "loss": 1.0675,
+      "step": 12368
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.960221438457075e-05,
+      "loss": 1.0322,
+      "step": 12369
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.959328541452744e-05,
+      "loss": 1.0126,
+      "step": 12370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.958435644448413e-05,
+      "loss": 1.0626,
+      "step": 12371
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.957542747444084e-05,
+      "loss": 0.9025,
+      "step": 12372
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.956649850439753e-05,
+      "loss": 1.0271,
+      "step": 12373
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.955756953435422e-05,
+      "loss": 0.9296,
+      "step": 12374
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.954864056431092e-05,
+      "loss": 1.0951,
+      "step": 12375
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.953971159426761e-05,
+      "loss": 1.1191,
+      "step": 12376
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.95307826242243e-05,
+      "loss": 1.0284,
+      "step": 12377
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.952185365418099e-05,
+      "loss": 1.0123,
+      "step": 12378
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.951292468413768e-05,
+      "loss": 1.0908,
+      "step": 12379
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.950399571409437e-05,
+      "loss": 0.9933,
+      "step": 12380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.949506674405108e-05,
+      "loss": 1.1127,
+      "step": 12381
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.948613777400777e-05,
+      "loss": 0.9743,
+      "step": 12382
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.947720880396446e-05,
+      "loss": 1.0976,
+      "step": 12383
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.946827983392116e-05,
+      "loss": 1.1008,
+      "step": 12384
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.945935086387785e-05,
+      "loss": 1.0492,
+      "step": 12385
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.945042189383454e-05,
+      "loss": 1.1386,
+      "step": 12386
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.944149292379125e-05,
+      "loss": 0.9946,
+      "step": 12387
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.943256395374794e-05,
+      "loss": 0.9666,
+      "step": 12388
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.942363498370463e-05,
+      "loss": 1.1041,
+      "step": 12389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.941470601366133e-05,
+      "loss": 0.9752,
+      "step": 12390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.940577704361802e-05,
+      "loss": 1.0128,
+      "step": 12391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.939684807357471e-05,
+      "loss": 1.0451,
+      "step": 12392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.938791910353142e-05,
+      "loss": 0.9722,
+      "step": 12393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.937899013348811e-05,
+      "loss": 1.1389,
+      "step": 12394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.93700611634448e-05,
+      "loss": 1.1357,
+      "step": 12395
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.93611321934015e-05,
+      "loss": 1.061,
+      "step": 12396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.93522032233582e-05,
+      "loss": 0.9219,
+      "step": 12397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.934327425331488e-05,
+      "loss": 0.986,
+      "step": 12398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.933434528327157e-05,
+      "loss": 1.0217,
+      "step": 12399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.932541631322828e-05,
+      "loss": 1.0505,
+      "step": 12400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.931648734318497e-05,
+      "loss": 1.0014,
+      "step": 12401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.930755837314166e-05,
+      "loss": 0.9435,
+      "step": 12402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.929862940309835e-05,
+      "loss": 0.9034,
+      "step": 12403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.928970043305505e-05,
+      "loss": 0.9906,
+      "step": 12404
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.928077146301174e-05,
+      "loss": 1.0636,
+      "step": 12405
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.927184249296843e-05,
+      "loss": 1.0114,
+      "step": 12406
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.926291352292514e-05,
+      "loss": 1.0144,
+      "step": 12407
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.925398455288183e-05,
+      "loss": 1.0019,
+      "step": 12408
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.924505558283852e-05,
+      "loss": 1.1315,
+      "step": 12409
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.923612661279522e-05,
+      "loss": 1.0371,
+      "step": 12410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.922719764275191e-05,
+      "loss": 0.941,
+      "step": 12411
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.92182686727086e-05,
+      "loss": 1.014,
+      "step": 12412
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.920933970266531e-05,
+      "loss": 0.9501,
+      "step": 12413
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.9200410732622e-05,
+      "loss": 1.0491,
+      "step": 12414
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.919148176257869e-05,
+      "loss": 0.9734,
+      "step": 12415
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.91825527925354e-05,
+      "loss": 0.9067,
+      "step": 12416
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.917362382249208e-05,
+      "loss": 0.9963,
+      "step": 12417
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.916469485244877e-05,
+      "loss": 0.9964,
+      "step": 12418
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.915576588240548e-05,
+      "loss": 0.993,
+      "step": 12419
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.914683691236217e-05,
+      "loss": 0.9737,
+      "step": 12420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.913790794231886e-05,
+      "loss": 1.1044,
+      "step": 12421
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.912897897227555e-05,
+      "loss": 1.1095,
+      "step": 12422
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.912005000223224e-05,
+      "loss": 1.097,
+      "step": 12423
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.911112103218893e-05,
+      "loss": 0.9472,
+      "step": 12424
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.910219206214564e-05,
+      "loss": 1.0287,
+      "step": 12425
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.909326309210233e-05,
+      "loss": 1.0937,
+      "step": 12426
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.908433412205902e-05,
+      "loss": 1.034,
+      "step": 12427
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.907540515201572e-05,
+      "loss": 1.033,
+      "step": 12428
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.906647618197241e-05,
+      "loss": 0.9762,
+      "step": 12429
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.90575472119291e-05,
+      "loss": 0.9782,
+      "step": 12430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.90486182418858e-05,
+      "loss": 0.9735,
+      "step": 12431
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.90396892718425e-05,
+      "loss": 1.001,
+      "step": 12432
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.903076030179919e-05,
+      "loss": 1.036,
+      "step": 12433
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.902183133175589e-05,
+      "loss": 1.0668,
+      "step": 12434
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.901290236171258e-05,
+      "loss": 0.9858,
+      "step": 12435
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.900397339166927e-05,
+      "loss": 1.1231,
+      "step": 12436
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.899504442162598e-05,
+      "loss": 1.0558,
+      "step": 12437
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.898611545158267e-05,
+      "loss": 1.0605,
+      "step": 12438
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.897718648153936e-05,
+      "loss": 1.1571,
+      "step": 12439
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.896825751149606e-05,
+      "loss": 1.0003,
+      "step": 12440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.895932854145275e-05,
+      "loss": 0.9666,
+      "step": 12441
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.895039957140944e-05,
+      "loss": 1.1599,
+      "step": 12442
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.894147060136615e-05,
+      "loss": 0.9442,
+      "step": 12443
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.893254163132284e-05,
+      "loss": 0.9835,
+      "step": 12444
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.892361266127953e-05,
+      "loss": 0.9988,
+      "step": 12445
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.891468369123622e-05,
+      "loss": 1.0089,
+      "step": 12446
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.890575472119291e-05,
+      "loss": 1.0305,
+      "step": 12447
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.88968257511496e-05,
+      "loss": 0.9188,
+      "step": 12448
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.88878967811063e-05,
+      "loss": 1.0771,
+      "step": 12449
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.887896781106299e-05,
+      "loss": 1.0069,
+      "step": 12450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.887003884101968e-05,
+      "loss": 0.9701,
+      "step": 12451
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.886110987097639e-05,
+      "loss": 0.9576,
+      "step": 12452
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.885218090093308e-05,
+      "loss": 1.0496,
+      "step": 12453
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.884325193088978e-05,
+      "loss": 0.9249,
+      "step": 12454
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.883432296084647e-05,
+      "loss": 1.0644,
+      "step": 12455
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.882539399080316e-05,
+      "loss": 1.2601,
+      "step": 12456
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.881646502075987e-05,
+      "loss": 1.0988,
+      "step": 12457
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.880753605071656e-05,
+      "loss": 0.9788,
+      "step": 12458
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.879860708067325e-05,
+      "loss": 1.0649,
+      "step": 12459
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.878967811062995e-05,
+      "loss": 1.1336,
+      "step": 12460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.878074914058664e-05,
+      "loss": 0.9977,
+      "step": 12461
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.877182017054333e-05,
+      "loss": 1.1162,
+      "step": 12462
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.876289120050004e-05,
+      "loss": 0.9614,
+      "step": 12463
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.875396223045673e-05,
+      "loss": 1.0457,
+      "step": 12464
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.874503326041342e-05,
+      "loss": 1.1829,
+      "step": 12465
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.873610429037012e-05,
+      "loss": 1.0832,
+      "step": 12466
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.872717532032681e-05,
+      "loss": 1.0011,
+      "step": 12467
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.87182463502835e-05,
+      "loss": 0.9339,
+      "step": 12468
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.870931738024019e-05,
+      "loss": 1.1638,
+      "step": 12469
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.870038841019688e-05,
+      "loss": 0.9611,
+      "step": 12470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.869145944015357e-05,
+      "loss": 0.9736,
+      "step": 12471
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.868253047011028e-05,
+      "loss": 1.0278,
+      "step": 12472
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.867360150006697e-05,
+      "loss": 0.9768,
+      "step": 12473
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.866467253002366e-05,
+      "loss": 0.9763,
+      "step": 12474
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.865574355998036e-05,
+      "loss": 1.0731,
+      "step": 12475
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.864681458993705e-05,
+      "loss": 1.0939,
+      "step": 12476
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.863788561989374e-05,
+      "loss": 1.0478,
+      "step": 12477
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.862895664985045e-05,
+      "loss": 1.1215,
+      "step": 12478
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.862002767980714e-05,
+      "loss": 1.0452,
+      "step": 12479
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.861109870976383e-05,
+      "loss": 1.1335,
+      "step": 12480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.860216973972053e-05,
+      "loss": 1.1703,
+      "step": 12481
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.859324076967722e-05,
+      "loss": 0.9963,
+      "step": 12482
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.858431179963391e-05,
+      "loss": 1.1564,
+      "step": 12483
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.857538282959062e-05,
+      "loss": 1.1258,
+      "step": 12484
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.856645385954731e-05,
+      "loss": 1.0251,
+      "step": 12485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.8557524889504e-05,
+      "loss": 0.9646,
+      "step": 12486
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.85485959194607e-05,
+      "loss": 0.9959,
+      "step": 12487
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.853966694941739e-05,
+      "loss": 0.9969,
+      "step": 12488
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.853073797937408e-05,
+      "loss": 0.9701,
+      "step": 12489
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.852180900933077e-05,
+      "loss": 1.092,
+      "step": 12490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.851288003928748e-05,
+      "loss": 1.037,
+      "step": 12491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.850395106924417e-05,
+      "loss": 1.0789,
+      "step": 12492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.849502209920086e-05,
+      "loss": 0.9894,
+      "step": 12493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.848609312915755e-05,
+      "loss": 1.0024,
+      "step": 12494
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.847716415911424e-05,
+      "loss": 1.0946,
+      "step": 12495
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.846823518907094e-05,
+      "loss": 1.1393,
+      "step": 12496
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.845930621902763e-05,
+      "loss": 1.0476,
+      "step": 12497
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.845037724898432e-05,
+      "loss": 1.1733,
+      "step": 12498
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.844144827894103e-05,
+      "loss": 1.0435,
+      "step": 12499
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.843251930889772e-05,
+      "loss": 1.0696,
+      "step": 12500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.842359033885442e-05,
+      "loss": 1.0286,
+      "step": 12501
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.841466136881111e-05,
+      "loss": 0.9789,
+      "step": 12502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.84057323987678e-05,
+      "loss": 0.9669,
+      "step": 12503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.839680342872451e-05,
+      "loss": 1.0692,
+      "step": 12504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.83878744586812e-05,
+      "loss": 1.1127,
+      "step": 12505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.837894548863789e-05,
+      "loss": 1.0646,
+      "step": 12506
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.83700165185946e-05,
+      "loss": 1.1425,
+      "step": 12507
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.836108754855128e-05,
+      "loss": 0.9814,
+      "step": 12508
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.835215857850797e-05,
+      "loss": 0.9822,
+      "step": 12509
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.834322960846468e-05,
+      "loss": 1.1312,
+      "step": 12510
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.833430063842137e-05,
+      "loss": 0.8768,
+      "step": 12511
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.832537166837806e-05,
+      "loss": 1.0085,
+      "step": 12512
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.831644269833475e-05,
+      "loss": 1.0929,
+      "step": 12513
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.830751372829144e-05,
+      "loss": 1.0411,
+      "step": 12514
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.829858475824813e-05,
+      "loss": 0.8883,
+      "step": 12515
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.828965578820483e-05,
+      "loss": 1.1256,
+      "step": 12516
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.828072681816153e-05,
+      "loss": 0.9522,
+      "step": 12517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.827179784811822e-05,
+      "loss": 1.0232,
+      "step": 12518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.826286887807492e-05,
+      "loss": 1.0298,
+      "step": 12519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.825393990803161e-05,
+      "loss": 1.0668,
+      "step": 12520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.82450109379883e-05,
+      "loss": 0.9467,
+      "step": 12521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.8236081967945e-05,
+      "loss": 1.1452,
+      "step": 12522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.82271529979017e-05,
+      "loss": 1.0387,
+      "step": 12523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.821822402785839e-05,
+      "loss": 0.9522,
+      "step": 12524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.820929505781509e-05,
+      "loss": 0.9673,
+      "step": 12525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.820036608777178e-05,
+      "loss": 1.0617,
+      "step": 12526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.819143711772847e-05,
+      "loss": 1.1167,
+      "step": 12527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.818250814768517e-05,
+      "loss": 1.0332,
+      "step": 12528
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.817357917764187e-05,
+      "loss": 1.018,
+      "step": 12529
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.816465020759856e-05,
+      "loss": 1.0237,
+      "step": 12530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.815572123755526e-05,
+      "loss": 1.0901,
+      "step": 12531
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.814679226751195e-05,
+      "loss": 0.9985,
+      "step": 12532
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.813786329746864e-05,
+      "loss": 1.1238,
+      "step": 12533
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.812893432742534e-05,
+      "loss": 1.0466,
+      "step": 12534
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.812000535738204e-05,
+      "loss": 0.937,
+      "step": 12535
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.811107638733873e-05,
+      "loss": 0.9822,
+      "step": 12536
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.810214741729542e-05,
+      "loss": 1.116,
+      "step": 12537
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.809321844725211e-05,
+      "loss": 1.0781,
+      "step": 12538
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.80842894772088e-05,
+      "loss": 0.9842,
+      "step": 12539
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.80753605071655e-05,
+      "loss": 0.9335,
+      "step": 12540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.806643153712219e-05,
+      "loss": 1.1282,
+      "step": 12541
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.805750256707888e-05,
+      "loss": 1.0029,
+      "step": 12542
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.804857359703559e-05,
+      "loss": 1.0634,
+      "step": 12543
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.803964462699228e-05,
+      "loss": 1.1438,
+      "step": 12544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.803071565694897e-05,
+      "loss": 1.0076,
+      "step": 12545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.802178668690567e-05,
+      "loss": 0.9241,
+      "step": 12546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.801285771686236e-05,
+      "loss": 1.0584,
+      "step": 12547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.800392874681905e-05,
+      "loss": 1.0807,
+      "step": 12548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.799499977677576e-05,
+      "loss": 1.1207,
+      "step": 12549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.798607080673245e-05,
+      "loss": 1.0349,
+      "step": 12550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.797714183668915e-05,
+      "loss": 1.0971,
+      "step": 12551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.796821286664584e-05,
+      "loss": 1.0993,
+      "step": 12552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.795928389660253e-05,
+      "loss": 0.9826,
+      "step": 12553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.795035492655924e-05,
+      "loss": 1.0436,
+      "step": 12554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.794142595651593e-05,
+      "loss": 1.1312,
+      "step": 12555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.793249698647262e-05,
+      "loss": 1.0458,
+      "step": 12556
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.792356801642932e-05,
+      "loss": 0.9989,
+      "step": 12557
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.791463904638601e-05,
+      "loss": 0.9955,
+      "step": 12558
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.79057100763427e-05,
+      "loss": 0.9862,
+      "step": 12559
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.789678110629939e-05,
+      "loss": 1.0897,
+      "step": 12560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.788785213625608e-05,
+      "loss": 1.2463,
+      "step": 12561
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.787892316621277e-05,
+      "loss": 0.9488,
+      "step": 12562
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.786999419616948e-05,
+      "loss": 0.9622,
+      "step": 12563
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.786106522612617e-05,
+      "loss": 1.0271,
+      "step": 12564
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.785213625608286e-05,
+      "loss": 0.9891,
+      "step": 12565
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.784320728603956e-05,
+      "loss": 1.1079,
+      "step": 12566
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.783427831599625e-05,
+      "loss": 1.0777,
+      "step": 12567
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.782534934595294e-05,
+      "loss": 1.0382,
+      "step": 12568
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.781642037590965e-05,
+      "loss": 1.1003,
+      "step": 12569
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.780749140586634e-05,
+      "loss": 0.9938,
+      "step": 12570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.779856243582303e-05,
+      "loss": 0.997,
+      "step": 12571
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.778963346577973e-05,
+      "loss": 1.0408,
+      "step": 12572
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.778070449573642e-05,
+      "loss": 1.0421,
+      "step": 12573
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.777177552569311e-05,
+      "loss": 1.0398,
+      "step": 12574
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.776284655564982e-05,
+      "loss": 1.0783,
+      "step": 12575
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.775391758560651e-05,
+      "loss": 0.9056,
+      "step": 12576
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.77449886155632e-05,
+      "loss": 1.0441,
+      "step": 12577
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.77360596455199e-05,
+      "loss": 0.9807,
+      "step": 12578
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.772713067547659e-05,
+      "loss": 1.0776,
+      "step": 12579
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.771820170543328e-05,
+      "loss": 0.9703,
+      "step": 12580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.770927273538997e-05,
+      "loss": 1.0927,
+      "step": 12581
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.770034376534666e-05,
+      "loss": 1.0036,
+      "step": 12582
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.769141479530337e-05,
+      "loss": 1.1917,
+      "step": 12583
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.768248582526006e-05,
+      "loss": 1.0637,
+      "step": 12584
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.767355685521675e-05,
+      "loss": 1.0481,
+      "step": 12585
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.766462788517344e-05,
+      "loss": 1.0691,
+      "step": 12586
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.765569891513014e-05,
+      "loss": 1.2001,
+      "step": 12587
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.764676994508683e-05,
+      "loss": 1.051,
+      "step": 12588
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.763784097504352e-05,
+      "loss": 1.0504,
+      "step": 12589
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.762891200500023e-05,
+      "loss": 0.9019,
+      "step": 12590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.761998303495692e-05,
+      "loss": 1.116,
+      "step": 12591
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.761105406491361e-05,
+      "loss": 1.0306,
+      "step": 12592
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.760212509487031e-05,
+      "loss": 1.0895,
+      "step": 12593
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.7593196124827e-05,
+      "loss": 0.9937,
+      "step": 12594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.75842671547837e-05,
+      "loss": 1.0383,
+      "step": 12595
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.75753381847404e-05,
+      "loss": 1.0473,
+      "step": 12596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.756640921469709e-05,
+      "loss": 0.9637,
+      "step": 12597
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.755748024465378e-05,
+      "loss": 0.9214,
+      "step": 12598
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.754855127461048e-05,
+      "loss": 0.9614,
+      "step": 12599
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.753962230456717e-05,
+      "loss": 1.0782,
+      "step": 12600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.753069333452388e-05,
+      "loss": 0.9646,
+      "step": 12601
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.752176436448057e-05,
+      "loss": 0.9988,
+      "step": 12602
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.751283539443726e-05,
+      "loss": 1.0261,
+      "step": 12603
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.750390642439395e-05,
+      "loss": 1.1066,
+      "step": 12604
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.749497745435064e-05,
+      "loss": 0.9059,
+      "step": 12605
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.748604848430733e-05,
+      "loss": 0.9956,
+      "step": 12606
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.747711951426403e-05,
+      "loss": 0.9402,
+      "step": 12607
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.746819054422073e-05,
+      "loss": 1.0164,
+      "step": 12608
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.745926157417742e-05,
+      "loss": 1.0612,
+      "step": 12609
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.745033260413412e-05,
+      "loss": 1.0144,
+      "step": 12610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.744140363409081e-05,
+      "loss": 1.0183,
+      "step": 12611
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.74324746640475e-05,
+      "loss": 1.0443,
+      "step": 12612
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.74235456940042e-05,
+      "loss": 1.1645,
+      "step": 12613
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.74146167239609e-05,
+      "loss": 1.1014,
+      "step": 12614
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.740568775391759e-05,
+      "loss": 1.0204,
+      "step": 12615
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.739675878387429e-05,
+      "loss": 1.141,
+      "step": 12616
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.738782981383098e-05,
+      "loss": 1.0658,
+      "step": 12617
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.737890084378767e-05,
+      "loss": 0.9885,
+      "step": 12618
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.736997187374437e-05,
+      "loss": 1.1099,
+      "step": 12619
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.736104290370107e-05,
+      "loss": 1.1042,
+      "step": 12620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.735211393365776e-05,
+      "loss": 1.0042,
+      "step": 12621
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.734318496361446e-05,
+      "loss": 0.9635,
+      "step": 12622
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.733425599357115e-05,
+      "loss": 1.0959,
+      "step": 12623
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.732532702352784e-05,
+      "loss": 1.0499,
+      "step": 12624
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.731639805348454e-05,
+      "loss": 1.1298,
+      "step": 12625
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.730746908344124e-05,
+      "loss": 1.0337,
+      "step": 12626
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.729854011339793e-05,
+      "loss": 1.0542,
+      "step": 12627
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.728961114335462e-05,
+      "loss": 0.9318,
+      "step": 12628
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.72806821733113e-05,
+      "loss": 0.8836,
+      "step": 12629
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.7271753203268e-05,
+      "loss": 0.9597,
+      "step": 12630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.72628242332247e-05,
+      "loss": 0.9679,
+      "step": 12631
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.725389526318139e-05,
+      "loss": 1.0908,
+      "step": 12632
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.724496629313808e-05,
+      "loss": 0.9913,
+      "step": 12633
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.723603732309479e-05,
+      "loss": 0.9334,
+      "step": 12634
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.722710835305148e-05,
+      "loss": 1.003,
+      "step": 12635
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.721817938300817e-05,
+      "loss": 1.1059,
+      "step": 12636
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.720925041296487e-05,
+      "loss": 0.9756,
+      "step": 12637
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.720032144292156e-05,
+      "loss": 1.066,
+      "step": 12638
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.719139247287825e-05,
+      "loss": 0.9571,
+      "step": 12639
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.718246350283496e-05,
+      "loss": 1.0262,
+      "step": 12640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.717353453279165e-05,
+      "loss": 1.1372,
+      "step": 12641
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.716460556274834e-05,
+      "loss": 0.9387,
+      "step": 12642
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.715567659270504e-05,
+      "loss": 1.0865,
+      "step": 12643
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.714674762266173e-05,
+      "loss": 0.9628,
+      "step": 12644
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.713781865261842e-05,
+      "loss": 1.104,
+      "step": 12645
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.712888968257513e-05,
+      "loss": 1.0917,
+      "step": 12646
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.711996071253182e-05,
+      "loss": 1.0001,
+      "step": 12647
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.711103174248851e-05,
+      "loss": 1.0738,
+      "step": 12648
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.710210277244521e-05,
+      "loss": 1.0298,
+      "step": 12649
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.70931738024019e-05,
+      "loss": 1.1637,
+      "step": 12650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.708424483235859e-05,
+      "loss": 1.0468,
+      "step": 12651
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.707531586231528e-05,
+      "loss": 1.0155,
+      "step": 12652
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.706638689227197e-05,
+      "loss": 1.0633,
+      "step": 12653
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.705745792222868e-05,
+      "loss": 1.1579,
+      "step": 12654
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.704852895218537e-05,
+      "loss": 1.0251,
+      "step": 12655
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.703959998214206e-05,
+      "loss": 0.9662,
+      "step": 12656
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.703067101209876e-05,
+      "loss": 1.0139,
+      "step": 12657
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.702174204205545e-05,
+      "loss": 1.1319,
+      "step": 12658
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.701281307201214e-05,
+      "loss": 1.1626,
+      "step": 12659
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.700388410196885e-05,
+      "loss": 1.019,
+      "step": 12660
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.699495513192554e-05,
+      "loss": 1.0378,
+      "step": 12661
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.698602616188223e-05,
+      "loss": 1.0345,
+      "step": 12662
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.697709719183893e-05,
+      "loss": 0.9585,
+      "step": 12663
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.696816822179562e-05,
+      "loss": 1.1747,
+      "step": 12664
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.695923925175231e-05,
+      "loss": 1.0196,
+      "step": 12665
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.695031028170902e-05,
+      "loss": 0.9696,
+      "step": 12666
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.694138131166571e-05,
+      "loss": 0.9519,
+      "step": 12667
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.69324523416224e-05,
+      "loss": 1.0983,
+      "step": 12668
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.69235233715791e-05,
+      "loss": 0.9669,
+      "step": 12669
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.691459440153579e-05,
+      "loss": 1.0654,
+      "step": 12670
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.690566543149248e-05,
+      "loss": 0.9356,
+      "step": 12671
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.689673646144917e-05,
+      "loss": 1.0185,
+      "step": 12672
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.688780749140586e-05,
+      "loss": 1.0576,
+      "step": 12673
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.687887852136257e-05,
+      "loss": 1.0246,
+      "step": 12674
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.686994955131926e-05,
+      "loss": 1.0407,
+      "step": 12675
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.686102058127595e-05,
+      "loss": 0.9782,
+      "step": 12676
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.685209161123264e-05,
+      "loss": 0.9698,
+      "step": 12677
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.684316264118934e-05,
+      "loss": 1.0889,
+      "step": 12678
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.683423367114603e-05,
+      "loss": 1.0529,
+      "step": 12679
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.682530470110272e-05,
+      "loss": 1.1567,
+      "step": 12680
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.681637573105943e-05,
+      "loss": 0.9806,
+      "step": 12681
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.680744676101612e-05,
+      "loss": 0.9169,
+      "step": 12682
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.679851779097281e-05,
+      "loss": 0.9979,
+      "step": 12683
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.678958882092951e-05,
+      "loss": 0.9844,
+      "step": 12684
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.67806598508862e-05,
+      "loss": 0.9809,
+      "step": 12685
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.67717308808429e-05,
+      "loss": 1.0604,
+      "step": 12686
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.67628019107996e-05,
+      "loss": 1.1298,
+      "step": 12687
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.675387294075629e-05,
+      "loss": 1.0509,
+      "step": 12688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.674494397071298e-05,
+      "loss": 1.0629,
+      "step": 12689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.673601500066968e-05,
+      "loss": 0.9919,
+      "step": 12690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.672708603062637e-05,
+      "loss": 1.0204,
+      "step": 12691
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.671815706058306e-05,
+      "loss": 0.9225,
+      "step": 12692
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.670922809053977e-05,
+      "loss": 1.0781,
+      "step": 12693
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.670029912049646e-05,
+      "loss": 0.9782,
+      "step": 12694
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.669137015045315e-05,
+      "loss": 1.1358,
+      "step": 12695
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.668244118040984e-05,
+      "loss": 1.0105,
+      "step": 12696
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.667351221036653e-05,
+      "loss": 1.0753,
+      "step": 12697
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.666458324032323e-05,
+      "loss": 1.1746,
+      "step": 12698
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.665565427027992e-05,
+      "loss": 1.0946,
+      "step": 12699
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.664672530023662e-05,
+      "loss": 0.9977,
+      "step": 12700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.663779633019332e-05,
+      "loss": 1.1048,
+      "step": 12701
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.662886736015001e-05,
+      "loss": 1.0519,
+      "step": 12702
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.66199383901067e-05,
+      "loss": 0.9939,
+      "step": 12703
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.66110094200634e-05,
+      "loss": 1.1426,
+      "step": 12704
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.66020804500201e-05,
+      "loss": 1.0478,
+      "step": 12705
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.659315147997679e-05,
+      "loss": 1.1061,
+      "step": 12706
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.658422250993349e-05,
+      "loss": 1.0336,
+      "step": 12707
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.657529353989018e-05,
+      "loss": 0.983,
+      "step": 12708
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.656636456984687e-05,
+      "loss": 1.1546,
+      "step": 12709
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.655743559980357e-05,
+      "loss": 1.3113,
+      "step": 12710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.654850662976026e-05,
+      "loss": 1.021,
+      "step": 12711
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.653957765971696e-05,
+      "loss": 0.9892,
+      "step": 12712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.653064868967366e-05,
+      "loss": 1.0549,
+      "step": 12713
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.652171971963035e-05,
+      "loss": 1.0871,
+      "step": 12714
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.651279074958704e-05,
+      "loss": 1.0161,
+      "step": 12715
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.650386177954374e-05,
+      "loss": 0.9385,
+      "step": 12716
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.649493280950043e-05,
+      "loss": 0.9051,
+      "step": 12717
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.648600383945713e-05,
+      "loss": 1.0257,
+      "step": 12718
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.647707486941382e-05,
+      "loss": 0.9687,
+      "step": 12719
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.64681458993705e-05,
+      "loss": 1.0535,
+      "step": 12720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.64592169293272e-05,
+      "loss": 1.0123,
+      "step": 12721
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.64502879592839e-05,
+      "loss": 1.0454,
+      "step": 12722
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.644135898924059e-05,
+      "loss": 1.0532,
+      "step": 12723
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.643243001919728e-05,
+      "loss": 0.9114,
+      "step": 12724
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.642350104915399e-05,
+      "loss": 1.0636,
+      "step": 12725
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.641457207911068e-05,
+      "loss": 1.1122,
+      "step": 12726
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.640564310906737e-05,
+      "loss": 0.9828,
+      "step": 12727
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.639671413902407e-05,
+      "loss": 1.0498,
+      "step": 12728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.638778516898076e-05,
+      "loss": 0.9136,
+      "step": 12729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.637885619893745e-05,
+      "loss": 0.9707,
+      "step": 12730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.636992722889416e-05,
+      "loss": 0.9832,
+      "step": 12731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.636099825885085e-05,
+      "loss": 1.0248,
+      "step": 12732
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.635206928880754e-05,
+      "loss": 1.0237,
+      "step": 12733
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.634314031876424e-05,
+      "loss": 1.1447,
+      "step": 12734
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.633421134872093e-05,
+      "loss": 0.9782,
+      "step": 12735
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.632528237867762e-05,
+      "loss": 0.994,
+      "step": 12736
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.631635340863433e-05,
+      "loss": 1.1359,
+      "step": 12737
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.630742443859102e-05,
+      "loss": 1.079,
+      "step": 12738
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.62984954685477e-05,
+      "loss": 0.9282,
+      "step": 12739
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.628956649850441e-05,
+      "loss": 1.076,
+      "step": 12740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.62806375284611e-05,
+      "loss": 1.1052,
+      "step": 12741
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.627170855841779e-05,
+      "loss": 1.0389,
+      "step": 12742
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.626277958837448e-05,
+      "loss": 1.0975,
+      "step": 12743
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.625385061833117e-05,
+      "loss": 0.9364,
+      "step": 12744
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.624492164828786e-05,
+      "loss": 1.0937,
+      "step": 12745
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.623599267824457e-05,
+      "loss": 1.0241,
+      "step": 12746
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.622706370820126e-05,
+      "loss": 0.943,
+      "step": 12747
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.621813473815796e-05,
+      "loss": 1.0194,
+      "step": 12748
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.620920576811465e-05,
+      "loss": 1.0792,
+      "step": 12749
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.620027679807134e-05,
+      "loss": 0.9329,
+      "step": 12750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.619134782802805e-05,
+      "loss": 1.0578,
+      "step": 12751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.618241885798474e-05,
+      "loss": 1.1085,
+      "step": 12752
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.617348988794143e-05,
+      "loss": 1.0004,
+      "step": 12753
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.616456091789813e-05,
+      "loss": 1.1028,
+      "step": 12754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.615563194785482e-05,
+      "loss": 1.0895,
+      "step": 12755
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.614670297781151e-05,
+      "loss": 1.0239,
+      "step": 12756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.613777400776822e-05,
+      "loss": 1.0237,
+      "step": 12757
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.612884503772491e-05,
+      "loss": 1.085,
+      "step": 12758
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.61199160676816e-05,
+      "loss": 1.0561,
+      "step": 12759
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.61109870976383e-05,
+      "loss": 1.0682,
+      "step": 12760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.610205812759499e-05,
+      "loss": 1.0151,
+      "step": 12761
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.609312915755168e-05,
+      "loss": 0.9573,
+      "step": 12762
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.608420018750837e-05,
+      "loss": 1.0533,
+      "step": 12763
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.607527121746506e-05,
+      "loss": 1.0681,
+      "step": 12764
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.606634224742177e-05,
+      "loss": 1.1841,
+      "step": 12765
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.605741327737846e-05,
+      "loss": 1.0523,
+      "step": 12766
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.604848430733515e-05,
+      "loss": 1.1526,
+      "step": 12767
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.603955533729184e-05,
+      "loss": 1.0854,
+      "step": 12768
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.603062636724854e-05,
+      "loss": 1.0476,
+      "step": 12769
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.602169739720523e-05,
+      "loss": 1.0138,
+      "step": 12770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.601276842716192e-05,
+      "loss": 1.2752,
+      "step": 12771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.600383945711863e-05,
+      "loss": 1.0168,
+      "step": 12772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.599491048707532e-05,
+      "loss": 1.003,
+      "step": 12773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.598598151703201e-05,
+      "loss": 0.9527,
+      "step": 12774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.597705254698871e-05,
+      "loss": 0.9876,
+      "step": 12775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.59681235769454e-05,
+      "loss": 0.9503,
+      "step": 12776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.59591946069021e-05,
+      "loss": 1.0044,
+      "step": 12777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.59502656368588e-05,
+      "loss": 0.9471,
+      "step": 12778
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.594133666681549e-05,
+      "loss": 1.0423,
+      "step": 12779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.593240769677218e-05,
+      "loss": 1.0663,
+      "step": 12780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.592347872672888e-05,
+      "loss": 0.9034,
+      "step": 12781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.591454975668557e-05,
+      "loss": 1.0464,
+      "step": 12782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.590562078664226e-05,
+      "loss": 1.0544,
+      "step": 12783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.589669181659897e-05,
+      "loss": 0.9905,
+      "step": 12784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.588776284655566e-05,
+      "loss": 1.0784,
+      "step": 12785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.587883387651235e-05,
+      "loss": 1.0319,
+      "step": 12786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.586990490646904e-05,
+      "loss": 1.2373,
+      "step": 12787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.586097593642573e-05,
+      "loss": 1.1396,
+      "step": 12788
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.585204696638242e-05,
+      "loss": 0.9809,
+      "step": 12789
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.584311799633912e-05,
+      "loss": 0.9515,
+      "step": 12790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.583418902629581e-05,
+      "loss": 1.0744,
+      "step": 12791
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.58252600562525e-05,
+      "loss": 1.0467,
+      "step": 12792
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.581633108620921e-05,
+      "loss": 1.0464,
+      "step": 12793
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.58074021161659e-05,
+      "loss": 1.057,
+      "step": 12794
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.57984731461226e-05,
+      "loss": 1.0277,
+      "step": 12795
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.57895441760793e-05,
+      "loss": 0.9455,
+      "step": 12796
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.578061520603598e-05,
+      "loss": 1.0241,
+      "step": 12797
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.577168623599269e-05,
+      "loss": 0.8783,
+      "step": 12798
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.576275726594938e-05,
+      "loss": 1.0339,
+      "step": 12799
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.575382829590607e-05,
+      "loss": 1.0158,
+      "step": 12800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.574489932586277e-05,
+      "loss": 1.0679,
+      "step": 12801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.573597035581946e-05,
+      "loss": 1.0524,
+      "step": 12802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.572704138577615e-05,
+      "loss": 1.0538,
+      "step": 12803
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.571811241573286e-05,
+      "loss": 1.012,
+      "step": 12804
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.570918344568955e-05,
+      "loss": 0.9514,
+      "step": 12805
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.570025447564624e-05,
+      "loss": 1.0487,
+      "step": 12806
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.569132550560294e-05,
+      "loss": 0.9521,
+      "step": 12807
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.568239653555963e-05,
+      "loss": 0.9374,
+      "step": 12808
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.567346756551632e-05,
+      "loss": 1.0912,
+      "step": 12809
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.566453859547302e-05,
+      "loss": 1.0986,
+      "step": 12810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.56556096254297e-05,
+      "loss": 1.1069,
+      "step": 12811
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.56466806553864e-05,
+      "loss": 0.9776,
+      "step": 12812
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.56377516853431e-05,
+      "loss": 1.0556,
+      "step": 12813
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.562882271529979e-05,
+      "loss": 0.9356,
+      "step": 12814
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.561989374525648e-05,
+      "loss": 1.017,
+      "step": 12815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.561096477521319e-05,
+      "loss": 1.0331,
+      "step": 12816
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.560203580516988e-05,
+      "loss": 1.0339,
+      "step": 12817
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.559310683512657e-05,
+      "loss": 1.0942,
+      "step": 12818
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.558417786508327e-05,
+      "loss": 1.0587,
+      "step": 12819
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.557524889503996e-05,
+      "loss": 1.038,
+      "step": 12820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.556631992499665e-05,
+      "loss": 1.079,
+      "step": 12821
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.555739095495336e-05,
+      "loss": 1.1069,
+      "step": 12822
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.554846198491005e-05,
+      "loss": 1.0101,
+      "step": 12823
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.553953301486674e-05,
+      "loss": 0.9316,
+      "step": 12824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.553060404482344e-05,
+      "loss": 0.9528,
+      "step": 12825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.552167507478013e-05,
+      "loss": 1.1796,
+      "step": 12826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.551274610473682e-05,
+      "loss": 0.9911,
+      "step": 12827
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.550381713469353e-05,
+      "loss": 1.1253,
+      "step": 12828
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.549488816465022e-05,
+      "loss": 0.9892,
+      "step": 12829
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.54859591946069e-05,
+      "loss": 1.239,
+      "step": 12830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.547703022456361e-05,
+      "loss": 1.1488,
+      "step": 12831
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.54681012545203e-05,
+      "loss": 0.9924,
+      "step": 12832
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.545917228447699e-05,
+      "loss": 1.002,
+      "step": 12833
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.545024331443368e-05,
+      "loss": 0.9896,
+      "step": 12834
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.544131434439037e-05,
+      "loss": 1.0996,
+      "step": 12835
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.543238537434706e-05,
+      "loss": 0.9166,
+      "step": 12836
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.542345640430377e-05,
+      "loss": 1.0183,
+      "step": 12837
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.541452743426046e-05,
+      "loss": 1.0136,
+      "step": 12838
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.540559846421715e-05,
+      "loss": 0.8757,
+      "step": 12839
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.539666949417385e-05,
+      "loss": 1.0697,
+      "step": 12840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.538774052413054e-05,
+      "loss": 1.159,
+      "step": 12841
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.537881155408723e-05,
+      "loss": 1.0596,
+      "step": 12842
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.536988258404394e-05,
+      "loss": 1.0588,
+      "step": 12843
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.536095361400063e-05,
+      "loss": 1.1032,
+      "step": 12844
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.535202464395733e-05,
+      "loss": 1.0001,
+      "step": 12845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.534309567391402e-05,
+      "loss": 0.9852,
+      "step": 12846
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.533416670387071e-05,
+      "loss": 0.9921,
+      "step": 12847
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.532523773382742e-05,
+      "loss": 1.1235,
+      "step": 12848
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.53163087637841e-05,
+      "loss": 1.0163,
+      "step": 12849
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.53073797937408e-05,
+      "loss": 1.0243,
+      "step": 12850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.52984508236975e-05,
+      "loss": 0.9107,
+      "step": 12851
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.528952185365419e-05,
+      "loss": 1.1306,
+      "step": 12852
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.528059288361088e-05,
+      "loss": 1.0299,
+      "step": 12853
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.527166391356757e-05,
+      "loss": 1.0054,
+      "step": 12854
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.526273494352426e-05,
+      "loss": 0.9783,
+      "step": 12855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.525380597348095e-05,
+      "loss": 1.0098,
+      "step": 12856
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.524487700343766e-05,
+      "loss": 0.9328,
+      "step": 12857
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.523594803339435e-05,
+      "loss": 0.9669,
+      "step": 12858
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.522701906335104e-05,
+      "loss": 1.0292,
+      "step": 12859
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.521809009330774e-05,
+      "loss": 0.9883,
+      "step": 12860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.520916112326443e-05,
+      "loss": 1.0619,
+      "step": 12861
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.520023215322112e-05,
+      "loss": 1.135,
+      "step": 12862
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.519130318317783e-05,
+      "loss": 0.9969,
+      "step": 12863
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.518237421313452e-05,
+      "loss": 0.9534,
+      "step": 12864
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.517344524309121e-05,
+      "loss": 1.0907,
+      "step": 12865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.516451627304791e-05,
+      "loss": 1.0671,
+      "step": 12866
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.51555873030046e-05,
+      "loss": 0.9757,
+      "step": 12867
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.51466583329613e-05,
+      "loss": 1.2101,
+      "step": 12868
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.5137729362918e-05,
+      "loss": 1.0097,
+      "step": 12869
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.512880039287469e-05,
+      "loss": 1.2874,
+      "step": 12870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.511987142283138e-05,
+      "loss": 0.9162,
+      "step": 12871
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.511094245278808e-05,
+      "loss": 1.0998,
+      "step": 12872
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.510201348274477e-05,
+      "loss": 0.9803,
+      "step": 12873
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.509308451270146e-05,
+      "loss": 0.9294,
+      "step": 12874
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.508415554265817e-05,
+      "loss": 1.0451,
+      "step": 12875
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.507522657261486e-05,
+      "loss": 1.0812,
+      "step": 12876
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.506629760257155e-05,
+      "loss": 0.9192,
+      "step": 12877
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.505736863252824e-05,
+      "loss": 1.0598,
+      "step": 12878
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.504843966248493e-05,
+      "loss": 1.122,
+      "step": 12879
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.503951069244162e-05,
+      "loss": 1.028,
+      "step": 12880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.503058172239832e-05,
+      "loss": 1.0616,
+      "step": 12881
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.502165275235501e-05,
+      "loss": 1.0549,
+      "step": 12882
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.50127237823117e-05,
+      "loss": 1.1118,
+      "step": 12883
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.500379481226841e-05,
+      "loss": 1.1251,
+      "step": 12884
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.49948658422251e-05,
+      "loss": 1.0237,
+      "step": 12885
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.498593687218179e-05,
+      "loss": 1.0494,
+      "step": 12886
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.49770079021385e-05,
+      "loss": 1.0613,
+      "step": 12887
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.496807893209518e-05,
+      "loss": 1.0527,
+      "step": 12888
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.495914996205187e-05,
+      "loss": 1.0091,
+      "step": 12889
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.495022099200858e-05,
+      "loss": 0.883,
+      "step": 12890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.494129202196527e-05,
+      "loss": 1.015,
+      "step": 12891
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.493236305192197e-05,
+      "loss": 0.8898,
+      "step": 12892
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.492343408187866e-05,
+      "loss": 1.0037,
+      "step": 12893
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.491450511183535e-05,
+      "loss": 0.9882,
+      "step": 12894
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.490557614179206e-05,
+      "loss": 1.0045,
+      "step": 12895
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.489664717174875e-05,
+      "loss": 1.1022,
+      "step": 12896
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.488771820170544e-05,
+      "loss": 1.0239,
+      "step": 12897
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.487878923166214e-05,
+      "loss": 0.9914,
+      "step": 12898
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.486986026161883e-05,
+      "loss": 1.0214,
+      "step": 12899
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.486093129157552e-05,
+      "loss": 1.036,
+      "step": 12900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.485200232153221e-05,
+      "loss": 1.0196,
+      "step": 12901
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48430733514889e-05,
+      "loss": 1.0379,
+      "step": 12902
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48341443814456e-05,
+      "loss": 1.2251,
+      "step": 12903
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48252154114023e-05,
+      "loss": 1.0353,
+      "step": 12904
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.481628644135899e-05,
+      "loss": 1.0549,
+      "step": 12905
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.480735747131568e-05,
+      "loss": 1.0605,
+      "step": 12906
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.479842850127238e-05,
+      "loss": 1.1478,
+      "step": 12907
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.478949953122908e-05,
+      "loss": 1.0256,
+      "step": 12908
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.478057056118577e-05,
+      "loss": 1.0099,
+      "step": 12909
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.477164159114247e-05,
+      "loss": 1.0522,
+      "step": 12910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.476271262109916e-05,
+      "loss": 1.1063,
+      "step": 12911
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.475378365105585e-05,
+      "loss": 1.2611,
+      "step": 12912
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.474485468101255e-05,
+      "loss": 1.0124,
+      "step": 12913
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.473592571096925e-05,
+      "loss": 0.9138,
+      "step": 12914
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.472699674092594e-05,
+      "loss": 0.925,
+      "step": 12915
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.471806777088264e-05,
+      "loss": 1.0362,
+      "step": 12916
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.470913880083933e-05,
+      "loss": 1.3386,
+      "step": 12917
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.470020983079602e-05,
+      "loss": 0.9849,
+      "step": 12918
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.469128086075272e-05,
+      "loss": 1.0899,
+      "step": 12919
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.468235189070942e-05,
+      "loss": 1.0555,
+      "step": 12920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.46734229206661e-05,
+      "loss": 1.0741,
+      "step": 12921
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.46644939506228e-05,
+      "loss": 0.9269,
+      "step": 12922
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.46555649805795e-05,
+      "loss": 0.9593,
+      "step": 12923
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.464663601053619e-05,
+      "loss": 1.1483,
+      "step": 12924
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.463770704049288e-05,
+      "loss": 1.1221,
+      "step": 12925
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.462877807044957e-05,
+      "loss": 0.9114,
+      "step": 12926
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.461984910040626e-05,
+      "loss": 1.033,
+      "step": 12927
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.461092013036297e-05,
+      "loss": 1.0934,
+      "step": 12928
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.460199116031966e-05,
+      "loss": 1.1531,
+      "step": 12929
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.459306219027635e-05,
+      "loss": 0.9988,
+      "step": 12930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.458413322023305e-05,
+      "loss": 0.9361,
+      "step": 12931
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.457520425018974e-05,
+      "loss": 1.0527,
+      "step": 12932
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.456627528014643e-05,
+      "loss": 1.1675,
+      "step": 12933
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.455734631010314e-05,
+      "loss": 1.0371,
+      "step": 12934
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.454841734005983e-05,
+      "loss": 1.1157,
+      "step": 12935
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.453948837001652e-05,
+      "loss": 1.0812,
+      "step": 12936
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.453055939997322e-05,
+      "loss": 1.1162,
+      "step": 12937
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.452163042992991e-05,
+      "loss": 1.0087,
+      "step": 12938
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.45127014598866e-05,
+      "loss": 0.9999,
+      "step": 12939
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.45037724898433e-05,
+      "loss": 1.0029,
+      "step": 12940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.44948435198e-05,
+      "loss": 1.0938,
+      "step": 12941
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.44859145497567e-05,
+      "loss": 1.0606,
+      "step": 12942
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.447698557971339e-05,
+      "loss": 1.0549,
+      "step": 12943
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.446805660967008e-05,
+      "loss": 1.3336,
+      "step": 12944
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.445912763962677e-05,
+      "loss": 1.0124,
+      "step": 12945
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.445019866958346e-05,
+      "loss": 1.0583,
+      "step": 12946
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.444126969954015e-05,
+      "loss": 1.0234,
+      "step": 12947
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.443234072949686e-05,
+      "loss": 1.0612,
+      "step": 12948
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.442341175945355e-05,
+      "loss": 1.0393,
+      "step": 12949
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.441448278941024e-05,
+      "loss": 1.0642,
+      "step": 12950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.440555381936694e-05,
+      "loss": 0.9613,
+      "step": 12951
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.439662484932363e-05,
+      "loss": 1.1403,
+      "step": 12952
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.438769587928032e-05,
+      "loss": 1.027,
+      "step": 12953
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.437876690923703e-05,
+      "loss": 1.0119,
+      "step": 12954
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.436983793919372e-05,
+      "loss": 1.0655,
+      "step": 12955
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.436090896915041e-05,
+      "loss": 1.1124,
+      "step": 12956
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.435197999910711e-05,
+      "loss": 1.0392,
+      "step": 12957
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.43430510290638e-05,
+      "loss": 0.9996,
+      "step": 12958
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.433412205902049e-05,
+      "loss": 0.9846,
+      "step": 12959
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.43251930889772e-05,
+      "loss": 1.2278,
+      "step": 12960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.431626411893389e-05,
+      "loss": 0.8173,
+      "step": 12961
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.430733514889058e-05,
+      "loss": 0.9692,
+      "step": 12962
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.429840617884728e-05,
+      "loss": 0.9705,
+      "step": 12963
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.428947720880397e-05,
+      "loss": 0.9751,
+      "step": 12964
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.428054823876066e-05,
+      "loss": 1.0625,
+      "step": 12965
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.427161926871737e-05,
+      "loss": 1.1443,
+      "step": 12966
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.426269029867406e-05,
+      "loss": 0.965,
+      "step": 12967
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.425376132863075e-05,
+      "loss": 1.0883,
+      "step": 12968
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.424483235858744e-05,
+      "loss": 1.0908,
+      "step": 12969
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.423590338854413e-05,
+      "loss": 1.0982,
+      "step": 12970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.422697441850082e-05,
+      "loss": 0.9526,
+      "step": 12971
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.421804544845752e-05,
+      "loss": 0.9941,
+      "step": 12972
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.420911647841421e-05,
+      "loss": 0.9289,
+      "step": 12973
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.42001875083709e-05,
+      "loss": 0.9972,
+      "step": 12974
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.419125853832761e-05,
+      "loss": 0.9821,
+      "step": 12975
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.41823295682843e-05,
+      "loss": 0.9899,
+      "step": 12976
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.417340059824099e-05,
+      "loss": 1.0401,
+      "step": 12977
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.41644716281977e-05,
+      "loss": 1.0593,
+      "step": 12978
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.415554265815438e-05,
+      "loss": 1.0006,
+      "step": 12979
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.414661368811107e-05,
+      "loss": 1.2238,
+      "step": 12980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.413768471806778e-05,
+      "loss": 0.9771,
+      "step": 12981
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.412875574802447e-05,
+      "loss": 0.9343,
+      "step": 12982
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.411982677798116e-05,
+      "loss": 1.0002,
+      "step": 12983
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.411089780793786e-05,
+      "loss": 0.9326,
+      "step": 12984
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.410196883789455e-05,
+      "loss": 1.1788,
+      "step": 12985
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.409303986785124e-05,
+      "loss": 1.0272,
+      "step": 12986
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.408411089780795e-05,
+      "loss": 0.9239,
+      "step": 12987
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.407518192776464e-05,
+      "loss": 1.1574,
+      "step": 12988
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.406625295772133e-05,
+      "loss": 0.9266,
+      "step": 12989
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.405732398767803e-05,
+      "loss": 0.9526,
+      "step": 12990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.404839501763472e-05,
+      "loss": 1.0873,
+      "step": 12991
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.403946604759141e-05,
+      "loss": 1.0741,
+      "step": 12992
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.40305370775481e-05,
+      "loss": 0.9888,
+      "step": 12993
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.40216081075048e-05,
+      "loss": 1.0182,
+      "step": 12994
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.40126791374615e-05,
+      "loss": 1.0819,
+      "step": 12995
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.400375016741819e-05,
+      "loss": 0.966,
+      "step": 12996
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.399482119737488e-05,
+      "loss": 1.0285,
+      "step": 12997
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.398589222733158e-05,
+      "loss": 0.9991,
+      "step": 12998
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.397696325728827e-05,
+      "loss": 0.9623,
+      "step": 12999
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.396803428724497e-05,
+      "loss": 1.0765,
+      "step": 13000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.395910531720167e-05,
+      "loss": 1.0224,
+      "step": 13001
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.395017634715836e-05,
+      "loss": 1.1042,
+      "step": 13002
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.394124737711505e-05,
+      "loss": 0.9436,
+      "step": 13003
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.393231840707175e-05,
+      "loss": 0.9791,
+      "step": 13004
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.392338943702844e-05,
+      "loss": 0.8924,
+      "step": 13005
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.391446046698514e-05,
+      "loss": 1.1114,
+      "step": 13006
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.390553149694184e-05,
+      "loss": 0.9641,
+      "step": 13007
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.389660252689853e-05,
+      "loss": 1.1003,
+      "step": 13008
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.388767355685522e-05,
+      "loss": 1.0771,
+      "step": 13009
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.387874458681192e-05,
+      "loss": 1.0196,
+      "step": 13010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.386981561676861e-05,
+      "loss": 1.0364,
+      "step": 13011
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.38608866467253e-05,
+      "loss": 1.121,
+      "step": 13012
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.3851957676682e-05,
+      "loss": 0.9969,
+      "step": 13013
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.38430287066387e-05,
+      "loss": 1.0823,
+      "step": 13014
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.383409973659539e-05,
+      "loss": 0.9533,
+      "step": 13015
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.382517076655208e-05,
+      "loss": 1.0081,
+      "step": 13016
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.381624179650877e-05,
+      "loss": 0.9135,
+      "step": 13017
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.380731282646546e-05,
+      "loss": 1.0459,
+      "step": 13018
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.379838385642217e-05,
+      "loss": 1.0964,
+      "step": 13019
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.378945488637886e-05,
+      "loss": 0.8226,
+      "step": 13020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.378052591633555e-05,
+      "loss": 1.2316,
+      "step": 13021
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.377159694629225e-05,
+      "loss": 1.1273,
+      "step": 13022
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.376266797624894e-05,
+      "loss": 0.9868,
+      "step": 13023
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.375373900620563e-05,
+      "loss": 0.9976,
+      "step": 13024
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.374481003616234e-05,
+      "loss": 1.1253,
+      "step": 13025
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.373588106611903e-05,
+      "loss": 1.0681,
+      "step": 13026
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.372695209607572e-05,
+      "loss": 1.0464,
+      "step": 13027
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.371802312603242e-05,
+      "loss": 1.0983,
+      "step": 13028
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.370909415598911e-05,
+      "loss": 1.0427,
+      "step": 13029
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.37001651859458e-05,
+      "loss": 0.9312,
+      "step": 13030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.36912362159025e-05,
+      "loss": 1.0561,
+      "step": 13031
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.36823072458592e-05,
+      "loss": 0.9322,
+      "step": 13032
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.367337827581589e-05,
+      "loss": 1.1199,
+      "step": 13033
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.366444930577259e-05,
+      "loss": 1.0883,
+      "step": 13034
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.365552033572928e-05,
+      "loss": 1.0511,
+      "step": 13035
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.364659136568597e-05,
+      "loss": 1.0129,
+      "step": 13036
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.363766239564266e-05,
+      "loss": 1.0868,
+      "step": 13037
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.362873342559935e-05,
+      "loss": 1.1101,
+      "step": 13038
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.361980445555606e-05,
+      "loss": 1.0005,
+      "step": 13039
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.361087548551275e-05,
+      "loss": 0.9868,
+      "step": 13040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.360194651546944e-05,
+      "loss": 1.0466,
+      "step": 13041
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.359301754542614e-05,
+      "loss": 1.0778,
+      "step": 13042
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.358408857538283e-05,
+      "loss": 0.9875,
+      "step": 13043
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.357515960533952e-05,
+      "loss": 1.0826,
+      "step": 13044
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.356623063529623e-05,
+      "loss": 1.095,
+      "step": 13045
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.355730166525292e-05,
+      "loss": 1.0045,
+      "step": 13046
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.354837269520961e-05,
+      "loss": 0.9229,
+      "step": 13047
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.353944372516631e-05,
+      "loss": 0.9793,
+      "step": 13048
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.3530514755123e-05,
+      "loss": 1.0928,
+      "step": 13049
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.352158578507969e-05,
+      "loss": 0.8928,
+      "step": 13050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.35126568150364e-05,
+      "loss": 0.9466,
+      "step": 13051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.350372784499309e-05,
+      "loss": 1.0993,
+      "step": 13052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.349479887494978e-05,
+      "loss": 1.0041,
+      "step": 13053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.348586990490648e-05,
+      "loss": 1.06,
+      "step": 13054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.347694093486317e-05,
+      "loss": 1.412,
+      "step": 13055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.346801196481986e-05,
+      "loss": 0.9741,
+      "step": 13056
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.345908299477657e-05,
+      "loss": 0.9503,
+      "step": 13057
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.345015402473326e-05,
+      "loss": 0.9697,
+      "step": 13058
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.344122505468995e-05,
+      "loss": 1.0425,
+      "step": 13059
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.343229608464664e-05,
+      "loss": 0.9732,
+      "step": 13060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.342336711460333e-05,
+      "loss": 1.1034,
+      "step": 13061
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.341443814456002e-05,
+      "loss": 1.0737,
+      "step": 13062
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.340550917451672e-05,
+      "loss": 1.0853,
+      "step": 13063
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.339658020447341e-05,
+      "loss": 0.9916,
+      "step": 13064
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.33876512344301e-05,
+      "loss": 1.1493,
+      "step": 13065
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.337872226438681e-05,
+      "loss": 1.0947,
+      "step": 13066
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.33697932943435e-05,
+      "loss": 0.9791,
+      "step": 13067
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.336086432430019e-05,
+      "loss": 1.158,
+      "step": 13068
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.335193535425689e-05,
+      "loss": 1.0413,
+      "step": 13069
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.334300638421358e-05,
+      "loss": 1.018,
+      "step": 13070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.333407741417027e-05,
+      "loss": 1.0702,
+      "step": 13071
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.332514844412698e-05,
+      "loss": 1.0961,
+      "step": 13072
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.331621947408367e-05,
+      "loss": 1.0437,
+      "step": 13073
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.330729050404036e-05,
+      "loss": 0.9449,
+      "step": 13074
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.329836153399706e-05,
+      "loss": 1.2341,
+      "step": 13075
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.328943256395375e-05,
+      "loss": 0.9172,
+      "step": 13076
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.328050359391044e-05,
+      "loss": 0.8954,
+      "step": 13077
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.327157462386715e-05,
+      "loss": 0.995,
+      "step": 13078
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.326264565382384e-05,
+      "loss": 0.9746,
+      "step": 13079
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.325371668378053e-05,
+      "loss": 1.0667,
+      "step": 13080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.324478771373723e-05,
+      "loss": 0.9308,
+      "step": 13081
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.323585874369392e-05,
+      "loss": 0.9719,
+      "step": 13082
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.322692977365061e-05,
+      "loss": 1.0361,
+      "step": 13083
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.32180008036073e-05,
+      "loss": 1.0452,
+      "step": 13084
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.3209071833564e-05,
+      "loss": 1.0286,
+      "step": 13085
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.320014286352069e-05,
+      "loss": 0.9034,
+      "step": 13086
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.319121389347739e-05,
+      "loss": 1.0132,
+      "step": 13087
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.318228492343408e-05,
+      "loss": 1.0766,
+      "step": 13088
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.317335595339078e-05,
+      "loss": 1.0115,
+      "step": 13089
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.316442698334747e-05,
+      "loss": 0.8776,
+      "step": 13090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.315549801330417e-05,
+      "loss": 0.997,
+      "step": 13091
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.314656904326087e-05,
+      "loss": 1.0058,
+      "step": 13092
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.313764007321756e-05,
+      "loss": 1.0348,
+      "step": 13093
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.312871110317425e-05,
+      "loss": 1.1273,
+      "step": 13094
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.311978213313095e-05,
+      "loss": 1.0093,
+      "step": 13095
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.311085316308764e-05,
+      "loss": 0.9632,
+      "step": 13096
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.310192419304434e-05,
+      "loss": 1.028,
+      "step": 13097
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.309299522300104e-05,
+      "loss": 0.9655,
+      "step": 13098
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.308406625295773e-05,
+      "loss": 1.0566,
+      "step": 13099
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.307513728291442e-05,
+      "loss": 0.9999,
+      "step": 13100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.306620831287112e-05,
+      "loss": 1.0874,
+      "step": 13101
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.305727934282781e-05,
+      "loss": 1.041,
+      "step": 13102
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.30483503727845e-05,
+      "loss": 1.0795,
+      "step": 13103
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.30394214027412e-05,
+      "loss": 0.9184,
+      "step": 13104
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.30304924326979e-05,
+      "loss": 1.2009,
+      "step": 13105
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.302156346265459e-05,
+      "loss": 1.0294,
+      "step": 13106
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.301263449261128e-05,
+      "loss": 0.9804,
+      "step": 13107
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.300370552256797e-05,
+      "loss": 1.0039,
+      "step": 13108
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.299477655252466e-05,
+      "loss": 0.8799,
+      "step": 13109
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.298584758248137e-05,
+      "loss": 1.5216,
+      "step": 13110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.297691861243806e-05,
+      "loss": 1.1017,
+      "step": 13111
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.296798964239475e-05,
+      "loss": 1.0748,
+      "step": 13112
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.295906067235145e-05,
+      "loss": 1.1277,
+      "step": 13113
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.295013170230814e-05,
+      "loss": 0.9262,
+      "step": 13114
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.294120273226483e-05,
+      "loss": 1.097,
+      "step": 13115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.293227376222154e-05,
+      "loss": 1.0495,
+      "step": 13116
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.292334479217823e-05,
+      "loss": 0.9919,
+      "step": 13117
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.291441582213492e-05,
+      "loss": 0.9828,
+      "step": 13118
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.290548685209162e-05,
+      "loss": 1.0894,
+      "step": 13119
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.289655788204831e-05,
+      "loss": 1.1254,
+      "step": 13120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.2887628912005e-05,
+      "loss": 0.9441,
+      "step": 13121
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.28786999419617e-05,
+      "loss": 1.0218,
+      "step": 13122
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.28697709719184e-05,
+      "loss": 0.8995,
+      "step": 13123
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.286084200187509e-05,
+      "loss": 1.0377,
+      "step": 13124
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.285191303183179e-05,
+      "loss": 1.0865,
+      "step": 13125
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.284298406178848e-05,
+      "loss": 1.0198,
+      "step": 13126
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.283405509174517e-05,
+      "loss": 0.9616,
+      "step": 13127
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.282512612170186e-05,
+      "loss": 1.0246,
+      "step": 13128
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.281619715165855e-05,
+      "loss": 0.9306,
+      "step": 13129
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.280726818161524e-05,
+      "loss": 1.0929,
+      "step": 13130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.279833921157195e-05,
+      "loss": 1.1383,
+      "step": 13131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.278941024152864e-05,
+      "loss": 1.088,
+      "step": 13132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.278048127148533e-05,
+      "loss": 0.9447,
+      "step": 13133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.277155230144203e-05,
+      "loss": 1.0524,
+      "step": 13134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.276262333139872e-05,
+      "loss": 0.989,
+      "step": 13135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.275369436135541e-05,
+      "loss": 1.0485,
+      "step": 13136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.274476539131212e-05,
+      "loss": 1.012,
+      "step": 13137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.273583642126881e-05,
+      "loss": 0.8866,
+      "step": 13138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.272690745122551e-05,
+      "loss": 1.0085,
+      "step": 13139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.27179784811822e-05,
+      "loss": 1.014,
+      "step": 13140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.270904951113889e-05,
+      "loss": 1.0196,
+      "step": 13141
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.27001205410956e-05,
+      "loss": 0.8453,
+      "step": 13142
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.269119157105229e-05,
+      "loss": 0.9912,
+      "step": 13143
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.268226260100898e-05,
+      "loss": 1.1438,
+      "step": 13144
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.267333363096568e-05,
+      "loss": 1.0105,
+      "step": 13145
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.266440466092237e-05,
+      "loss": 0.9491,
+      "step": 13146
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.265547569087906e-05,
+      "loss": 0.9341,
+      "step": 13147
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.264654672083577e-05,
+      "loss": 1.0544,
+      "step": 13148
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.263761775079246e-05,
+      "loss": 0.9543,
+      "step": 13149
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.262868878074915e-05,
+      "loss": 1.0684,
+      "step": 13150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.261975981070584e-05,
+      "loss": 0.9049,
+      "step": 13151
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.261083084066253e-05,
+      "loss": 1.1953,
+      "step": 13152
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.260190187061922e-05,
+      "loss": 0.956,
+      "step": 13153
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.259297290057592e-05,
+      "loss": 1.1149,
+      "step": 13154
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.258404393053261e-05,
+      "loss": 1.1085,
+      "step": 13155
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.25751149604893e-05,
+      "loss": 1.1027,
+      "step": 13156
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.256618599044601e-05,
+      "loss": 1.1199,
+      "step": 13157
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.25572570204027e-05,
+      "loss": 0.9775,
+      "step": 13158
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.254832805035939e-05,
+      "loss": 0.9656,
+      "step": 13159
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.253939908031609e-05,
+      "loss": 1.1269,
+      "step": 13160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.253047011027278e-05,
+      "loss": 0.9244,
+      "step": 13161
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.252154114022947e-05,
+      "loss": 1.1378,
+      "step": 13162
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.251261217018618e-05,
+      "loss": 1.1969,
+      "step": 13163
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.250368320014287e-05,
+      "loss": 0.9843,
+      "step": 13164
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.249475423009956e-05,
+      "loss": 1.0882,
+      "step": 13165
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.248582526005626e-05,
+      "loss": 1.0802,
+      "step": 13166
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.247689629001295e-05,
+      "loss": 0.9842,
+      "step": 13167
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.246796731996964e-05,
+      "loss": 1.0326,
+      "step": 13168
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.245903834992635e-05,
+      "loss": 1.05,
+      "step": 13169
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.245010937988304e-05,
+      "loss": 1.0327,
+      "step": 13170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.244118040983973e-05,
+      "loss": 0.9874,
+      "step": 13171
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.243225143979643e-05,
+      "loss": 1.0732,
+      "step": 13172
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.242332246975312e-05,
+      "loss": 0.9122,
+      "step": 13173
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.241439349970981e-05,
+      "loss": 0.9852,
+      "step": 13174
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.24054645296665e-05,
+      "loss": 0.9549,
+      "step": 13175
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.23965355596232e-05,
+      "loss": 1.0612,
+      "step": 13176
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.238760658957989e-05,
+      "loss": 0.9387,
+      "step": 13177
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.237867761953659e-05,
+      "loss": 1.01,
+      "step": 13178
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.236974864949328e-05,
+      "loss": 1.3054,
+      "step": 13179
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.236081967944997e-05,
+      "loss": 1.0764,
+      "step": 13180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.235189070940667e-05,
+      "loss": 1.0047,
+      "step": 13181
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.234296173936336e-05,
+      "loss": 1.1894,
+      "step": 13182
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.233403276932006e-05,
+      "loss": 0.9559,
+      "step": 13183
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.232510379927676e-05,
+      "loss": 0.9936,
+      "step": 13184
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.231617482923345e-05,
+      "loss": 1.0852,
+      "step": 13185
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.230724585919015e-05,
+      "loss": 1.0408,
+      "step": 13186
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.229831688914684e-05,
+      "loss": 1.1054,
+      "step": 13187
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.228938791910353e-05,
+      "loss": 1.0203,
+      "step": 13188
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.228045894906024e-05,
+      "loss": 0.9686,
+      "step": 13189
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.227152997901693e-05,
+      "loss": 0.9626,
+      "step": 13190
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.226260100897362e-05,
+      "loss": 0.9336,
+      "step": 13191
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.225367203893032e-05,
+      "loss": 0.9655,
+      "step": 13192
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.224474306888701e-05,
+      "loss": 0.9519,
+      "step": 13193
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.22358140988437e-05,
+      "loss": 0.9859,
+      "step": 13194
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.22268851288004e-05,
+      "loss": 0.9415,
+      "step": 13195
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.221795615875709e-05,
+      "loss": 1.0351,
+      "step": 13196
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.220902718871379e-05,
+      "loss": 0.9933,
+      "step": 13197
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.220009821867048e-05,
+      "loss": 1.1555,
+      "step": 13198
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.219116924862717e-05,
+      "loss": 1.189,
+      "step": 13199
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.218224027858386e-05,
+      "loss": 0.9278,
+      "step": 13200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.217331130854057e-05,
+      "loss": 1.0534,
+      "step": 13201
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.216438233849726e-05,
+      "loss": 1.0729,
+      "step": 13202
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.215545336845395e-05,
+      "loss": 0.9456,
+      "step": 13203
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.214652439841065e-05,
+      "loss": 1.0449,
+      "step": 13204
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.213759542836734e-05,
+      "loss": 0.9258,
+      "step": 13205
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.212866645832403e-05,
+      "loss": 1.013,
+      "step": 13206
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.211973748828074e-05,
+      "loss": 0.9665,
+      "step": 13207
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.211080851823743e-05,
+      "loss": 0.9585,
+      "step": 13208
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.210187954819412e-05,
+      "loss": 1.022,
+      "step": 13209
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.209295057815082e-05,
+      "loss": 1.0456,
+      "step": 13210
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.208402160810751e-05,
+      "loss": 0.9925,
+      "step": 13211
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.20750926380642e-05,
+      "loss": 1.0289,
+      "step": 13212
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.20661636680209e-05,
+      "loss": 0.8821,
+      "step": 13213
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.20572346979776e-05,
+      "loss": 1.0395,
+      "step": 13214
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.204830572793429e-05,
+      "loss": 1.0003,
+      "step": 13215
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.203937675789099e-05,
+      "loss": 1.0206,
+      "step": 13216
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.203044778784768e-05,
+      "loss": 1.065,
+      "step": 13217
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.202151881780437e-05,
+      "loss": 0.9692,
+      "step": 13218
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.201258984776106e-05,
+      "loss": 1.0277,
+      "step": 13219
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.200366087771775e-05,
+      "loss": 1.0064,
+      "step": 13220
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.199473190767444e-05,
+      "loss": 0.8816,
+      "step": 13221
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.198580293763115e-05,
+      "loss": 1.0464,
+      "step": 13222
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.197687396758784e-05,
+      "loss": 1.0354,
+      "step": 13223
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.196794499754453e-05,
+      "loss": 0.9486,
+      "step": 13224
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.195901602750123e-05,
+      "loss": 1.0061,
+      "step": 13225
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.195008705745792e-05,
+      "loss": 1.0414,
+      "step": 13226
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.194115808741461e-05,
+      "loss": 0.9483,
+      "step": 13227
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.193222911737132e-05,
+      "loss": 1.0261,
+      "step": 13228
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.192330014732801e-05,
+      "loss": 1.0839,
+      "step": 13229
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.19143711772847e-05,
+      "loss": 1.0371,
+      "step": 13230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.19054422072414e-05,
+      "loss": 1.1144,
+      "step": 13231
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.189651323719809e-05,
+      "loss": 1.0435,
+      "step": 13232
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.188758426715478e-05,
+      "loss": 1.119,
+      "step": 13233
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.187865529711149e-05,
+      "loss": 0.9674,
+      "step": 13234
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.186972632706818e-05,
+      "loss": 1.0739,
+      "step": 13235
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.186079735702488e-05,
+      "loss": 1.1247,
+      "step": 13236
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.185186838698157e-05,
+      "loss": 1.1579,
+      "step": 13237
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.184293941693826e-05,
+      "loss": 1.0158,
+      "step": 13238
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.183401044689497e-05,
+      "loss": 0.9454,
+      "step": 13239
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.182508147685166e-05,
+      "loss": 1.0531,
+      "step": 13240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.181615250680835e-05,
+      "loss": 1.0044,
+      "step": 13241
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.180722353676504e-05,
+      "loss": 0.9624,
+      "step": 13242
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.179829456672173e-05,
+      "loss": 1.1224,
+      "step": 13243
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.178936559667842e-05,
+      "loss": 1.0718,
+      "step": 13244
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.178043662663512e-05,
+      "loss": 0.903,
+      "step": 13245
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.177150765659181e-05,
+      "loss": 1.0211,
+      "step": 13246
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.17625786865485e-05,
+      "loss": 1.0785,
+      "step": 13247
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.175364971650521e-05,
+      "loss": 1.1234,
+      "step": 13248
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.17447207464619e-05,
+      "loss": 0.8842,
+      "step": 13249
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.173579177641859e-05,
+      "loss": 1.0836,
+      "step": 13250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.172686280637529e-05,
+      "loss": 1.0622,
+      "step": 13251
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.171793383633198e-05,
+      "loss": 1.0321,
+      "step": 13252
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.170900486628867e-05,
+      "loss": 1.1687,
+      "step": 13253
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.170007589624538e-05,
+      "loss": 1.1128,
+      "step": 13254
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.169114692620207e-05,
+      "loss": 0.9665,
+      "step": 13255
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.168221795615876e-05,
+      "loss": 0.8702,
+      "step": 13256
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.167328898611546e-05,
+      "loss": 0.9402,
+      "step": 13257
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.166436001607215e-05,
+      "loss": 1.0552,
+      "step": 13258
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.165543104602884e-05,
+      "loss": 0.9905,
+      "step": 13259
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.164650207598555e-05,
+      "loss": 1.0112,
+      "step": 13260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.163757310594224e-05,
+      "loss": 0.9108,
+      "step": 13261
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.162864413589893e-05,
+      "loss": 1.0198,
+      "step": 13262
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.161971516585563e-05,
+      "loss": 1.0016,
+      "step": 13263
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.161078619581232e-05,
+      "loss": 1.0263,
+      "step": 13264
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.160185722576901e-05,
+      "loss": 1.0465,
+      "step": 13265
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.15929282557257e-05,
+      "loss": 1.0149,
+      "step": 13266
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.15839992856824e-05,
+      "loss": 0.9725,
+      "step": 13267
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.157507031563908e-05,
+      "loss": 0.9393,
+      "step": 13268
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.156614134559579e-05,
+      "loss": 1.1028,
+      "step": 13269
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.155721237555248e-05,
+      "loss": 0.9991,
+      "step": 13270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.154828340550917e-05,
+      "loss": 0.9956,
+      "step": 13271
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.153935443546587e-05,
+      "loss": 1.0821,
+      "step": 13272
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.153042546542256e-05,
+      "loss": 1.0636,
+      "step": 13273
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.152149649537925e-05,
+      "loss": 1.068,
+      "step": 13274
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.151256752533596e-05,
+      "loss": 0.9851,
+      "step": 13275
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.150363855529265e-05,
+      "loss": 1.0891,
+      "step": 13276
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.149470958524934e-05,
+      "loss": 1.0403,
+      "step": 13277
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.148578061520604e-05,
+      "loss": 0.9876,
+      "step": 13278
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.147685164516273e-05,
+      "loss": 1.1007,
+      "step": 13279
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.146792267511942e-05,
+      "loss": 1.0195,
+      "step": 13280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.145899370507613e-05,
+      "loss": 1.0814,
+      "step": 13281
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.145006473503282e-05,
+      "loss": 1.0043,
+      "step": 13282
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.144113576498951e-05,
+      "loss": 1.0144,
+      "step": 13283
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.143220679494621e-05,
+      "loss": 0.9399,
+      "step": 13284
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.14232778249029e-05,
+      "loss": 1.0477,
+      "step": 13285
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.14143488548596e-05,
+      "loss": 1.0557,
+      "step": 13286
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.140541988481629e-05,
+      "loss": 0.9031,
+      "step": 13287
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.139649091477299e-05,
+      "loss": 1.0974,
+      "step": 13288
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.138756194472968e-05,
+      "loss": 1.0642,
+      "step": 13289
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.137863297468637e-05,
+      "loss": 1.0469,
+      "step": 13290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.136970400464306e-05,
+      "loss": 1.0403,
+      "step": 13291
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.136077503459976e-05,
+      "loss": 1.0782,
+      "step": 13292
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.135184606455646e-05,
+      "loss": 0.9823,
+      "step": 13293
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.134291709451315e-05,
+      "loss": 1.1034,
+      "step": 13294
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.133398812446985e-05,
+      "loss": 0.9528,
+      "step": 13295
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.132505915442654e-05,
+      "loss": 1.1291,
+      "step": 13296
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.131613018438323e-05,
+      "loss": 1.0598,
+      "step": 13297
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.130720121433993e-05,
+      "loss": 1.0007,
+      "step": 13298
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.129827224429663e-05,
+      "loss": 0.9862,
+      "step": 13299
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.128934327425332e-05,
+      "loss": 1.1296,
+      "step": 13300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.128041430421002e-05,
+      "loss": 1.1187,
+      "step": 13301
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.127148533416671e-05,
+      "loss": 1.0518,
+      "step": 13302
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.12625563641234e-05,
+      "loss": 1.0658,
+      "step": 13303
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.12536273940801e-05,
+      "loss": 0.9176,
+      "step": 13304
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.12446984240368e-05,
+      "loss": 1.0699,
+      "step": 13305
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.123576945399349e-05,
+      "loss": 1.0992,
+      "step": 13306
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.122684048395019e-05,
+      "loss": 0.9362,
+      "step": 13307
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.121791151390688e-05,
+      "loss": 1.0042,
+      "step": 13308
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.120898254386357e-05,
+      "loss": 0.9846,
+      "step": 13309
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.120005357382026e-05,
+      "loss": 0.9884,
+      "step": 13310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.119112460377695e-05,
+      "loss": 0.9998,
+      "step": 13311
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.118219563373364e-05,
+      "loss": 0.979,
+      "step": 13312
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.117326666369035e-05,
+      "loss": 0.9479,
+      "step": 13313
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.116433769364704e-05,
+      "loss": 1.1296,
+      "step": 13314
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.115540872360373e-05,
+      "loss": 0.9487,
+      "step": 13315
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.114647975356043e-05,
+      "loss": 1.1029,
+      "step": 13316
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.113755078351712e-05,
+      "loss": 1.0478,
+      "step": 13317
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.112862181347381e-05,
+      "loss": 0.9945,
+      "step": 13318
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.111969284343052e-05,
+      "loss": 1.0741,
+      "step": 13319
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.11107638733872e-05,
+      "loss": 1.0346,
+      "step": 13320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.11018349033439e-05,
+      "loss": 1.0101,
+      "step": 13321
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.10929059333006e-05,
+      "loss": 1.0532,
+      "step": 13322
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.108397696325729e-05,
+      "loss": 1.0991,
+      "step": 13323
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.107504799321398e-05,
+      "loss": 1.0205,
+      "step": 13324
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.106611902317069e-05,
+      "loss": 1.0857,
+      "step": 13325
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.105719005312738e-05,
+      "loss": 0.94,
+      "step": 13326
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.104826108308407e-05,
+      "loss": 0.9901,
+      "step": 13327
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.103933211304077e-05,
+      "loss": 0.9934,
+      "step": 13328
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.103040314299746e-05,
+      "loss": 1.0491,
+      "step": 13329
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.102147417295415e-05,
+      "loss": 1.0173,
+      "step": 13330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.101254520291086e-05,
+      "loss": 0.964,
+      "step": 13331
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.100361623286755e-05,
+      "loss": 0.9803,
+      "step": 13332
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.099468726282424e-05,
+      "loss": 0.9972,
+      "step": 13333
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.098575829278093e-05,
+      "loss": 1.0696,
+      "step": 13334
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.097682932273762e-05,
+      "loss": 1.0099,
+      "step": 13335
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.096790035269432e-05,
+      "loss": 1.3374,
+      "step": 13336
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.095897138265101e-05,
+      "loss": 0.8924,
+      "step": 13337
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.09500424126077e-05,
+      "loss": 1.117,
+      "step": 13338
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.094111344256441e-05,
+      "loss": 1.0282,
+      "step": 13339
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.09321844725211e-05,
+      "loss": 1.0301,
+      "step": 13340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.092325550247779e-05,
+      "loss": 1.086,
+      "step": 13341
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.091432653243449e-05,
+      "loss": 1.0879,
+      "step": 13342
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.090539756239118e-05,
+      "loss": 1.0785,
+      "step": 13343
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.089646859234787e-05,
+      "loss": 1.0087,
+      "step": 13344
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.088753962230458e-05,
+      "loss": 1.0147,
+      "step": 13345
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.087861065226127e-05,
+      "loss": 0.9821,
+      "step": 13346
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.086968168221796e-05,
+      "loss": 1.0975,
+      "step": 13347
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.086075271217466e-05,
+      "loss": 1.1355,
+      "step": 13348
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.085182374213135e-05,
+      "loss": 0.9912,
+      "step": 13349
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.084289477208804e-05,
+      "loss": 0.9848,
+      "step": 13350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.083396580204475e-05,
+      "loss": 0.951,
+      "step": 13351
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.082503683200144e-05,
+      "loss": 0.9656,
+      "step": 13352
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.081610786195813e-05,
+      "loss": 1.0161,
+      "step": 13353
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.080717889191483e-05,
+      "loss": 0.9995,
+      "step": 13354
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.079824992187152e-05,
+      "loss": 0.964,
+      "step": 13355
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.078932095182821e-05,
+      "loss": 1.1116,
+      "step": 13356
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.07803919817849e-05,
+      "loss": 1.0521,
+      "step": 13357
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.07714630117416e-05,
+      "loss": 0.9849,
+      "step": 13358
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.076253404169828e-05,
+      "loss": 1.0078,
+      "step": 13359
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.075360507165499e-05,
+      "loss": 1.1811,
+      "step": 13360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.074467610161168e-05,
+      "loss": 1.0413,
+      "step": 13361
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.073574713156837e-05,
+      "loss": 0.9854,
+      "step": 13362
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.072681816152507e-05,
+      "loss": 0.9879,
+      "step": 13363
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.071788919148176e-05,
+      "loss": 1.0704,
+      "step": 13364
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.070896022143845e-05,
+      "loss": 0.9778,
+      "step": 13365
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.070003125139516e-05,
+      "loss": 0.9002,
+      "step": 13366
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.069110228135185e-05,
+      "loss": 0.9766,
+      "step": 13367
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.068217331130854e-05,
+      "loss": 0.976,
+      "step": 13368
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.067324434126524e-05,
+      "loss": 1.0355,
+      "step": 13369
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.066431537122193e-05,
+      "loss": 1.0214,
+      "step": 13370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.065538640117862e-05,
+      "loss": 1.1119,
+      "step": 13371
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.064645743113533e-05,
+      "loss": 1.0635,
+      "step": 13372
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.063752846109202e-05,
+      "loss": 0.9527,
+      "step": 13373
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.062859949104871e-05,
+      "loss": 1.0217,
+      "step": 13374
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.061967052100541e-05,
+      "loss": 1.035,
+      "step": 13375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.06107415509621e-05,
+      "loss": 1.1206,
+      "step": 13376
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.06018125809188e-05,
+      "loss": 1.0215,
+      "step": 13377
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.059288361087548e-05,
+      "loss": 1.0451,
+      "step": 13378
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.058395464083219e-05,
+      "loss": 1.2242,
+      "step": 13379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.057502567078888e-05,
+      "loss": 1.0926,
+      "step": 13380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.056609670074557e-05,
+      "loss": 1.0002,
+      "step": 13381
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.055716773070226e-05,
+      "loss": 1.1882,
+      "step": 13382
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.054823876065896e-05,
+      "loss": 0.9243,
+      "step": 13383
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.053930979061565e-05,
+      "loss": 0.9988,
+      "step": 13384
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.053038082057235e-05,
+      "loss": 0.9973,
+      "step": 13385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.052145185052905e-05,
+      "loss": 1.1112,
+      "step": 13386
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.051252288048574e-05,
+      "loss": 1.1071,
+      "step": 13387
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.050359391044243e-05,
+      "loss": 0.9799,
+      "step": 13388
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.049466494039913e-05,
+      "loss": 1.0346,
+      "step": 13389
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.048573597035582e-05,
+      "loss": 0.803,
+      "step": 13390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.047680700031252e-05,
+      "loss": 1.1,
+      "step": 13391
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.046787803026922e-05,
+      "loss": 0.9619,
+      "step": 13392
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.045894906022591e-05,
+      "loss": 1.0645,
+      "step": 13393
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.04500200901826e-05,
+      "loss": 0.918,
+      "step": 13394
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.04410911201393e-05,
+      "loss": 0.9343,
+      "step": 13395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.0432162150096e-05,
+      "loss": 0.9277,
+      "step": 13396
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.042323318005269e-05,
+      "loss": 1.0977,
+      "step": 13397
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.041430421000939e-05,
+      "loss": 0.9776,
+      "step": 13398
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.040537523996608e-05,
+      "loss": 1.164,
+      "step": 13399
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.039644626992277e-05,
+      "loss": 1.0239,
+      "step": 13400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.038751729987946e-05,
+      "loss": 1.0891,
+      "step": 13401
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.037858832983615e-05,
+      "loss": 1.0709,
+      "step": 13402
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.036965935979284e-05,
+      "loss": 0.9148,
+      "step": 13403
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.036073038974955e-05,
+      "loss": 0.9585,
+      "step": 13404
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.035180141970624e-05,
+      "loss": 1.003,
+      "step": 13405
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.034287244966293e-05,
+      "loss": 0.9813,
+      "step": 13406
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.033394347961963e-05,
+      "loss": 1.0789,
+      "step": 13407
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.032501450957632e-05,
+      "loss": 1.0457,
+      "step": 13408
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.031608553953301e-05,
+      "loss": 0.9429,
+      "step": 13409
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.030715656948972e-05,
+      "loss": 0.9552,
+      "step": 13410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.02982275994464e-05,
+      "loss": 1.0055,
+      "step": 13411
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.02892986294031e-05,
+      "loss": 0.9229,
+      "step": 13412
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.02803696593598e-05,
+      "loss": 1.0236,
+      "step": 13413
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.027144068931649e-05,
+      "loss": 1.0435,
+      "step": 13414
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.026251171927318e-05,
+      "loss": 1.1027,
+      "step": 13415
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.025358274922989e-05,
+      "loss": 1.0052,
+      "step": 13416
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.024465377918658e-05,
+      "loss": 1.0619,
+      "step": 13417
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.023572480914327e-05,
+      "loss": 0.9729,
+      "step": 13418
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.022679583909997e-05,
+      "loss": 0.9128,
+      "step": 13419
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.021786686905666e-05,
+      "loss": 1.2363,
+      "step": 13420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.020893789901335e-05,
+      "loss": 1.1427,
+      "step": 13421
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.020000892897006e-05,
+      "loss": 0.9846,
+      "step": 13422
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.019107995892675e-05,
+      "loss": 0.9959,
+      "step": 13423
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.018215098888344e-05,
+      "loss": 1.1025,
+      "step": 13424
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.017322201884013e-05,
+      "loss": 0.9705,
+      "step": 13425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.016429304879682e-05,
+      "loss": 0.9835,
+      "step": 13426
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.015536407875351e-05,
+      "loss": 1.1366,
+      "step": 13427
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.014643510871021e-05,
+      "loss": 0.9861,
+      "step": 13428
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.01375061386669e-05,
+      "loss": 0.9624,
+      "step": 13429
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.01285771686236e-05,
+      "loss": 0.996,
+      "step": 13430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.01196481985803e-05,
+      "loss": 1.0916,
+      "step": 13431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.011071922853699e-05,
+      "loss": 1.1086,
+      "step": 13432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.010179025849369e-05,
+      "loss": 0.8894,
+      "step": 13433
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.009286128845038e-05,
+      "loss": 1.0464,
+      "step": 13434
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.008393231840707e-05,
+      "loss": 0.9473,
+      "step": 13435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.007500334836378e-05,
+      "loss": 1.0664,
+      "step": 13436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.006607437832047e-05,
+      "loss": 1.0121,
+      "step": 13437
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.005714540827716e-05,
+      "loss": 1.1071,
+      "step": 13438
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.004821643823386e-05,
+      "loss": 1.087,
+      "step": 13439
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.003928746819055e-05,
+      "loss": 1.099,
+      "step": 13440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.003035849814724e-05,
+      "loss": 1.111,
+      "step": 13441
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.002142952810395e-05,
+      "loss": 1.0286,
+      "step": 13442
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.001250055806064e-05,
+      "loss": 1.057,
+      "step": 13443
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.000357158801733e-05,
+      "loss": 1.2091,
+      "step": 13444
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.999464261797403e-05,
+      "loss": 1.0428,
+      "step": 13445
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.998571364793072e-05,
+      "loss": 1.021,
+      "step": 13446
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.997678467788741e-05,
+      "loss": 1.0846,
+      "step": 13447
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.99678557078441e-05,
+      "loss": 1.0927,
+      "step": 13448
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.99589267378008e-05,
+      "loss": 1.0025,
+      "step": 13449
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.994999776775748e-05,
+      "loss": 0.9405,
+      "step": 13450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.994106879771419e-05,
+      "loss": 0.9221,
+      "step": 13451
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.993213982767088e-05,
+      "loss": 0.9094,
+      "step": 13452
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.992321085762757e-05,
+      "loss": 1.0645,
+      "step": 13453
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.991428188758427e-05,
+      "loss": 1.0761,
+      "step": 13454
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.990535291754096e-05,
+      "loss": 0.9992,
+      "step": 13455
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.989642394749765e-05,
+      "loss": 1.0995,
+      "step": 13456
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.988749497745436e-05,
+      "loss": 1.0382,
+      "step": 13457
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.987856600741105e-05,
+      "loss": 1.1265,
+      "step": 13458
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.986963703736774e-05,
+      "loss": 1.1351,
+      "step": 13459
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.986070806732444e-05,
+      "loss": 1.0822,
+      "step": 13460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.985177909728113e-05,
+      "loss": 1.1071,
+      "step": 13461
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.984285012723782e-05,
+      "loss": 1.1022,
+      "step": 13462
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.983392115719453e-05,
+      "loss": 1.0331,
+      "step": 13463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.982499218715122e-05,
+      "loss": 1.0547,
+      "step": 13464
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.981606321710791e-05,
+      "loss": 1.0535,
+      "step": 13465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.980713424706461e-05,
+      "loss": 1.1661,
+      "step": 13466
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.97982052770213e-05,
+      "loss": 1.0122,
+      "step": 13467
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.9789276306978e-05,
+      "loss": 1.0301,
+      "step": 13468
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.978034733693468e-05,
+      "loss": 1.032,
+      "step": 13469
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.977141836689137e-05,
+      "loss": 0.992,
+      "step": 13470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.976248939684808e-05,
+      "loss": 0.9642,
+      "step": 13471
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.975356042680477e-05,
+      "loss": 0.9032,
+      "step": 13472
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.974463145676146e-05,
+      "loss": 1.0259,
+      "step": 13473
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.973570248671815e-05,
+      "loss": 1.1774,
+      "step": 13474
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.972677351667485e-05,
+      "loss": 0.9309,
+      "step": 13475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.971784454663154e-05,
+      "loss": 0.9984,
+      "step": 13476
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.970891557658824e-05,
+      "loss": 1.0271,
+      "step": 13477
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.969998660654494e-05,
+      "loss": 1.0764,
+      "step": 13478
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.969105763650163e-05,
+      "loss": 0.9874,
+      "step": 13479
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.968212866645833e-05,
+      "loss": 1.0338,
+      "step": 13480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.967319969641502e-05,
+      "loss": 0.9767,
+      "step": 13481
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.966427072637171e-05,
+      "loss": 1.0726,
+      "step": 13482
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.965534175632842e-05,
+      "loss": 1.0158,
+      "step": 13483
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.964641278628511e-05,
+      "loss": 1.2401,
+      "step": 13484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.96374838162418e-05,
+      "loss": 0.9132,
+      "step": 13485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.96285548461985e-05,
+      "loss": 1.0379,
+      "step": 13486
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.96196258761552e-05,
+      "loss": 0.9951,
+      "step": 13487
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.961069690611188e-05,
+      "loss": 1.1146,
+      "step": 13488
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.960176793606859e-05,
+      "loss": 1.0811,
+      "step": 13489
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.959283896602528e-05,
+      "loss": 0.9696,
+      "step": 13490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.958390999598197e-05,
+      "loss": 1.0593,
+      "step": 13491
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.957498102593866e-05,
+      "loss": 1.0812,
+      "step": 13492
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.956605205589535e-05,
+      "loss": 1.0029,
+      "step": 13493
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.955712308585204e-05,
+      "loss": 0.9401,
+      "step": 13494
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.954819411580875e-05,
+      "loss": 1.0102,
+      "step": 13495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.953926514576544e-05,
+      "loss": 1.087,
+      "step": 13496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.953033617572213e-05,
+      "loss": 1.0038,
+      "step": 13497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.952140720567883e-05,
+      "loss": 1.0474,
+      "step": 13498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.951247823563552e-05,
+      "loss": 0.9226,
+      "step": 13499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.950354926559221e-05,
+      "loss": 0.9751,
+      "step": 13500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.949462029554892e-05,
+      "loss": 0.9231,
+      "step": 13501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.94856913255056e-05,
+      "loss": 0.9832,
+      "step": 13502
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.94767623554623e-05,
+      "loss": 0.9702,
+      "step": 13503
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.9467833385419e-05,
+      "loss": 1.0286,
+      "step": 13504
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.945890441537569e-05,
+      "loss": 0.94,
+      "step": 13505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.944997544533238e-05,
+      "loss": 1.1394,
+      "step": 13506
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.944104647528909e-05,
+      "loss": 0.9337,
+      "step": 13507
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.943211750524578e-05,
+      "loss": 1.0259,
+      "step": 13508
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.942318853520247e-05,
+      "loss": 1.1258,
+      "step": 13509
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.941425956515917e-05,
+      "loss": 1.1213,
+      "step": 13510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.940533059511586e-05,
+      "loss": 1.004,
+      "step": 13511
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.939640162507255e-05,
+      "loss": 1.0376,
+      "step": 13512
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.938747265502926e-05,
+      "loss": 1.0686,
+      "step": 13513
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.937854368498595e-05,
+      "loss": 1.0279,
+      "step": 13514
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.936961471494264e-05,
+      "loss": 1.0108,
+      "step": 13515
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.936068574489933e-05,
+      "loss": 1.0272,
+      "step": 13516
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.935175677485602e-05,
+      "loss": 1.0549,
+      "step": 13517
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.934282780481271e-05,
+      "loss": 1.0807,
+      "step": 13518
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.933389883476941e-05,
+      "loss": 0.9304,
+      "step": 13519
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.93249698647261e-05,
+      "loss": 1.0744,
+      "step": 13520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.931604089468279e-05,
+      "loss": 0.9189,
+      "step": 13521
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.93071119246395e-05,
+      "loss": 1.0489,
+      "step": 13522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.929818295459619e-05,
+      "loss": 0.9824,
+      "step": 13523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.928925398455288e-05,
+      "loss": 1.0155,
+      "step": 13524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.928032501450958e-05,
+      "loss": 1.1107,
+      "step": 13525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.927139604446627e-05,
+      "loss": 0.9882,
+      "step": 13526
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.926246707442296e-05,
+      "loss": 1.0172,
+      "step": 13527
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.925353810437967e-05,
+      "loss": 1.1171,
+      "step": 13528
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.924460913433636e-05,
+      "loss": 0.9484,
+      "step": 13529
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.923568016429306e-05,
+      "loss": 1.0059,
+      "step": 13530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.922675119424975e-05,
+      "loss": 1.0873,
+      "step": 13531
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.921782222420644e-05,
+      "loss": 1.0222,
+      "step": 13532
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.920889325416315e-05,
+      "loss": 0.9717,
+      "step": 13533
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.919996428411984e-05,
+      "loss": 1.0215,
+      "step": 13534
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.919103531407653e-05,
+      "loss": 1.1038,
+      "step": 13535
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.918210634403322e-05,
+      "loss": 1.0338,
+      "step": 13536
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.917317737398992e-05,
+      "loss": 0.9303,
+      "step": 13537
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.916424840394661e-05,
+      "loss": 1.0836,
+      "step": 13538
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.91553194339033e-05,
+      "loss": 0.9722,
+      "step": 13539
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.914639046386e-05,
+      "loss": 0.9445,
+      "step": 13540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.913746149381668e-05,
+      "loss": 1.0758,
+      "step": 13541
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.912853252377339e-05,
+      "loss": 1.0269,
+      "step": 13542
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.911960355373008e-05,
+      "loss": 0.9436,
+      "step": 13543
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.911067458368677e-05,
+      "loss": 1.01,
+      "step": 13544
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.910174561364347e-05,
+      "loss": 0.9826,
+      "step": 13545
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.909281664360016e-05,
+      "loss": 1.0566,
+      "step": 13546
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.908388767355685e-05,
+      "loss": 1.209,
+      "step": 13547
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.907495870351356e-05,
+      "loss": 0.9369,
+      "step": 13548
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.906602973347025e-05,
+      "loss": 0.9958,
+      "step": 13549
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.905710076342694e-05,
+      "loss": 0.9064,
+      "step": 13550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.904817179338364e-05,
+      "loss": 1.0276,
+      "step": 13551
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.903924282334033e-05,
+      "loss": 1.0597,
+      "step": 13552
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.903031385329702e-05,
+      "loss": 1.1138,
+      "step": 13553
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.902138488325373e-05,
+      "loss": 0.9999,
+      "step": 13554
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.901245591321042e-05,
+      "loss": 1.0211,
+      "step": 13555
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.900352694316711e-05,
+      "loss": 1.0443,
+      "step": 13556
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.899459797312381e-05,
+      "loss": 1.0351,
+      "step": 13557
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.89856690030805e-05,
+      "loss": 1.0264,
+      "step": 13558
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.89767400330372e-05,
+      "loss": 1.1161,
+      "step": 13559
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.896781106299388e-05,
+      "loss": 0.9321,
+      "step": 13560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.895888209295057e-05,
+      "loss": 0.9385,
+      "step": 13561
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.894995312290728e-05,
+      "loss": 0.9888,
+      "step": 13562
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.894102415286397e-05,
+      "loss": 1.0194,
+      "step": 13563
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.893209518282066e-05,
+      "loss": 1.0075,
+      "step": 13564
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.892316621277735e-05,
+      "loss": 0.8907,
+      "step": 13565
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.891423724273405e-05,
+      "loss": 1.1093,
+      "step": 13566
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.890530827269074e-05,
+      "loss": 1.1246,
+      "step": 13567
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.889637930264744e-05,
+      "loss": 0.9646,
+      "step": 13568
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.888745033260414e-05,
+      "loss": 0.9619,
+      "step": 13569
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.887852136256083e-05,
+      "loss": 0.954,
+      "step": 13570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.886959239251752e-05,
+      "loss": 0.9888,
+      "step": 13571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.886066342247422e-05,
+      "loss": 0.8434,
+      "step": 13572
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.885173445243091e-05,
+      "loss": 1.0645,
+      "step": 13573
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.88428054823876e-05,
+      "loss": 1.1329,
+      "step": 13574
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.883387651234431e-05,
+      "loss": 1.0362,
+      "step": 13575
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.8824947542301e-05,
+      "loss": 1.0298,
+      "step": 13576
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.88160185722577e-05,
+      "loss": 1.018,
+      "step": 13577
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.88070896022144e-05,
+      "loss": 0.9929,
+      "step": 13578
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.879816063217108e-05,
+      "loss": 1.0833,
+      "step": 13579
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.878923166212779e-05,
+      "loss": 1.1606,
+      "step": 13580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.878030269208448e-05,
+      "loss": 0.9134,
+      "step": 13581
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.877137372204117e-05,
+      "loss": 1.0445,
+      "step": 13582
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.876244475199786e-05,
+      "loss": 0.9136,
+      "step": 13583
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.875351578195455e-05,
+      "loss": 1.0717,
+      "step": 13584
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.874458681191124e-05,
+      "loss": 1.015,
+      "step": 13585
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.873565784186795e-05,
+      "loss": 1.0865,
+      "step": 13586
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.872672887182464e-05,
+      "loss": 1.0209,
+      "step": 13587
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.871779990178133e-05,
+      "loss": 1.1312,
+      "step": 13588
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.870887093173803e-05,
+      "loss": 0.9569,
+      "step": 13589
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.869994196169472e-05,
+      "loss": 0.9725,
+      "step": 13590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.869101299165141e-05,
+      "loss": 1.1535,
+      "step": 13591
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.868208402160812e-05,
+      "loss": 1.036,
+      "step": 13592
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.86731550515648e-05,
+      "loss": 1.0645,
+      "step": 13593
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.86642260815215e-05,
+      "loss": 0.978,
+      "step": 13594
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.86552971114782e-05,
+      "loss": 0.9598,
+      "step": 13595
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.864636814143489e-05,
+      "loss": 1.1894,
+      "step": 13596
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.863743917139158e-05,
+      "loss": 1.0325,
+      "step": 13597
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.862851020134829e-05,
+      "loss": 0.9567,
+      "step": 13598
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.861958123130498e-05,
+      "loss": 1.0188,
+      "step": 13599
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.861065226126167e-05,
+      "loss": 0.9849,
+      "step": 13600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.860172329121837e-05,
+      "loss": 1.1092,
+      "step": 13601
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.859279432117506e-05,
+      "loss": 1.0856,
+      "step": 13602
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.858386535113175e-05,
+      "loss": 1.13,
+      "step": 13603
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.857493638108846e-05,
+      "loss": 1.0766,
+      "step": 13604
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.856600741104515e-05,
+      "loss": 0.8942,
+      "step": 13605
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.855707844100184e-05,
+      "loss": 1.1163,
+      "step": 13606
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.854814947095853e-05,
+      "loss": 1.1018,
+      "step": 13607
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.853922050091522e-05,
+      "loss": 0.9294,
+      "step": 13608
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.853029153087191e-05,
+      "loss": 1.0571,
+      "step": 13609
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.852136256082861e-05,
+      "loss": 0.9726,
+      "step": 13610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.85124335907853e-05,
+      "loss": 1.0013,
+      "step": 13611
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.850350462074199e-05,
+      "loss": 1.0,
+      "step": 13612
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.84945756506987e-05,
+      "loss": 1.0828,
+      "step": 13613
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.848564668065539e-05,
+      "loss": 0.9448,
+      "step": 13614
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.847671771061208e-05,
+      "loss": 1.1003,
+      "step": 13615
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.846778874056878e-05,
+      "loss": 1.2426,
+      "step": 13616
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.845885977052547e-05,
+      "loss": 0.9916,
+      "step": 13617
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.844993080048216e-05,
+      "loss": 1.0656,
+      "step": 13618
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.844100183043887e-05,
+      "loss": 1.1156,
+      "step": 13619
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.843207286039556e-05,
+      "loss": 0.8985,
+      "step": 13620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.842314389035225e-05,
+      "loss": 1.0103,
+      "step": 13621
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.841421492030895e-05,
+      "loss": 1.0308,
+      "step": 13622
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.840528595026564e-05,
+      "loss": 0.9383,
+      "step": 13623
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.839635698022233e-05,
+      "loss": 1.0667,
+      "step": 13624
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.838742801017904e-05,
+      "loss": 1.1433,
+      "step": 13625
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.837849904013573e-05,
+      "loss": 0.8593,
+      "step": 13626
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.836957007009242e-05,
+      "loss": 1.0053,
+      "step": 13627
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.836064110004912e-05,
+      "loss": 0.9386,
+      "step": 13628
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.835171213000581e-05,
+      "loss": 1.0883,
+      "step": 13629
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.83427831599625e-05,
+      "loss": 1.0173,
+      "step": 13630
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.833385418991919e-05,
+      "loss": 1.1484,
+      "step": 13631
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.832492521987588e-05,
+      "loss": 1.0273,
+      "step": 13632
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.831599624983259e-05,
+      "loss": 0.8931,
+      "step": 13633
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.830706727978928e-05,
+      "loss": 0.9275,
+      "step": 13634
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.829813830974597e-05,
+      "loss": 1.0484,
+      "step": 13635
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.828920933970267e-05,
+      "loss": 0.9868,
+      "step": 13636
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.828028036965936e-05,
+      "loss": 0.9944,
+      "step": 13637
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.827135139961605e-05,
+      "loss": 1.0574,
+      "step": 13638
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.826242242957276e-05,
+      "loss": 1.1107,
+      "step": 13639
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.825349345952945e-05,
+      "loss": 1.1025,
+      "step": 13640
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.824456448948614e-05,
+      "loss": 1.0369,
+      "step": 13641
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.823563551944284e-05,
+      "loss": 1.1388,
+      "step": 13642
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.822670654939953e-05,
+      "loss": 0.9424,
+      "step": 13643
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.821777757935622e-05,
+      "loss": 0.9922,
+      "step": 13644
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.820884860931293e-05,
+      "loss": 1.0067,
+      "step": 13645
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.819991963926962e-05,
+      "loss": 1.0739,
+      "step": 13646
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.819099066922631e-05,
+      "loss": 0.9797,
+      "step": 13647
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.818206169918301e-05,
+      "loss": 1.0051,
+      "step": 13648
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.81731327291397e-05,
+      "loss": 1.0146,
+      "step": 13649
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.81642037590964e-05,
+      "loss": 1.0252,
+      "step": 13650
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.815527478905308e-05,
+      "loss": 1.0506,
+      "step": 13651
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.814634581900977e-05,
+      "loss": 0.9359,
+      "step": 13652
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.813741684896648e-05,
+      "loss": 1.0377,
+      "step": 13653
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.812848787892317e-05,
+      "loss": 0.9815,
+      "step": 13654
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.811955890887986e-05,
+      "loss": 0.8833,
+      "step": 13655
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.811062993883655e-05,
+      "loss": 1.0254,
+      "step": 13656
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.810170096879325e-05,
+      "loss": 1.0143,
+      "step": 13657
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.809277199874994e-05,
+      "loss": 1.0368,
+      "step": 13658
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.808384302870663e-05,
+      "loss": 0.9498,
+      "step": 13659
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.807491405866334e-05,
+      "loss": 0.9491,
+      "step": 13660
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.806598508862003e-05,
+      "loss": 1.0567,
+      "step": 13661
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.805705611857672e-05,
+      "loss": 0.9307,
+      "step": 13662
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.804812714853342e-05,
+      "loss": 1.2184,
+      "step": 13663
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.803919817849011e-05,
+      "loss": 1.1363,
+      "step": 13664
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.80302692084468e-05,
+      "loss": 0.9749,
+      "step": 13665
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.802134023840351e-05,
+      "loss": 1.0901,
+      "step": 13666
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.80124112683602e-05,
+      "loss": 0.9693,
+      "step": 13667
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.800348229831689e-05,
+      "loss": 0.9537,
+      "step": 13668
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.79945533282736e-05,
+      "loss": 1.0794,
+      "step": 13669
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.798562435823028e-05,
+      "loss": 1.1296,
+      "step": 13670
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.797669538818697e-05,
+      "loss": 0.9344,
+      "step": 13671
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.796776641814368e-05,
+      "loss": 1.0704,
+      "step": 13672
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.795883744810037e-05,
+      "loss": 0.9592,
+      "step": 13673
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.794990847805706e-05,
+      "loss": 1.066,
+      "step": 13674
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.794097950801375e-05,
+      "loss": 1.0647,
+      "step": 13675
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.793205053797044e-05,
+      "loss": 1.0312,
+      "step": 13676
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.792312156792714e-05,
+      "loss": 0.9345,
+      "step": 13677
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.791419259788384e-05,
+      "loss": 1.0549,
+      "step": 13678
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.790526362784053e-05,
+      "loss": 0.9386,
+      "step": 13679
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.789633465779723e-05,
+      "loss": 0.9557,
+      "step": 13680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.788740568775392e-05,
+      "loss": 1.0047,
+      "step": 13681
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.787847671771061e-05,
+      "loss": 1.0684,
+      "step": 13682
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.786954774766731e-05,
+      "loss": 1.1424,
+      "step": 13683
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.7860618777624e-05,
+      "loss": 1.0542,
+      "step": 13684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.78516898075807e-05,
+      "loss": 1.0024,
+      "step": 13685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.78427608375374e-05,
+      "loss": 1.074,
+      "step": 13686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.783383186749409e-05,
+      "loss": 0.9795,
+      "step": 13687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.782490289745078e-05,
+      "loss": 1.0273,
+      "step": 13688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.781597392740748e-05,
+      "loss": 1.013,
+      "step": 13689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.780704495736418e-05,
+      "loss": 1.0637,
+      "step": 13690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.779811598732087e-05,
+      "loss": 0.9705,
+      "step": 13691
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.778918701727757e-05,
+      "loss": 1.0138,
+      "step": 13692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.778025804723426e-05,
+      "loss": 1.0384,
+      "step": 13693
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.777132907719095e-05,
+      "loss": 0.9502,
+      "step": 13694
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.776240010714765e-05,
+      "loss": 1.0818,
+      "step": 13695
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.775347113710435e-05,
+      "loss": 0.8853,
+      "step": 13696
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.774454216706104e-05,
+      "loss": 1.0455,
+      "step": 13697
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.773561319701773e-05,
+      "loss": 1.1303,
+      "step": 13698
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.772668422697442e-05,
+      "loss": 0.9935,
+      "step": 13699
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.771775525693111e-05,
+      "loss": 1.0481,
+      "step": 13700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.770882628688781e-05,
+      "loss": 0.9862,
+      "step": 13701
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.76998973168445e-05,
+      "loss": 1.0125,
+      "step": 13702
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.769096834680119e-05,
+      "loss": 1.0685,
+      "step": 13703
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.76820393767579e-05,
+      "loss": 1.0345,
+      "step": 13704
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.767311040671459e-05,
+      "loss": 1.0959,
+      "step": 13705
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.766418143667128e-05,
+      "loss": 0.9932,
+      "step": 13706
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.765525246662798e-05,
+      "loss": 1.0013,
+      "step": 13707
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.764632349658467e-05,
+      "loss": 1.0801,
+      "step": 13708
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.763739452654136e-05,
+      "loss": 0.9868,
+      "step": 13709
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.762846555649807e-05,
+      "loss": 0.8795,
+      "step": 13710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.761953658645476e-05,
+      "loss": 1.0922,
+      "step": 13711
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.761060761641145e-05,
+      "loss": 1.066,
+      "step": 13712
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.760167864636815e-05,
+      "loss": 1.0918,
+      "step": 13713
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.759274967632484e-05,
+      "loss": 0.9946,
+      "step": 13714
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.758382070628153e-05,
+      "loss": 0.9677,
+      "step": 13715
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.757489173623824e-05,
+      "loss": 1.0136,
+      "step": 13716
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.756596276619493e-05,
+      "loss": 0.9748,
+      "step": 13717
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.755703379615162e-05,
+      "loss": 0.9836,
+      "step": 13718
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.754810482610832e-05,
+      "loss": 1.0603,
+      "step": 13719
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.753917585606501e-05,
+      "loss": 0.9489,
+      "step": 13720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.75302468860217e-05,
+      "loss": 0.9317,
+      "step": 13721
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.752131791597839e-05,
+      "loss": 1.0447,
+      "step": 13722
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.751238894593508e-05,
+      "loss": 1.0434,
+      "step": 13723
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.750345997589179e-05,
+      "loss": 1.1071,
+      "step": 13724
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.749453100584848e-05,
+      "loss": 1.0379,
+      "step": 13725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.748560203580517e-05,
+      "loss": 1.1358,
+      "step": 13726
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.747667306576187e-05,
+      "loss": 1.0058,
+      "step": 13727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.746774409571856e-05,
+      "loss": 1.2072,
+      "step": 13728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.745881512567525e-05,
+      "loss": 0.9894,
+      "step": 13729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.744988615563196e-05,
+      "loss": 1.6627,
+      "step": 13730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.744095718558865e-05,
+      "loss": 1.0489,
+      "step": 13731
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.743202821554534e-05,
+      "loss": 0.9694,
+      "step": 13732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.742309924550204e-05,
+      "loss": 0.9589,
+      "step": 13733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.741417027545873e-05,
+      "loss": 1.0455,
+      "step": 13734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.740524130541542e-05,
+      "loss": 1.0829,
+      "step": 13735
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.739631233537213e-05,
+      "loss": 1.0924,
+      "step": 13736
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.738738336532882e-05,
+      "loss": 0.9726,
+      "step": 13737
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.737845439528551e-05,
+      "loss": 1.0661,
+      "step": 13738
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.736952542524221e-05,
+      "loss": 0.9394,
+      "step": 13739
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.73605964551989e-05,
+      "loss": 1.1954,
+      "step": 13740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.735166748515559e-05,
+      "loss": 1.0785,
+      "step": 13741
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.734273851511228e-05,
+      "loss": 0.9204,
+      "step": 13742
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.733380954506897e-05,
+      "loss": 1.0661,
+      "step": 13743
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.732488057502566e-05,
+      "loss": 0.9834,
+      "step": 13744
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.731595160498237e-05,
+      "loss": 0.9945,
+      "step": 13745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.730702263493906e-05,
+      "loss": 1.0812,
+      "step": 13746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.729809366489575e-05,
+      "loss": 1.1977,
+      "step": 13747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.728916469485245e-05,
+      "loss": 1.0587,
+      "step": 13748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.728023572480914e-05,
+      "loss": 1.0769,
+      "step": 13749
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.727130675476583e-05,
+      "loss": 1.1012,
+      "step": 13750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.726237778472254e-05,
+      "loss": 1.077,
+      "step": 13751
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.725344881467923e-05,
+      "loss": 1.0083,
+      "step": 13752
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.724451984463592e-05,
+      "loss": 1.0133,
+      "step": 13753
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.723559087459262e-05,
+      "loss": 1.1665,
+      "step": 13754
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.722666190454931e-05,
+      "loss": 0.9693,
+      "step": 13755
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.7217732934506e-05,
+      "loss": 0.9471,
+      "step": 13756
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.720880396446271e-05,
+      "loss": 1.0044,
+      "step": 13757
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.71998749944194e-05,
+      "loss": 1.1243,
+      "step": 13758
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.719094602437609e-05,
+      "loss": 1.0055,
+      "step": 13759
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.71820170543328e-05,
+      "loss": 0.9664,
+      "step": 13760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.717308808428948e-05,
+      "loss": 1.0345,
+      "step": 13761
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.716415911424617e-05,
+      "loss": 1.0281,
+      "step": 13762
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.715523014420288e-05,
+      "loss": 0.9504,
+      "step": 13763
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.714630117415957e-05,
+      "loss": 1.0322,
+      "step": 13764
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.713737220411626e-05,
+      "loss": 1.0632,
+      "step": 13765
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.712844323407295e-05,
+      "loss": 1.0126,
+      "step": 13766
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.711951426402964e-05,
+      "loss": 1.0004,
+      "step": 13767
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.711058529398633e-05,
+      "loss": 0.9913,
+      "step": 13768
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.710165632394303e-05,
+      "loss": 1.0259,
+      "step": 13769
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.709272735389973e-05,
+      "loss": 1.0396,
+      "step": 13770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.708379838385642e-05,
+      "loss": 0.9967,
+      "step": 13771
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.707486941381312e-05,
+      "loss": 1.0596,
+      "step": 13772
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.706594044376981e-05,
+      "loss": 1.0934,
+      "step": 13773
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.705701147372651e-05,
+      "loss": 1.0831,
+      "step": 13774
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.70480825036832e-05,
+      "loss": 1.0594,
+      "step": 13775
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.70391535336399e-05,
+      "loss": 1.2265,
+      "step": 13776
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.70302245635966e-05,
+      "loss": 1.0142,
+      "step": 13777
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.702129559355329e-05,
+      "loss": 0.9765,
+      "step": 13778
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.701236662350998e-05,
+      "loss": 0.917,
+      "step": 13779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.700343765346668e-05,
+      "loss": 1.0551,
+      "step": 13780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.699450868342337e-05,
+      "loss": 1.136,
+      "step": 13781
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.698557971338007e-05,
+      "loss": 1.1482,
+      "step": 13782
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.697665074333677e-05,
+      "loss": 0.972,
+      "step": 13783
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.696772177329346e-05,
+      "loss": 0.9936,
+      "step": 13784
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.695879280325015e-05,
+      "loss": 0.9906,
+      "step": 13785
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.694986383320685e-05,
+      "loss": 0.9203,
+      "step": 13786
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.694093486316354e-05,
+      "loss": 1.0495,
+      "step": 13787
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.693200589312024e-05,
+      "loss": 0.9502,
+      "step": 13788
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.8793,
+      "step": 13789
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.691414795303362e-05,
+      "loss": 1.0169,
+      "step": 13790
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.69052189829903e-05,
+      "loss": 1.017,
+      "step": 13791
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.689629001294701e-05,
+      "loss": 1.1701,
+      "step": 13792
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.68873610429037e-05,
+      "loss": 1.0519,
+      "step": 13793
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.687843207286039e-05,
+      "loss": 1.0747,
+      "step": 13794
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.68695031028171e-05,
+      "loss": 1.0527,
+      "step": 13795
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.686057413277379e-05,
+      "loss": 1.0948,
+      "step": 13796
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.685164516273048e-05,
+      "loss": 1.0002,
+      "step": 13797
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.684271619268718e-05,
+      "loss": 1.0402,
+      "step": 13798
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.683378722264387e-05,
+      "loss": 0.9434,
+      "step": 13799
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.682485825260056e-05,
+      "loss": 0.9684,
+      "step": 13800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.681592928255727e-05,
+      "loss": 1.0254,
+      "step": 13801
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.680700031251396e-05,
+      "loss": 1.1808,
+      "step": 13802
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.679807134247065e-05,
+      "loss": 0.9498,
+      "step": 13803
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.678914237242735e-05,
+      "loss": 1.0305,
+      "step": 13804
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.678021340238404e-05,
+      "loss": 1.1391,
+      "step": 13805
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.677128443234073e-05,
+      "loss": 0.9653,
+      "step": 13806
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.676235546229744e-05,
+      "loss": 1.0138,
+      "step": 13807
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.675342649225413e-05,
+      "loss": 1.1058,
+      "step": 13808
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.674449752221082e-05,
+      "loss": 1.0081,
+      "step": 13809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.673556855216751e-05,
+      "loss": 0.9543,
+      "step": 13810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.672663958212421e-05,
+      "loss": 0.9746,
+      "step": 13811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.67177106120809e-05,
+      "loss": 1.1307,
+      "step": 13812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.670878164203759e-05,
+      "loss": 1.2453,
+      "step": 13813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.669985267199428e-05,
+      "loss": 0.9655,
+      "step": 13814
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.669092370195097e-05,
+      "loss": 1.0267,
+      "step": 13815
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.668199473190768e-05,
+      "loss": 1.1487,
+      "step": 13816
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.667306576186437e-05,
+      "loss": 1.0181,
+      "step": 13817
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.666413679182106e-05,
+      "loss": 0.9064,
+      "step": 13818
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.665520782177776e-05,
+      "loss": 1.0686,
+      "step": 13819
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.664627885173445e-05,
+      "loss": 0.9717,
+      "step": 13820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.663734988169116e-05,
+      "loss": 1.0626,
+      "step": 13821
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.662842091164785e-05,
+      "loss": 1.0246,
+      "step": 13822
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.661949194160454e-05,
+      "loss": 1.0869,
+      "step": 13823
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.661056297156124e-05,
+      "loss": 1.151,
+      "step": 13824
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.660163400151793e-05,
+      "loss": 0.941,
+      "step": 13825
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.659270503147462e-05,
+      "loss": 1.0147,
+      "step": 13826
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.658377606143133e-05,
+      "loss": 1.0269,
+      "step": 13827
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.657484709138802e-05,
+      "loss": 0.9255,
+      "step": 13828
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.656591812134471e-05,
+      "loss": 1.019,
+      "step": 13829
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.655698915130141e-05,
+      "loss": 1.0665,
+      "step": 13830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.65480601812581e-05,
+      "loss": 0.971,
+      "step": 13831
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.653913121121479e-05,
+      "loss": 1.1768,
+      "step": 13832
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.653020224117148e-05,
+      "loss": 1.0386,
+      "step": 13833
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.652127327112817e-05,
+      "loss": 1.1899,
+      "step": 13834
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.651234430108486e-05,
+      "loss": 1.1636,
+      "step": 13835
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.650341533104157e-05,
+      "loss": 1.0886,
+      "step": 13836
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.649448636099826e-05,
+      "loss": 0.9699,
+      "step": 13837
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.648555739095495e-05,
+      "loss": 1.0165,
+      "step": 13838
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.647662842091165e-05,
+      "loss": 0.9329,
+      "step": 13839
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.646769945086834e-05,
+      "loss": 1.0812,
+      "step": 13840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.645877048082503e-05,
+      "loss": 1.0744,
+      "step": 13841
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.644984151078174e-05,
+      "loss": 1.0405,
+      "step": 13842
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.644091254073843e-05,
+      "loss": 1.0722,
+      "step": 13843
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.643198357069512e-05,
+      "loss": 0.973,
+      "step": 13844
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.642305460065182e-05,
+      "loss": 1.024,
+      "step": 13845
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.641412563060851e-05,
+      "loss": 1.2293,
+      "step": 13846
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.64051966605652e-05,
+      "loss": 1.0189,
+      "step": 13847
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.639626769052191e-05,
+      "loss": 1.0123,
+      "step": 13848
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.63873387204786e-05,
+      "loss": 0.9914,
+      "step": 13849
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.637840975043529e-05,
+      "loss": 0.9947,
+      "step": 13850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.636948078039199e-05,
+      "loss": 1.0347,
+      "step": 13851
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.636055181034868e-05,
+      "loss": 0.9333,
+      "step": 13852
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.635162284030537e-05,
+      "loss": 0.9729,
+      "step": 13853
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.634269387026208e-05,
+      "loss": 0.9996,
+      "step": 13854
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.633376490021877e-05,
+      "loss": 1.1309,
+      "step": 13855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.632483593017546e-05,
+      "loss": 1.03,
+      "step": 13856
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.631590696013215e-05,
+      "loss": 0.9919,
+      "step": 13857
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.630697799008884e-05,
+      "loss": 1.0397,
+      "step": 13858
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.629804902004553e-05,
+      "loss": 0.967,
+      "step": 13859
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.628912005000223e-05,
+      "loss": 0.9755,
+      "step": 13860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.628019107995892e-05,
+      "loss": 1.1941,
+      "step": 13861
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.627126210991562e-05,
+      "loss": 1.115,
+      "step": 13862
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.626233313987232e-05,
+      "loss": 1.0016,
+      "step": 13863
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.625340416982901e-05,
+      "loss": 1.06,
+      "step": 13864
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.62444751997857e-05,
+      "loss": 1.0737,
+      "step": 13865
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.62355462297424e-05,
+      "loss": 0.988,
+      "step": 13866
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.62266172596991e-05,
+      "loss": 1.3742,
+      "step": 13867
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.621768828965579e-05,
+      "loss": 0.9862,
+      "step": 13868
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.620875931961249e-05,
+      "loss": 1.1505,
+      "step": 13869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.619983034956918e-05,
+      "loss": 0.9986,
+      "step": 13870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.619090137952588e-05,
+      "loss": 0.9712,
+      "step": 13871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.618197240948257e-05,
+      "loss": 0.9168,
+      "step": 13872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.617304343943926e-05,
+      "loss": 0.9137,
+      "step": 13873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.616411446939597e-05,
+      "loss": 0.9222,
+      "step": 13874
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.615518549935266e-05,
+      "loss": 0.983,
+      "step": 13875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.614625652930935e-05,
+      "loss": 1.3093,
+      "step": 13876
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.613732755926605e-05,
+      "loss": 1.2111,
+      "step": 13877
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.612839858922274e-05,
+      "loss": 1.0303,
+      "step": 13878
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.611946961917943e-05,
+      "loss": 1.0936,
+      "step": 13879
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.611054064913613e-05,
+      "loss": 0.9762,
+      "step": 13880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.610161167909282e-05,
+      "loss": 1.0229,
+      "step": 13881
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.60926827090495e-05,
+      "loss": 1.0554,
+      "step": 13882
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.608375373900621e-05,
+      "loss": 1.0832,
+      "step": 13883
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.60748247689629e-05,
+      "loss": 1.0219,
+      "step": 13884
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.606589579891959e-05,
+      "loss": 1.0505,
+      "step": 13885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.60569668288763e-05,
+      "loss": 1.019,
+      "step": 13886
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.604803785883299e-05,
+      "loss": 1.0106,
+      "step": 13887
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.603910888878968e-05,
+      "loss": 0.9776,
+      "step": 13888
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.603017991874638e-05,
+      "loss": 1.0313,
+      "step": 13889
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.602125094870307e-05,
+      "loss": 1.0728,
+      "step": 13890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.601232197865976e-05,
+      "loss": 1.0395,
+      "step": 13891
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.600339300861647e-05,
+      "loss": 0.9588,
+      "step": 13892
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.599446403857316e-05,
+      "loss": 0.9836,
+      "step": 13893
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.598553506852985e-05,
+      "loss": 1.1126,
+      "step": 13894
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.597660609848655e-05,
+      "loss": 0.9814,
+      "step": 13895
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.596767712844324e-05,
+      "loss": 0.9967,
+      "step": 13896
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.595874815839993e-05,
+      "loss": 1.1279,
+      "step": 13897
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.594981918835664e-05,
+      "loss": 0.9592,
+      "step": 13898
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.594089021831333e-05,
+      "loss": 0.9909,
+      "step": 13899
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.593196124827002e-05,
+      "loss": 0.9815,
+      "step": 13900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.59230322782267e-05,
+      "loss": 0.938,
+      "step": 13901
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.591410330818341e-05,
+      "loss": 1.1352,
+      "step": 13902
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.59051743381401e-05,
+      "loss": 0.9999,
+      "step": 13903
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.589624536809679e-05,
+      "loss": 1.0329,
+      "step": 13904
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.588731639805348e-05,
+      "loss": 0.9707,
+      "step": 13905
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.587838742801017e-05,
+      "loss": 1.0466,
+      "step": 13906
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.586945845796688e-05,
+      "loss": 1.094,
+      "step": 13907
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.586052948792357e-05,
+      "loss": 1.0739,
+      "step": 13908
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.585160051788026e-05,
+      "loss": 1.0328,
+      "step": 13909
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.584267154783696e-05,
+      "loss": 1.0371,
+      "step": 13910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.583374257779365e-05,
+      "loss": 0.9416,
+      "step": 13911
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.582481360775034e-05,
+      "loss": 1.2365,
+      "step": 13912
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.581588463770705e-05,
+      "loss": 0.904,
+      "step": 13913
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.580695566766374e-05,
+      "loss": 0.9788,
+      "step": 13914
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.579802669762043e-05,
+      "loss": 1.0748,
+      "step": 13915
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.578909772757713e-05,
+      "loss": 0.9791,
+      "step": 13916
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.578016875753382e-05,
+      "loss": 0.9542,
+      "step": 13917
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.577123978749051e-05,
+      "loss": 1.0867,
+      "step": 13918
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.576231081744722e-05,
+      "loss": 1.0819,
+      "step": 13919
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.575338184740391e-05,
+      "loss": 1.4429,
+      "step": 13920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.574445287736061e-05,
+      "loss": 0.988,
+      "step": 13921
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.57355239073173e-05,
+      "loss": 0.9528,
+      "step": 13922
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.572659493727399e-05,
+      "loss": 1.0782,
+      "step": 13923
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.571766596723068e-05,
+      "loss": 0.9789,
+      "step": 13924
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.570873699718737e-05,
+      "loss": 0.9623,
+      "step": 13925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.569980802714406e-05,
+      "loss": 1.0363,
+      "step": 13926
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.569087905710077e-05,
+      "loss": 1.0273,
+      "step": 13927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.568195008705746e-05,
+      "loss": 1.0988,
+      "step": 13928
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.567302111701415e-05,
+      "loss": 1.0301,
+      "step": 13929
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.566409214697085e-05,
+      "loss": 1.0549,
+      "step": 13930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.565516317692754e-05,
+      "loss": 0.9695,
+      "step": 13931
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.564623420688423e-05,
+      "loss": 1.028,
+      "step": 13932
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.563730523684094e-05,
+      "loss": 1.0845,
+      "step": 13933
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.562837626679763e-05,
+      "loss": 0.9959,
+      "step": 13934
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.561944729675432e-05,
+      "loss": 1.0105,
+      "step": 13935
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.561051832671102e-05,
+      "loss": 1.0638,
+      "step": 13936
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.560158935666771e-05,
+      "loss": 1.0284,
+      "step": 13937
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.55926603866244e-05,
+      "loss": 0.9336,
+      "step": 13938
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.558373141658111e-05,
+      "loss": 0.9531,
+      "step": 13939
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.55748024465378e-05,
+      "loss": 0.9432,
+      "step": 13940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.556587347649449e-05,
+      "loss": 1.0388,
+      "step": 13941
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.555694450645119e-05,
+      "loss": 1.0005,
+      "step": 13942
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.554801553640788e-05,
+      "loss": 1.0421,
+      "step": 13943
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.553908656636457e-05,
+      "loss": 1.1017,
+      "step": 13944
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.553015759632128e-05,
+      "loss": 1.0283,
+      "step": 13945
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.552122862627797e-05,
+      "loss": 0.9857,
+      "step": 13946
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.551229965623466e-05,
+      "loss": 1.0621,
+      "step": 13947
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.550337068619135e-05,
+      "loss": 0.9586,
+      "step": 13948
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.549444171614804e-05,
+      "loss": 1.2633,
+      "step": 13949
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.548551274610473e-05,
+      "loss": 0.993,
+      "step": 13950
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.547658377606143e-05,
+      "loss": 1.1459,
+      "step": 13951
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.546765480601812e-05,
+      "loss": 0.9582,
+      "step": 13952
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.545872583597482e-05,
+      "loss": 1.03,
+      "step": 13953
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.544979686593152e-05,
+      "loss": 1.0374,
+      "step": 13954
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.544086789588821e-05,
+      "loss": 1.0164,
+      "step": 13955
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.54319389258449e-05,
+      "loss": 0.9805,
+      "step": 13956
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.54230099558016e-05,
+      "loss": 1.0175,
+      "step": 13957
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.54140809857583e-05,
+      "loss": 1.0492,
+      "step": 13958
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.540515201571499e-05,
+      "loss": 0.8943,
+      "step": 13959
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.539622304567169e-05,
+      "loss": 1.2851,
+      "step": 13960
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.538729407562838e-05,
+      "loss": 1.0174,
+      "step": 13961
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.537836510558507e-05,
+      "loss": 1.031,
+      "step": 13962
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.536943613554177e-05,
+      "loss": 0.9691,
+      "step": 13963
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.536050716549846e-05,
+      "loss": 0.9977,
+      "step": 13964
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.535157819545515e-05,
+      "loss": 0.9356,
+      "step": 13965
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.534264922541186e-05,
+      "loss": 0.991,
+      "step": 13966
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.533372025536855e-05,
+      "loss": 1.1239,
+      "step": 13967
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.532479128532525e-05,
+      "loss": 1.0655,
+      "step": 13968
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.531586231528194e-05,
+      "loss": 1.0247,
+      "step": 13969
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.530693334523863e-05,
+      "loss": 0.945,
+      "step": 13970
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.529800437519532e-05,
+      "loss": 1.0317,
+      "step": 13971
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.528907540515202e-05,
+      "loss": 1.0941,
+      "step": 13972
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.52801464351087e-05,
+      "loss": 1.0812,
+      "step": 13973
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.527121746506541e-05,
+      "loss": 0.9088,
+      "step": 13974
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.52622884950221e-05,
+      "loss": 0.9675,
+      "step": 13975
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.525335952497879e-05,
+      "loss": 0.9981,
+      "step": 13976
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.52444305549355e-05,
+      "loss": 0.9661,
+      "step": 13977
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.523550158489219e-05,
+      "loss": 1.0346,
+      "step": 13978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.522657261484888e-05,
+      "loss": 1.0617,
+      "step": 13979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.521764364480558e-05,
+      "loss": 1.0454,
+      "step": 13980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.520871467476227e-05,
+      "loss": 0.9945,
+      "step": 13981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.519978570471896e-05,
+      "loss": 1.0094,
+      "step": 13982
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.519085673467566e-05,
+      "loss": 1.0009,
+      "step": 13983
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.518192776463236e-05,
+      "loss": 0.9927,
+      "step": 13984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.517299879458905e-05,
+      "loss": 1.0852,
+      "step": 13985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.516406982454575e-05,
+      "loss": 1.0433,
+      "step": 13986
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.515514085450244e-05,
+      "loss": 0.9993,
+      "step": 13987
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.514621188445913e-05,
+      "loss": 0.9816,
+      "step": 13988
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.513728291441583e-05,
+      "loss": 1.0052,
+      "step": 13989
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.512835394437253e-05,
+      "loss": 1.3647,
+      "step": 13990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.511942497432922e-05,
+      "loss": 0.914,
+      "step": 13991
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.51104960042859e-05,
+      "loss": 1.0104,
+      "step": 13992
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.510156703424261e-05,
+      "loss": 1.004,
+      "step": 13993
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.50926380641993e-05,
+      "loss": 0.972,
+      "step": 13994
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.508370909415599e-05,
+      "loss": 1.1323,
+      "step": 13995
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.507478012411268e-05,
+      "loss": 1.0607,
+      "step": 13996
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.506585115406937e-05,
+      "loss": 0.8974,
+      "step": 13997
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.505692218402608e-05,
+      "loss": 1.0494,
+      "step": 13998
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.504799321398277e-05,
+      "loss": 0.9859,
+      "step": 13999
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.503906424393946e-05,
+      "loss": 1.1023,
+      "step": 14000
+    },
+    {
+      "epoch": 1.87,
+      "eval_loss": 1.025034785270691,
+      "eval_runtime": 3047.4063,
+      "eval_samples_per_second": 6.171,
+      "eval_steps_per_second": 0.771,
+      "step": 14000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.503013527389616e-05,
+      "loss": 1.0073,
+      "step": 14001
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.502120630385285e-05,
+      "loss": 1.0132,
+      "step": 14002
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.501227733380954e-05,
+      "loss": 1.0496,
+      "step": 14003
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.500334836376625e-05,
+      "loss": 1.243,
+      "step": 14004
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.499441939372294e-05,
+      "loss": 0.892,
+      "step": 14005
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.498549042367963e-05,
+      "loss": 1.0429,
+      "step": 14006
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.497656145363633e-05,
+      "loss": 1.0251,
+      "step": 14007
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.496763248359302e-05,
+      "loss": 1.0724,
+      "step": 14008
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.495870351354971e-05,
+      "loss": 1.0073,
+      "step": 14009
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.494977454350642e-05,
+      "loss": 0.9842,
+      "step": 14010
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.494084557346311e-05,
+      "loss": 1.0966,
+      "step": 14011
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.49319166034198e-05,
+      "loss": 1.2416,
+      "step": 14012
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.49229876333765e-05,
+      "loss": 1.0647,
+      "step": 14013
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.491405866333319e-05,
+      "loss": 0.9732,
+      "step": 14014
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.490512969328988e-05,
+      "loss": 1.1182,
+      "step": 14015
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.489620072324657e-05,
+      "loss": 1.08,
+      "step": 14016
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.488727175320326e-05,
+      "loss": 0.9501,
+      "step": 14017
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.487834278315997e-05,
+      "loss": 0.9305,
+      "step": 14018
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.486941381311666e-05,
+      "loss": 0.9813,
+      "step": 14019
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.486048484307335e-05,
+      "loss": 1.03,
+      "step": 14020
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.485155587303005e-05,
+      "loss": 1.0889,
+      "step": 14021
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.484262690298674e-05,
+      "loss": 1.0667,
+      "step": 14022
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.483369793294343e-05,
+      "loss": 0.9407,
+      "step": 14023
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.482476896290014e-05,
+      "loss": 1.04,
+      "step": 14024
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.481583999285683e-05,
+      "loss": 1.0365,
+      "step": 14025
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.480691102281352e-05,
+      "loss": 0.9993,
+      "step": 14026
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.479798205277022e-05,
+      "loss": 0.9793,
+      "step": 14027
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.478905308272691e-05,
+      "loss": 1.1475,
+      "step": 14028
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.47801241126836e-05,
+      "loss": 0.868,
+      "step": 14029
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.477119514264031e-05,
+      "loss": 0.9408,
+      "step": 14030
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4762266172597e-05,
+      "loss": 0.9071,
+      "step": 14031
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.475333720255369e-05,
+      "loss": 0.8902,
+      "step": 14032
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.474440823251039e-05,
+      "loss": 1.0849,
+      "step": 14033
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.473547926246708e-05,
+      "loss": 1.0747,
+      "step": 14034
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.472655029242377e-05,
+      "loss": 1.0193,
+      "step": 14035
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.471762132238048e-05,
+      "loss": 0.9332,
+      "step": 14036
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.470869235233717e-05,
+      "loss": 1.0247,
+      "step": 14037
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.469976338229386e-05,
+      "loss": 1.0348,
+      "step": 14038
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.469083441225055e-05,
+      "loss": 1.1949,
+      "step": 14039
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.468190544220724e-05,
+      "loss": 1.0271,
+      "step": 14040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.467297647216393e-05,
+      "loss": 1.01,
+      "step": 14041
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.466404750212063e-05,
+      "loss": 1.1136,
+      "step": 14042
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.465511853207732e-05,
+      "loss": 1.0268,
+      "step": 14043
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.464618956203401e-05,
+      "loss": 1.0655,
+      "step": 14044
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.463726059199072e-05,
+      "loss": 0.9868,
+      "step": 14045
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.462833162194741e-05,
+      "loss": 0.9913,
+      "step": 14046
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.46194026519041e-05,
+      "loss": 0.9842,
+      "step": 14047
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.46104736818608e-05,
+      "loss": 1.0357,
+      "step": 14048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.46015447118175e-05,
+      "loss": 0.9444,
+      "step": 14049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.459261574177418e-05,
+      "loss": 1.0623,
+      "step": 14050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.458368677173089e-05,
+      "loss": 0.9594,
+      "step": 14051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.457475780168758e-05,
+      "loss": 1.1087,
+      "step": 14052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.456582883164427e-05,
+      "loss": 0.9771,
+      "step": 14053
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.455689986160097e-05,
+      "loss": 0.9461,
+      "step": 14054
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.454797089155766e-05,
+      "loss": 1.0544,
+      "step": 14055
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.453904192151435e-05,
+      "loss": 1.0733,
+      "step": 14056
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.453011295147106e-05,
+      "loss": 1.1325,
+      "step": 14057
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.452118398142775e-05,
+      "loss": 1.0511,
+      "step": 14058
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.451225501138444e-05,
+      "loss": 0.9729,
+      "step": 14059
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.450332604134114e-05,
+      "loss": 1.0164,
+      "step": 14060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.449439707129783e-05,
+      "loss": 1.0016,
+      "step": 14061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.448546810125452e-05,
+      "loss": 1.0083,
+      "step": 14062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.447653913121122e-05,
+      "loss": 1.0262,
+      "step": 14063
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.44676101611679e-05,
+      "loss": 1.0452,
+      "step": 14064
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.44586811911246e-05,
+      "loss": 1.0858,
+      "step": 14065
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.44497522210813e-05,
+      "loss": 1.0734,
+      "step": 14066
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.444082325103799e-05,
+      "loss": 1.0126,
+      "step": 14067
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.44318942809947e-05,
+      "loss": 1.013,
+      "step": 14068
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.442296531095139e-05,
+      "loss": 1.0835,
+      "step": 14069
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.441403634090808e-05,
+      "loss": 1.033,
+      "step": 14070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.440510737086478e-05,
+      "loss": 0.954,
+      "step": 14071
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.439617840082147e-05,
+      "loss": 1.0504,
+      "step": 14072
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.438724943077816e-05,
+      "loss": 0.9991,
+      "step": 14073
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.437832046073486e-05,
+      "loss": 0.9077,
+      "step": 14074
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.436939149069156e-05,
+      "loss": 1.0788,
+      "step": 14075
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.436046252064825e-05,
+      "loss": 0.9541,
+      "step": 14076
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.435153355060495e-05,
+      "loss": 1.1095,
+      "step": 14077
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.434260458056164e-05,
+      "loss": 1.0405,
+      "step": 14078
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.433367561051833e-05,
+      "loss": 0.9687,
+      "step": 14079
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.432474664047503e-05,
+      "loss": 1.0028,
+      "step": 14080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.431581767043173e-05,
+      "loss": 1.0773,
+      "step": 14081
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.430688870038842e-05,
+      "loss": 1.1684,
+      "step": 14082
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.42979597303451e-05,
+      "loss": 1.0399,
+      "step": 14083
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.42890307603018e-05,
+      "loss": 0.971,
+      "step": 14084
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.42801017902585e-05,
+      "loss": 0.9055,
+      "step": 14085
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.427117282021519e-05,
+      "loss": 0.8801,
+      "step": 14086
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.426224385017188e-05,
+      "loss": 0.9773,
+      "step": 14087
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.425331488012857e-05,
+      "loss": 0.9152,
+      "step": 14088
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.424438591008528e-05,
+      "loss": 1.017,
+      "step": 14089
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.423545694004197e-05,
+      "loss": 0.9249,
+      "step": 14090
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.422652796999866e-05,
+      "loss": 1.1307,
+      "step": 14091
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.421759899995536e-05,
+      "loss": 0.9973,
+      "step": 14092
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.420867002991205e-05,
+      "loss": 0.9972,
+      "step": 14093
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.419974105986874e-05,
+      "loss": 0.967,
+      "step": 14094
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.419081208982545e-05,
+      "loss": 1.0493,
+      "step": 14095
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.418188311978214e-05,
+      "loss": 0.9528,
+      "step": 14096
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.417295414973883e-05,
+      "loss": 1.1006,
+      "step": 14097
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.416402517969553e-05,
+      "loss": 0.9853,
+      "step": 14098
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.415509620965222e-05,
+      "loss": 1.0484,
+      "step": 14099
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.414616723960891e-05,
+      "loss": 1.0311,
+      "step": 14100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.413723826956562e-05,
+      "loss": 1.0376,
+      "step": 14101
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.41283092995223e-05,
+      "loss": 1.0581,
+      "step": 14102
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.4119380329479e-05,
+      "loss": 1.0528,
+      "step": 14103
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.41104513594357e-05,
+      "loss": 1.0488,
+      "step": 14104
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.410152238939239e-05,
+      "loss": 1.0567,
+      "step": 14105
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.409259341934908e-05,
+      "loss": 1.0068,
+      "step": 14106
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.408366444930577e-05,
+      "loss": 1.0142,
+      "step": 14107
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.407473547926246e-05,
+      "loss": 0.8877,
+      "step": 14108
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.406580650921915e-05,
+      "loss": 0.979,
+      "step": 14109
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.405687753917586e-05,
+      "loss": 1.0196,
+      "step": 14110
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.404794856913255e-05,
+      "loss": 1.1009,
+      "step": 14111
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.403901959908924e-05,
+      "loss": 1.0053,
+      "step": 14112
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.403009062904594e-05,
+      "loss": 1.0094,
+      "step": 14113
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.402116165900263e-05,
+      "loss": 0.9882,
+      "step": 14114
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.401223268895934e-05,
+      "loss": 1.1008,
+      "step": 14115
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.400330371891603e-05,
+      "loss": 0.9212,
+      "step": 14116
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.399437474887272e-05,
+      "loss": 1.0048,
+      "step": 14117
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.398544577882942e-05,
+      "loss": 1.0252,
+      "step": 14118
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.397651680878611e-05,
+      "loss": 1.0623,
+      "step": 14119
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.39675878387428e-05,
+      "loss": 0.9571,
+      "step": 14120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.395865886869951e-05,
+      "loss": 1.0065,
+      "step": 14121
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.39497298986562e-05,
+      "loss": 1.2055,
+      "step": 14122
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.394080092861289e-05,
+      "loss": 1.0739,
+      "step": 14123
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.393187195856959e-05,
+      "loss": 1.0685,
+      "step": 14124
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.392294298852628e-05,
+      "loss": 1.1421,
+      "step": 14125
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.391401401848297e-05,
+      "loss": 1.0601,
+      "step": 14126
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.390508504843968e-05,
+      "loss": 1.186,
+      "step": 14127
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.389615607839637e-05,
+      "loss": 1.2473,
+      "step": 14128
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.388722710835306e-05,
+      "loss": 1.0538,
+      "step": 14129
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.387829813830975e-05,
+      "loss": 0.9651,
+      "step": 14130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.386936916826644e-05,
+      "loss": 1.0137,
+      "step": 14131
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.386044019822313e-05,
+      "loss": 0.9359,
+      "step": 14132
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.385151122817983e-05,
+      "loss": 1.1462,
+      "step": 14133
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.384258225813652e-05,
+      "loss": 1.0551,
+      "step": 14134
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.383365328809321e-05,
+      "loss": 0.9372,
+      "step": 14135
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.382472431804992e-05,
+      "loss": 0.9803,
+      "step": 14136
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.381579534800661e-05,
+      "loss": 0.9759,
+      "step": 14137
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.38068663779633e-05,
+      "loss": 1.0883,
+      "step": 14138
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.379793740792e-05,
+      "loss": 0.9115,
+      "step": 14139
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.37890084378767e-05,
+      "loss": 1.0529,
+      "step": 14140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.378007946783338e-05,
+      "loss": 0.9459,
+      "step": 14141
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.377115049779009e-05,
+      "loss": 1.0131,
+      "step": 14142
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.376222152774678e-05,
+      "loss": 1.0224,
+      "step": 14143
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.375329255770347e-05,
+      "loss": 1.0219,
+      "step": 14144
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.374436358766017e-05,
+      "loss": 0.8836,
+      "step": 14145
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.373543461761686e-05,
+      "loss": 0.9883,
+      "step": 14146
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.372650564757355e-05,
+      "loss": 1.1752,
+      "step": 14147
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.371757667753026e-05,
+      "loss": 1.0159,
+      "step": 14148
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.370864770748695e-05,
+      "loss": 1.1469,
+      "step": 14149
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.369971873744364e-05,
+      "loss": 0.9255,
+      "step": 14150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.369078976740034e-05,
+      "loss": 0.9297,
+      "step": 14151
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.368186079735703e-05,
+      "loss": 1.0913,
+      "step": 14152
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.367293182731372e-05,
+      "loss": 1.1715,
+      "step": 14153
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.366400285727041e-05,
+      "loss": 0.9296,
+      "step": 14154
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.36550738872271e-05,
+      "loss": 1.0682,
+      "step": 14155
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.36461449171838e-05,
+      "loss": 1.0433,
+      "step": 14156
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.36372159471405e-05,
+      "loss": 1.0606,
+      "step": 14157
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.362828697709719e-05,
+      "loss": 1.0913,
+      "step": 14158
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.361935800705388e-05,
+      "loss": 0.9713,
+      "step": 14159
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.361042903701058e-05,
+      "loss": 1.0535,
+      "step": 14160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.360150006696728e-05,
+      "loss": 0.9534,
+      "step": 14161
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.359257109692397e-05,
+      "loss": 0.9304,
+      "step": 14162
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.358364212688067e-05,
+      "loss": 0.9896,
+      "step": 14163
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.357471315683736e-05,
+      "loss": 0.9395,
+      "step": 14164
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.356578418679406e-05,
+      "loss": 1.1093,
+      "step": 14165
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.355685521675075e-05,
+      "loss": 0.9484,
+      "step": 14166
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.354792624670745e-05,
+      "loss": 0.9994,
+      "step": 14167
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.353899727666415e-05,
+      "loss": 1.046,
+      "step": 14168
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.353006830662084e-05,
+      "loss": 1.2401,
+      "step": 14169
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.352113933657753e-05,
+      "loss": 1.0515,
+      "step": 14170
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.351221036653423e-05,
+      "loss": 1.1523,
+      "step": 14171
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.350328139649092e-05,
+      "loss": 1.022,
+      "step": 14172
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.349435242644762e-05,
+      "loss": 1.0005,
+      "step": 14173
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.34854234564043e-05,
+      "loss": 1.0168,
+      "step": 14174
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.3476494486361e-05,
+      "loss": 1.051,
+      "step": 14175
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.34675655163177e-05,
+      "loss": 1.0336,
+      "step": 14176
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.345863654627439e-05,
+      "loss": 0.9938,
+      "step": 14177
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.344970757623108e-05,
+      "loss": 0.9996,
+      "step": 14178
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.344077860618777e-05,
+      "loss": 0.9129,
+      "step": 14179
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.343184963614448e-05,
+      "loss": 1.0789,
+      "step": 14180
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.342292066610117e-05,
+      "loss": 0.9276,
+      "step": 14181
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.341399169605786e-05,
+      "loss": 0.9609,
+      "step": 14182
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.340506272601456e-05,
+      "loss": 1.0101,
+      "step": 14183
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.339613375597125e-05,
+      "loss": 1.1121,
+      "step": 14184
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.338720478592794e-05,
+      "loss": 1.067,
+      "step": 14185
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.337827581588465e-05,
+      "loss": 1.0499,
+      "step": 14186
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.336934684584134e-05,
+      "loss": 1.1139,
+      "step": 14187
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.336041787579803e-05,
+      "loss": 1.2733,
+      "step": 14188
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.335148890575473e-05,
+      "loss": 1.0408,
+      "step": 14189
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.334255993571142e-05,
+      "loss": 1.0201,
+      "step": 14190
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.333363096566811e-05,
+      "loss": 1.0721,
+      "step": 14191
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.332470199562482e-05,
+      "loss": 0.9772,
+      "step": 14192
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.33157730255815e-05,
+      "loss": 1.0481,
+      "step": 14193
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.33068440555382e-05,
+      "loss": 0.9286,
+      "step": 14194
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.32979150854949e-05,
+      "loss": 1.024,
+      "step": 14195
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.328898611545159e-05,
+      "loss": 1.0038,
+      "step": 14196
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.328005714540828e-05,
+      "loss": 0.9848,
+      "step": 14197
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.327112817536497e-05,
+      "loss": 0.9859,
+      "step": 14198
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.326219920532166e-05,
+      "loss": 0.9968,
+      "step": 14199
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.325327023527835e-05,
+      "loss": 1.0503,
+      "step": 14200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.324434126523506e-05,
+      "loss": 0.9803,
+      "step": 14201
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.323541229519175e-05,
+      "loss": 1.018,
+      "step": 14202
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.322648332514844e-05,
+      "loss": 0.9557,
+      "step": 14203
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.321755435510514e-05,
+      "loss": 1.027,
+      "step": 14204
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.320862538506183e-05,
+      "loss": 0.9795,
+      "step": 14205
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.319969641501852e-05,
+      "loss": 0.9564,
+      "step": 14206
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.319076744497523e-05,
+      "loss": 1.0986,
+      "step": 14207
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.318183847493192e-05,
+      "loss": 1.0812,
+      "step": 14208
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.317290950488861e-05,
+      "loss": 1.1795,
+      "step": 14209
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.316398053484531e-05,
+      "loss": 0.8406,
+      "step": 14210
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.3155051564802e-05,
+      "loss": 1.0968,
+      "step": 14211
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.314612259475869e-05,
+      "loss": 0.9897,
+      "step": 14212
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.31371936247154e-05,
+      "loss": 1.017,
+      "step": 14213
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.312826465467209e-05,
+      "loss": 1.0875,
+      "step": 14214
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.311933568462879e-05,
+      "loss": 1.019,
+      "step": 14215
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.311040671458548e-05,
+      "loss": 1.1402,
+      "step": 14216
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.310147774454217e-05,
+      "loss": 1.0394,
+      "step": 14217
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.309254877449888e-05,
+      "loss": 0.8938,
+      "step": 14218
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.308361980445557e-05,
+      "loss": 0.9724,
+      "step": 14219
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.307469083441226e-05,
+      "loss": 0.9869,
+      "step": 14220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.306576186436895e-05,
+      "loss": 1.0152,
+      "step": 14221
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.305683289432564e-05,
+      "loss": 1.0141,
+      "step": 14222
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.304790392428233e-05,
+      "loss": 1.1278,
+      "step": 14223
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.303897495423903e-05,
+      "loss": 1.0724,
+      "step": 14224
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.303004598419572e-05,
+      "loss": 1.0256,
+      "step": 14225
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.302111701415241e-05,
+      "loss": 1.0,
+      "step": 14226
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.301218804410912e-05,
+      "loss": 1.1593,
+      "step": 14227
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.300325907406581e-05,
+      "loss": 0.9299,
+      "step": 14228
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.29943301040225e-05,
+      "loss": 1.1674,
+      "step": 14229
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.29854011339792e-05,
+      "loss": 0.9817,
+      "step": 14230
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.29764721639359e-05,
+      "loss": 1.0887,
+      "step": 14231
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.296754319389258e-05,
+      "loss": 1.0005,
+      "step": 14232
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.295861422384929e-05,
+      "loss": 1.0426,
+      "step": 14233
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.294968525380598e-05,
+      "loss": 0.9286,
+      "step": 14234
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.294075628376267e-05,
+      "loss": 0.9828,
+      "step": 14235
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.293182731371937e-05,
+      "loss": 0.9872,
+      "step": 14236
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.292289834367606e-05,
+      "loss": 0.8825,
+      "step": 14237
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.291396937363275e-05,
+      "loss": 1.0769,
+      "step": 14238
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.290504040358946e-05,
+      "loss": 0.9079,
+      "step": 14239
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.289611143354615e-05,
+      "loss": 1.0205,
+      "step": 14240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.288718246350284e-05,
+      "loss": 1.0075,
+      "step": 14241
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.287825349345954e-05,
+      "loss": 0.8894,
+      "step": 14242
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.286932452341623e-05,
+      "loss": 0.9814,
+      "step": 14243
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.286039555337292e-05,
+      "loss": 1.0053,
+      "step": 14244
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.285146658332961e-05,
+      "loss": 0.9319,
+      "step": 14245
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.28425376132863e-05,
+      "loss": 0.9676,
+      "step": 14246
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.2833608643243e-05,
+      "loss": 1.0857,
+      "step": 14247
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.28246796731997e-05,
+      "loss": 1.0666,
+      "step": 14248
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.281575070315639e-05,
+      "loss": 1.0536,
+      "step": 14249
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.280682173311308e-05,
+      "loss": 0.9646,
+      "step": 14250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.279789276306978e-05,
+      "loss": 1.0192,
+      "step": 14251
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.278896379302647e-05,
+      "loss": 1.0064,
+      "step": 14252
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.278003482298317e-05,
+      "loss": 1.0317,
+      "step": 14253
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.277110585293987e-05,
+      "loss": 0.9626,
+      "step": 14254
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.276217688289656e-05,
+      "loss": 1.1106,
+      "step": 14255
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.275324791285325e-05,
+      "loss": 0.845,
+      "step": 14256
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.274431894280995e-05,
+      "loss": 1.0567,
+      "step": 14257
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.273538997276664e-05,
+      "loss": 0.9632,
+      "step": 14258
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.272646100272334e-05,
+      "loss": 1.1142,
+      "step": 14259
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.271753203268004e-05,
+      "loss": 0.9964,
+      "step": 14260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.270860306263673e-05,
+      "loss": 1.0769,
+      "step": 14261
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.269967409259343e-05,
+      "loss": 0.8648,
+      "step": 14262
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.269074512255012e-05,
+      "loss": 1.0218,
+      "step": 14263
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.268181615250681e-05,
+      "loss": 0.9577,
+      "step": 14264
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.26728871824635e-05,
+      "loss": 1.1199,
+      "step": 14265
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.26639582124202e-05,
+      "loss": 1.0762,
+      "step": 14266
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.26550292423769e-05,
+      "loss": 1.1214,
+      "step": 14267
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.264610027233359e-05,
+      "loss": 1.0259,
+      "step": 14268
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.263717130229028e-05,
+      "loss": 0.923,
+      "step": 14269
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.262824233224697e-05,
+      "loss": 1.0572,
+      "step": 14270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.261931336220368e-05,
+      "loss": 1.1209,
+      "step": 14271
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.261038439216037e-05,
+      "loss": 0.9975,
+      "step": 14272
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.260145542211706e-05,
+      "loss": 1.0676,
+      "step": 14273
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.259252645207376e-05,
+      "loss": 1.0021,
+      "step": 14274
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.258359748203045e-05,
+      "loss": 1.0459,
+      "step": 14275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.257466851198714e-05,
+      "loss": 1.111,
+      "step": 14276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.256573954194385e-05,
+      "loss": 0.9603,
+      "step": 14277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.255681057190054e-05,
+      "loss": 0.9879,
+      "step": 14278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.254788160185723e-05,
+      "loss": 0.8498,
+      "step": 14279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.253895263181393e-05,
+      "loss": 0.9879,
+      "step": 14280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.253002366177062e-05,
+      "loss": 1.1413,
+      "step": 14281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.252109469172731e-05,
+      "loss": 0.9598,
+      "step": 14282
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.251216572168402e-05,
+      "loss": 1.1162,
+      "step": 14283
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.25032367516407e-05,
+      "loss": 1.0348,
+      "step": 14284
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.24943077815974e-05,
+      "loss": 0.9712,
+      "step": 14285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.24853788115541e-05,
+      "loss": 0.9994,
+      "step": 14286
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.247644984151079e-05,
+      "loss": 1.1448,
+      "step": 14287
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.246752087146748e-05,
+      "loss": 1.0362,
+      "step": 14288
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.245859190142417e-05,
+      "loss": 0.9589,
+      "step": 14289
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.244966293138086e-05,
+      "loss": 1.0117,
+      "step": 14290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.244073396133755e-05,
+      "loss": 1.1474,
+      "step": 14291
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.243180499129426e-05,
+      "loss": 1.0985,
+      "step": 14292
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.242287602125095e-05,
+      "loss": 0.924,
+      "step": 14293
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.241394705120764e-05,
+      "loss": 1.1099,
+      "step": 14294
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.240501808116434e-05,
+      "loss": 1.0909,
+      "step": 14295
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.239608911112103e-05,
+      "loss": 1.1176,
+      "step": 14296
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.238716014107772e-05,
+      "loss": 1.019,
+      "step": 14297
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.237823117103443e-05,
+      "loss": 1.0927,
+      "step": 14298
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.236930220099112e-05,
+      "loss": 0.9635,
+      "step": 14299
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.236037323094781e-05,
+      "loss": 0.908,
+      "step": 14300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.235144426090451e-05,
+      "loss": 1.0301,
+      "step": 14301
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.23425152908612e-05,
+      "loss": 1.0115,
+      "step": 14302
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.233358632081789e-05,
+      "loss": 1.1269,
+      "step": 14303
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.23246573507746e-05,
+      "loss": 1.1409,
+      "step": 14304
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.231572838073129e-05,
+      "loss": 0.8296,
+      "step": 14305
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.230679941068798e-05,
+      "loss": 1.0055,
+      "step": 14306
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.229787044064468e-05,
+      "loss": 1.0507,
+      "step": 14307
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.228894147060137e-05,
+      "loss": 0.9806,
+      "step": 14308
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.228001250055806e-05,
+      "loss": 1.1339,
+      "step": 14309
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.227108353051477e-05,
+      "loss": 1.1025,
+      "step": 14310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.226215456047146e-05,
+      "loss": 0.9861,
+      "step": 14311
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.225322559042815e-05,
+      "loss": 1.0382,
+      "step": 14312
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.224429662038484e-05,
+      "loss": 1.1105,
+      "step": 14313
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.223536765034153e-05,
+      "loss": 1.0781,
+      "step": 14314
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.222643868029823e-05,
+      "loss": 0.9997,
+      "step": 14315
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.221750971025492e-05,
+      "loss": 0.9571,
+      "step": 14316
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.220858074021161e-05,
+      "loss": 0.9588,
+      "step": 14317
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.219965177016832e-05,
+      "loss": 1.1433,
+      "step": 14318
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.219072280012501e-05,
+      "loss": 0.9185,
+      "step": 14319
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.21817938300817e-05,
+      "loss": 1.0402,
+      "step": 14320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.21728648600384e-05,
+      "loss": 1.0251,
+      "step": 14321
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.216393588999509e-05,
+      "loss": 0.9886,
+      "step": 14322
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.215500691995178e-05,
+      "loss": 1.0215,
+      "step": 14323
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.214607794990849e-05,
+      "loss": 1.0788,
+      "step": 14324
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.213714897986518e-05,
+      "loss": 1.1506,
+      "step": 14325
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.212822000982187e-05,
+      "loss": 0.9807,
+      "step": 14326
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.211929103977857e-05,
+      "loss": 1.0153,
+      "step": 14327
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.211036206973526e-05,
+      "loss": 1.0847,
+      "step": 14328
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.210143309969195e-05,
+      "loss": 1.0362,
+      "step": 14329
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.209250412964866e-05,
+      "loss": 0.922,
+      "step": 14330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.208357515960535e-05,
+      "loss": 0.9915,
+      "step": 14331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.207464618956204e-05,
+      "loss": 1.1564,
+      "step": 14332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.206571721951874e-05,
+      "loss": 0.9216,
+      "step": 14333
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.205678824947543e-05,
+      "loss": 0.8332,
+      "step": 14334
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.204785927943212e-05,
+      "loss": 1.1143,
+      "step": 14335
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.203893030938881e-05,
+      "loss": 1.1371,
+      "step": 14336
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.20300013393455e-05,
+      "loss": 0.8711,
+      "step": 14337
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.20210723693022e-05,
+      "loss": 1.0811,
+      "step": 14338
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.20121433992589e-05,
+      "loss": 0.9771,
+      "step": 14339
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.200321442921559e-05,
+      "loss": 0.8781,
+      "step": 14340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.199428545917228e-05,
+      "loss": 1.0041,
+      "step": 14341
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.198535648912898e-05,
+      "loss": 1.127,
+      "step": 14342
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.197642751908567e-05,
+      "loss": 1.0211,
+      "step": 14343
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.196749854904236e-05,
+      "loss": 0.8641,
+      "step": 14344
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.195856957899907e-05,
+      "loss": 0.9862,
+      "step": 14345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.194964060895576e-05,
+      "loss": 1.0353,
+      "step": 14346
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.194071163891245e-05,
+      "loss": 1.0111,
+      "step": 14347
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.193178266886915e-05,
+      "loss": 0.998,
+      "step": 14348
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.192285369882584e-05,
+      "loss": 1.0098,
+      "step": 14349
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.191392472878253e-05,
+      "loss": 1.0164,
+      "step": 14350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.190499575873924e-05,
+      "loss": 0.9728,
+      "step": 14351
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.189606678869593e-05,
+      "loss": 1.184,
+      "step": 14352
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.188713781865262e-05,
+      "loss": 1.4042,
+      "step": 14353
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.187820884860932e-05,
+      "loss": 0.9866,
+      "step": 14354
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.186927987856601e-05,
+      "loss": 1.124,
+      "step": 14355
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.18603509085227e-05,
+      "loss": 1.0348,
+      "step": 14356
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.18514219384794e-05,
+      "loss": 1.1039,
+      "step": 14357
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.184249296843609e-05,
+      "loss": 1.0413,
+      "step": 14358
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.183356399839279e-05,
+      "loss": 1.0061,
+      "step": 14359
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.182463502834948e-05,
+      "loss": 1.0267,
+      "step": 14360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.181570605830617e-05,
+      "loss": 0.9906,
+      "step": 14361
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.180677708826287e-05,
+      "loss": 1.179,
+      "step": 14362
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.179784811821957e-05,
+      "loss": 1.0233,
+      "step": 14363
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.178891914817626e-05,
+      "loss": 1.0118,
+      "step": 14364
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.177999017813296e-05,
+      "loss": 0.9449,
+      "step": 14365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.177106120808965e-05,
+      "loss": 0.8771,
+      "step": 14366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.176213223804634e-05,
+      "loss": 1.0132,
+      "step": 14367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.175320326800304e-05,
+      "loss": 1.0325,
+      "step": 14368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.174427429795974e-05,
+      "loss": 1.0086,
+      "step": 14369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.173534532791643e-05,
+      "loss": 1.0375,
+      "step": 14370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.172641635787313e-05,
+      "loss": 1.0563,
+      "step": 14371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.171748738782982e-05,
+      "loss": 1.296,
+      "step": 14372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.170855841778651e-05,
+      "loss": 1.122,
+      "step": 14373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.169962944774321e-05,
+      "loss": 0.9893,
+      "step": 14374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.16907004776999e-05,
+      "loss": 1.1078,
+      "step": 14375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.16817715076566e-05,
+      "loss": 1.0693,
+      "step": 14376
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.16728425376133e-05,
+      "loss": 0.8975,
+      "step": 14377
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.166391356756999e-05,
+      "loss": 1.0239,
+      "step": 14378
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.165498459752668e-05,
+      "loss": 1.0334,
+      "step": 14379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.164605562748337e-05,
+      "loss": 1.0134,
+      "step": 14380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.163712665744006e-05,
+      "loss": 1.0347,
+      "step": 14381
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.162819768739675e-05,
+      "loss": 1.0499,
+      "step": 14382
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.161926871735346e-05,
+      "loss": 1.0161,
+      "step": 14383
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.161033974731015e-05,
+      "loss": 0.9102,
+      "step": 14384
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.160141077726684e-05,
+      "loss": 1.1165,
+      "step": 14385
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.159248180722354e-05,
+      "loss": 0.9606,
+      "step": 14386
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.158355283718023e-05,
+      "loss": 1.1076,
+      "step": 14387
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.157462386713692e-05,
+      "loss": 0.9472,
+      "step": 14388
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.156569489709363e-05,
+      "loss": 0.9634,
+      "step": 14389
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.155676592705032e-05,
+      "loss": 1.1006,
+      "step": 14390
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.154783695700701e-05,
+      "loss": 1.0373,
+      "step": 14391
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.153890798696371e-05,
+      "loss": 1.0229,
+      "step": 14392
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.15299790169204e-05,
+      "loss": 1.0501,
+      "step": 14393
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.152105004687709e-05,
+      "loss": 0.921,
+      "step": 14394
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.15121210768338e-05,
+      "loss": 0.9707,
+      "step": 14395
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.150319210679049e-05,
+      "loss": 1.0328,
+      "step": 14396
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.149426313674718e-05,
+      "loss": 1.0229,
+      "step": 14397
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.148533416670388e-05,
+      "loss": 1.0447,
+      "step": 14398
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.147640519666057e-05,
+      "loss": 1.1334,
+      "step": 14399
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.146747622661726e-05,
+      "loss": 1.3568,
+      "step": 14400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.145854725657397e-05,
+      "loss": 1.237,
+      "step": 14401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.144961828653066e-05,
+      "loss": 1.0804,
+      "step": 14402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.144068931648735e-05,
+      "loss": 1.008,
+      "step": 14403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.143176034644404e-05,
+      "loss": 1.1894,
+      "step": 14404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.142283137640073e-05,
+      "loss": 1.1245,
+      "step": 14405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.141390240635742e-05,
+      "loss": 0.9502,
+      "step": 14406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.140497343631412e-05,
+      "loss": 1.0595,
+      "step": 14407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.139604446627081e-05,
+      "loss": 0.969,
+      "step": 14408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.138711549622752e-05,
+      "loss": 0.9682,
+      "step": 14409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.137818652618421e-05,
+      "loss": 0.9542,
+      "step": 14410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.13692575561409e-05,
+      "loss": 0.9435,
+      "step": 14411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.13603285860976e-05,
+      "loss": 1.1081,
+      "step": 14412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.135139961605429e-05,
+      "loss": 1.0002,
+      "step": 14413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.134247064601098e-05,
+      "loss": 0.9848,
+      "step": 14414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.133354167596769e-05,
+      "loss": 1.0717,
+      "step": 14415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.132461270592438e-05,
+      "loss": 1.0992,
+      "step": 14416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.131568373588107e-05,
+      "loss": 0.9171,
+      "step": 14417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.130675476583777e-05,
+      "loss": 0.9493,
+      "step": 14418
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.129782579579446e-05,
+      "loss": 1.0341,
+      "step": 14419
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.128889682575115e-05,
+      "loss": 1.0251,
+      "step": 14420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.127996785570786e-05,
+      "loss": 1.0018,
+      "step": 14421
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.127103888566455e-05,
+      "loss": 1.0036,
+      "step": 14422
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.126210991562124e-05,
+      "loss": 1.0583,
+      "step": 14423
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.125318094557793e-05,
+      "loss": 0.9931,
+      "step": 14424
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.124425197553463e-05,
+      "loss": 1.1268,
+      "step": 14425
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.123532300549132e-05,
+      "loss": 1.0551,
+      "step": 14426
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.122639403544801e-05,
+      "loss": 0.9867,
+      "step": 14427
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.12174650654047e-05,
+      "loss": 1.0073,
+      "step": 14428
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.12085360953614e-05,
+      "loss": 0.9731,
+      "step": 14429
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.11996071253181e-05,
+      "loss": 0.9753,
+      "step": 14430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.119067815527479e-05,
+      "loss": 0.9916,
+      "step": 14431
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.118174918523148e-05,
+      "loss": 1.2311,
+      "step": 14432
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.117282021518818e-05,
+      "loss": 0.9476,
+      "step": 14433
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.116389124514487e-05,
+      "loss": 1.0932,
+      "step": 14434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.115496227510156e-05,
+      "loss": 1.0799,
+      "step": 14435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.114603330505827e-05,
+      "loss": 1.0326,
+      "step": 14436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.113710433501496e-05,
+      "loss": 1.1053,
+      "step": 14437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.112817536497165e-05,
+      "loss": 1.1043,
+      "step": 14438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.111924639492835e-05,
+      "loss": 1.0136,
+      "step": 14439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.111031742488504e-05,
+      "loss": 1.0482,
+      "step": 14440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.110138845484173e-05,
+      "loss": 0.9785,
+      "step": 14441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.109245948479844e-05,
+      "loss": 0.9663,
+      "step": 14442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.108353051475513e-05,
+      "loss": 1.1251,
+      "step": 14443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.107460154471182e-05,
+      "loss": 0.9389,
+      "step": 14444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.106567257466852e-05,
+      "loss": 0.9399,
+      "step": 14445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.105674360462521e-05,
+      "loss": 1.0252,
+      "step": 14446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.10478146345819e-05,
+      "loss": 1.0054,
+      "step": 14447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.10388856645386e-05,
+      "loss": 1.0377,
+      "step": 14448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.102995669449529e-05,
+      "loss": 1.133,
+      "step": 14449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.102102772445199e-05,
+      "loss": 1.0088,
+      "step": 14450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.101209875440868e-05,
+      "loss": 0.9397,
+      "step": 14451
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.100316978436537e-05,
+      "loss": 1.0672,
+      "step": 14452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.099424081432206e-05,
+      "loss": 1.0085,
+      "step": 14453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.098531184427877e-05,
+      "loss": 1.0345,
+      "step": 14454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.097638287423546e-05,
+      "loss": 0.9653,
+      "step": 14455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.096745390419215e-05,
+      "loss": 0.9715,
+      "step": 14456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.095852493414885e-05,
+      "loss": 0.9886,
+      "step": 14457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.094959596410554e-05,
+      "loss": 1.0384,
+      "step": 14458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.094066699406224e-05,
+      "loss": 1.0053,
+      "step": 14459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.093173802401893e-05,
+      "loss": 0.889,
+      "step": 14460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.092280905397563e-05,
+      "loss": 0.9893,
+      "step": 14461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.091388008393233e-05,
+      "loss": 0.9665,
+      "step": 14462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.090495111388902e-05,
+      "loss": 1.0859,
+      "step": 14463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.089602214384571e-05,
+      "loss": 1.1404,
+      "step": 14464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.088709317380241e-05,
+      "loss": 1.1843,
+      "step": 14465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.08781642037591e-05,
+      "loss": 0.9323,
+      "step": 14466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.08692352337158e-05,
+      "loss": 0.9716,
+      "step": 14467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.08603062636725e-05,
+      "loss": 0.8927,
+      "step": 14468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.085137729362919e-05,
+      "loss": 1.0518,
+      "step": 14469
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.084244832358588e-05,
+      "loss": 0.9226,
+      "step": 14470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.083351935354257e-05,
+      "loss": 1.1431,
+      "step": 14471
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.082459038349926e-05,
+      "loss": 1.0264,
+      "step": 14472
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.081566141345595e-05,
+      "loss": 1.0107,
+      "step": 14473
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.080673244341266e-05,
+      "loss": 1.0361,
+      "step": 14474
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.079780347336935e-05,
+      "loss": 1.0948,
+      "step": 14475
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.078887450332604e-05,
+      "loss": 1.0799,
+      "step": 14476
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.077994553328274e-05,
+      "loss": 1.0619,
+      "step": 14477
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.077101656323943e-05,
+      "loss": 1.0491,
+      "step": 14478
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.076208759319612e-05,
+      "loss": 1.1048,
+      "step": 14479
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.075315862315283e-05,
+      "loss": 1.0889,
+      "step": 14480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.074422965310952e-05,
+      "loss": 0.9888,
+      "step": 14481
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.073530068306621e-05,
+      "loss": 0.9919,
+      "step": 14482
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.072637171302291e-05,
+      "loss": 1.2432,
+      "step": 14483
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.07174427429796e-05,
+      "loss": 0.9537,
+      "step": 14484
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.070851377293629e-05,
+      "loss": 1.0691,
+      "step": 14485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.0699584802893e-05,
+      "loss": 1.0248,
+      "step": 14486
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.069065583284969e-05,
+      "loss": 1.1303,
+      "step": 14487
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.068172686280638e-05,
+      "loss": 1.0031,
+      "step": 14488
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.067279789276308e-05,
+      "loss": 1.0184,
+      "step": 14489
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.066386892271977e-05,
+      "loss": 0.993,
+      "step": 14490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.065493995267646e-05,
+      "loss": 1.1418,
+      "step": 14491
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.064601098263317e-05,
+      "loss": 1.0577,
+      "step": 14492
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.063708201258986e-05,
+      "loss": 1.1373,
+      "step": 14493
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.062815304254655e-05,
+      "loss": 0.9954,
+      "step": 14494
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.061922407250324e-05,
+      "loss": 1.0571,
+      "step": 14495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.061029510245993e-05,
+      "loss": 1.0258,
+      "step": 14496
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.060136613241662e-05,
+      "loss": 0.9904,
+      "step": 14497
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.059243716237332e-05,
+      "loss": 0.952,
+      "step": 14498
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.058350819233001e-05,
+      "loss": 1.0045,
+      "step": 14499
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.05745792222867e-05,
+      "loss": 1.0437,
+      "step": 14500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.056565025224341e-05,
+      "loss": 1.1853,
+      "step": 14501
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.05567212822001e-05,
+      "loss": 1.0064,
+      "step": 14502
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.054779231215679e-05,
+      "loss": 1.1308,
+      "step": 14503
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.053886334211349e-05,
+      "loss": 0.9534,
+      "step": 14504
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.052993437207018e-05,
+      "loss": 0.9755,
+      "step": 14505
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.052100540202689e-05,
+      "loss": 1.0076,
+      "step": 14506
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.051207643198358e-05,
+      "loss": 1.0241,
+      "step": 14507
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.050314746194027e-05,
+      "loss": 1.0345,
+      "step": 14508
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.049421849189697e-05,
+      "loss": 0.9897,
+      "step": 14509
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.048528952185366e-05,
+      "loss": 1.0189,
+      "step": 14510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.047636055181035e-05,
+      "loss": 1.0625,
+      "step": 14511
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.046743158176706e-05,
+      "loss": 1.0171,
+      "step": 14512
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.045850261172375e-05,
+      "loss": 1.0383,
+      "step": 14513
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.044957364168044e-05,
+      "loss": 1.0466,
+      "step": 14514
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.044064467163713e-05,
+      "loss": 1.1846,
+      "step": 14515
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.043171570159383e-05,
+      "loss": 0.9322,
+      "step": 14516
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.042278673155052e-05,
+      "loss": 0.9659,
+      "step": 14517
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.041385776150721e-05,
+      "loss": 1.0351,
+      "step": 14518
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.04049287914639e-05,
+      "loss": 0.9328,
+      "step": 14519
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.03959998214206e-05,
+      "loss": 1.0185,
+      "step": 14520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.03870708513773e-05,
+      "loss": 0.9721,
+      "step": 14521
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.037814188133399e-05,
+      "loss": 1.0193,
+      "step": 14522
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.036921291129068e-05,
+      "loss": 1.0373,
+      "step": 14523
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.036028394124738e-05,
+      "loss": 1.077,
+      "step": 14524
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.035135497120407e-05,
+      "loss": 1.0355,
+      "step": 14525
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.034242600116076e-05,
+      "loss": 1.1415,
+      "step": 14526
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.033349703111747e-05,
+      "loss": 1.0199,
+      "step": 14527
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.032456806107416e-05,
+      "loss": 0.9954,
+      "step": 14528
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.031563909103085e-05,
+      "loss": 0.9306,
+      "step": 14529
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.030671012098755e-05,
+      "loss": 1.014,
+      "step": 14530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.029778115094424e-05,
+      "loss": 1.0202,
+      "step": 14531
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.028885218090093e-05,
+      "loss": 1.01,
+      "step": 14532
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.027992321085764e-05,
+      "loss": 1.1029,
+      "step": 14533
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.027099424081433e-05,
+      "loss": 1.0024,
+      "step": 14534
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.026206527077102e-05,
+      "loss": 1.0544,
+      "step": 14535
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.025313630072772e-05,
+      "loss": 1.0993,
+      "step": 14536
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.024420733068441e-05,
+      "loss": 1.0569,
+      "step": 14537
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.02352783606411e-05,
+      "loss": 1.0166,
+      "step": 14538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.02263493905978e-05,
+      "loss": 1.0576,
+      "step": 14539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.021742042055449e-05,
+      "loss": 0.9862,
+      "step": 14540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.020849145051118e-05,
+      "loss": 1.0063,
+      "step": 14541
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.019956248046788e-05,
+      "loss": 1.0438,
+      "step": 14542
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.019063351042457e-05,
+      "loss": 1.0216,
+      "step": 14543
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018170454038126e-05,
+      "loss": 1.1443,
+      "step": 14544
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.017277557033796e-05,
+      "loss": 0.9003,
+      "step": 14545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.016384660029466e-05,
+      "loss": 1.0286,
+      "step": 14546
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.015491763025135e-05,
+      "loss": 1.0773,
+      "step": 14547
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.014598866020805e-05,
+      "loss": 1.0422,
+      "step": 14548
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.013705969016474e-05,
+      "loss": 0.9578,
+      "step": 14549
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.012813072012143e-05,
+      "loss": 1.0429,
+      "step": 14550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.011920175007813e-05,
+      "loss": 1.1164,
+      "step": 14551
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.011027278003483e-05,
+      "loss": 1.0586,
+      "step": 14552
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.010134380999152e-05,
+      "loss": 0.9599,
+      "step": 14553
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.009241483994822e-05,
+      "loss": 1.0848,
+      "step": 14554
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.008348586990491e-05,
+      "loss": 1.1215,
+      "step": 14555
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.007455689986161e-05,
+      "loss": 1.0107,
+      "step": 14556
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.00656279298183e-05,
+      "loss": 1.2787,
+      "step": 14557
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.0056698959775e-05,
+      "loss": 1.1418,
+      "step": 14558
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.00477699897317e-05,
+      "loss": 0.8966,
+      "step": 14559
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.003884101968839e-05,
+      "loss": 0.9574,
+      "step": 14560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.002991204964508e-05,
+      "loss": 0.965,
+      "step": 14561
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.002098307960177e-05,
+      "loss": 1.0903,
+      "step": 14562
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.001205410955846e-05,
+      "loss": 1.1291,
+      "step": 14563
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.000312513951515e-05,
+      "loss": 1.2723,
+      "step": 14564
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.999419616947186e-05,
+      "loss": 0.9563,
+      "step": 14565
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.998526719942855e-05,
+      "loss": 0.9538,
+      "step": 14566
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.997633822938524e-05,
+      "loss": 1.0558,
+      "step": 14567
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.996740925934194e-05,
+      "loss": 1.2654,
+      "step": 14568
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.995848028929863e-05,
+      "loss": 1.1012,
+      "step": 14569
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.994955131925532e-05,
+      "loss": 1.1343,
+      "step": 14570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.994062234921203e-05,
+      "loss": 1.0825,
+      "step": 14571
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.993169337916872e-05,
+      "loss": 1.0701,
+      "step": 14572
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.99227644091254e-05,
+      "loss": 1.1396,
+      "step": 14573
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.991383543908211e-05,
+      "loss": 1.1653,
+      "step": 14574
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.99049064690388e-05,
+      "loss": 1.0275,
+      "step": 14575
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.989597749899549e-05,
+      "loss": 1.1018,
+      "step": 14576
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.98870485289522e-05,
+      "loss": 1.0034,
+      "step": 14577
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.987811955890889e-05,
+      "loss": 0.9797,
+      "step": 14578
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.986919058886558e-05,
+      "loss": 0.9141,
+      "step": 14579
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.986026161882228e-05,
+      "loss": 1.0081,
+      "step": 14580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.985133264877897e-05,
+      "loss": 1.0612,
+      "step": 14581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.984240367873566e-05,
+      "loss": 0.9424,
+      "step": 14582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.983347470869237e-05,
+      "loss": 1.0373,
+      "step": 14583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.982454573864906e-05,
+      "loss": 1.0203,
+      "step": 14584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.981561676860575e-05,
+      "loss": 0.9926,
+      "step": 14585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.980668779856244e-05,
+      "loss": 1.0841,
+      "step": 14586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.979775882851913e-05,
+      "loss": 0.8652,
+      "step": 14587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.978882985847582e-05,
+      "loss": 0.9533,
+      "step": 14588
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.977990088843252e-05,
+      "loss": 0.9805,
+      "step": 14589
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.977097191838921e-05,
+      "loss": 0.9691,
+      "step": 14590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.97620429483459e-05,
+      "loss": 0.9392,
+      "step": 14591
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.975311397830261e-05,
+      "loss": 1.0738,
+      "step": 14592
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.97441850082593e-05,
+      "loss": 1.0001,
+      "step": 14593
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.973525603821599e-05,
+      "loss": 1.0042,
+      "step": 14594
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.972632706817269e-05,
+      "loss": 1.0547,
+      "step": 14595
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.971739809812938e-05,
+      "loss": 1.042,
+      "step": 14596
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.970846912808607e-05,
+      "loss": 1.0314,
+      "step": 14597
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.969954015804278e-05,
+      "loss": 1.1249,
+      "step": 14598
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.969061118799947e-05,
+      "loss": 1.0115,
+      "step": 14599
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.968168221795616e-05,
+      "loss": 1.0314,
+      "step": 14600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.967275324791286e-05,
+      "loss": 1.0695,
+      "step": 14601
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.966382427786955e-05,
+      "loss": 0.9909,
+      "step": 14602
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.965489530782624e-05,
+      "loss": 1.0842,
+      "step": 14603
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.964596633778295e-05,
+      "loss": 1.0146,
+      "step": 14604
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.963703736773964e-05,
+      "loss": 0.9996,
+      "step": 14605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.962810839769633e-05,
+      "loss": 1.026,
+      "step": 14606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.961917942765303e-05,
+      "loss": 0.9214,
+      "step": 14607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.961025045760972e-05,
+      "loss": 1.1516,
+      "step": 14608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.960132148756641e-05,
+      "loss": 0.9511,
+      "step": 14609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.95923925175231e-05,
+      "loss": 0.9982,
+      "step": 14610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.95834635474798e-05,
+      "loss": 1.0622,
+      "step": 14611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.95745345774365e-05,
+      "loss": 1.0028,
+      "step": 14612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.956560560739319e-05,
+      "loss": 1.0958,
+      "step": 14613
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.955667663734988e-05,
+      "loss": 1.0158,
+      "step": 14614
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.954774766730658e-05,
+      "loss": 1.0472,
+      "step": 14615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.953881869726327e-05,
+      "loss": 1.0885,
+      "step": 14616
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.952988972721996e-05,
+      "loss": 0.9511,
+      "step": 14617
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.952096075717667e-05,
+      "loss": 0.9023,
+      "step": 14618
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.951203178713336e-05,
+      "loss": 0.9922,
+      "step": 14619
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.950310281709005e-05,
+      "loss": 1.0182,
+      "step": 14620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.949417384704675e-05,
+      "loss": 1.0958,
+      "step": 14621
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.948524487700344e-05,
+      "loss": 1.0121,
+      "step": 14622
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.947631590696013e-05,
+      "loss": 0.9976,
+      "step": 14623
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.946738693691684e-05,
+      "loss": 0.9677,
+      "step": 14624
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.945845796687353e-05,
+      "loss": 1.0581,
+      "step": 14625
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.944952899683022e-05,
+      "loss": 1.0549,
+      "step": 14626
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.944060002678692e-05,
+      "loss": 0.8833,
+      "step": 14627
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.943167105674361e-05,
+      "loss": 1.0046,
+      "step": 14628
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.94227420867003e-05,
+      "loss": 0.9918,
+      "step": 14629
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.9413813116657e-05,
+      "loss": 1.0059,
+      "step": 14630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.940488414661368e-05,
+      "loss": 0.9748,
+      "step": 14631
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.939595517657038e-05,
+      "loss": 1.116,
+      "step": 14632
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.938702620652708e-05,
+      "loss": 0.9071,
+      "step": 14633
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.937809723648377e-05,
+      "loss": 1.0546,
+      "step": 14634
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.936916826644046e-05,
+      "loss": 1.0162,
+      "step": 14635
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.936023929639716e-05,
+      "loss": 0.9131,
+      "step": 14636
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.935131032635385e-05,
+      "loss": 1.0241,
+      "step": 14637
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.934238135631055e-05,
+      "loss": 1.0187,
+      "step": 14638
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.933345238626725e-05,
+      "loss": 0.9954,
+      "step": 14639
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.932452341622394e-05,
+      "loss": 0.9813,
+      "step": 14640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.931559444618063e-05,
+      "loss": 0.9948,
+      "step": 14641
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.930666547613733e-05,
+      "loss": 1.005,
+      "step": 14642
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.929773650609402e-05,
+      "loss": 1.0396,
+      "step": 14643
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.928880753605072e-05,
+      "loss": 0.9837,
+      "step": 14644
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.927987856600742e-05,
+      "loss": 1.0429,
+      "step": 14645
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.927094959596411e-05,
+      "loss": 0.9111,
+      "step": 14646
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.92620206259208e-05,
+      "loss": 1.1401,
+      "step": 14647
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.92530916558775e-05,
+      "loss": 0.9975,
+      "step": 14648
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.92441626858342e-05,
+      "loss": 1.0359,
+      "step": 14649
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.923523371579089e-05,
+      "loss": 1.1137,
+      "step": 14650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.922630474574759e-05,
+      "loss": 0.928,
+      "step": 14651
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.921737577570428e-05,
+      "loss": 1.0458,
+      "step": 14652
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.920844680566097e-05,
+      "loss": 1.0268,
+      "step": 14653
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.919951783561766e-05,
+      "loss": 1.0645,
+      "step": 14654
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.919058886557435e-05,
+      "loss": 1.1102,
+      "step": 14655
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.918165989553106e-05,
+      "loss": 0.9602,
+      "step": 14656
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.917273092548775e-05,
+      "loss": 1.0343,
+      "step": 14657
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.916380195544444e-05,
+      "loss": 1.0642,
+      "step": 14658
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.915487298540114e-05,
+      "loss": 1.1424,
+      "step": 14659
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.914594401535783e-05,
+      "loss": 0.945,
+      "step": 14660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.913701504531452e-05,
+      "loss": 1.1419,
+      "step": 14661
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.912808607527123e-05,
+      "loss": 1.0378,
+      "step": 14662
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.911915710522792e-05,
+      "loss": 1.1043,
+      "step": 14663
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.91102281351846e-05,
+      "loss": 1.0063,
+      "step": 14664
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.910129916514131e-05,
+      "loss": 1.047,
+      "step": 14665
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.9092370195098e-05,
+      "loss": 1.0225,
+      "step": 14666
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.908344122505469e-05,
+      "loss": 1.1696,
+      "step": 14667
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.90745122550114e-05,
+      "loss": 1.1064,
+      "step": 14668
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.906558328496809e-05,
+      "loss": 1.0519,
+      "step": 14669
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.905665431492478e-05,
+      "loss": 1.0373,
+      "step": 14670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.904772534488148e-05,
+      "loss": 1.1227,
+      "step": 14671
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.903879637483817e-05,
+      "loss": 1.0205,
+      "step": 14672
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.902986740479486e-05,
+      "loss": 1.0392,
+      "step": 14673
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.902093843475157e-05,
+      "loss": 1.0067,
+      "step": 14674
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.901200946470826e-05,
+      "loss": 0.9383,
+      "step": 14675
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.900308049466495e-05,
+      "loss": 1.0436,
+      "step": 14676
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.899415152462164e-05,
+      "loss": 0.9681,
+      "step": 14677
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.898522255457833e-05,
+      "loss": 0.9531,
+      "step": 14678
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.897629358453502e-05,
+      "loss": 1.0839,
+      "step": 14679
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.896736461449172e-05,
+      "loss": 1.1332,
+      "step": 14680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.895843564444841e-05,
+      "loss": 1.0025,
+      "step": 14681
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.89495066744051e-05,
+      "loss": 0.9571,
+      "step": 14682
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.89405777043618e-05,
+      "loss": 1.0333,
+      "step": 14683
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.89316487343185e-05,
+      "loss": 1.1728,
+      "step": 14684
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.892271976427519e-05,
+      "loss": 0.9447,
+      "step": 14685
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.891379079423189e-05,
+      "loss": 0.9266,
+      "step": 14686
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.890486182418858e-05,
+      "loss": 1.1589,
+      "step": 14687
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.889593285414527e-05,
+      "loss": 1.0965,
+      "step": 14688
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.888700388410198e-05,
+      "loss": 0.9936,
+      "step": 14689
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.887807491405867e-05,
+      "loss": 1.0262,
+      "step": 14690
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.886914594401536e-05,
+      "loss": 1.1061,
+      "step": 14691
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.886021697397206e-05,
+      "loss": 1.0242,
+      "step": 14692
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.885128800392875e-05,
+      "loss": 0.998,
+      "step": 14693
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.884235903388544e-05,
+      "loss": 0.9214,
+      "step": 14694
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.883343006384215e-05,
+      "loss": 1.0021,
+      "step": 14695
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.882450109379884e-05,
+      "loss": 1.1028,
+      "step": 14696
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.881557212375553e-05,
+      "loss": 0.8199,
+      "step": 14697
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.880664315371222e-05,
+      "loss": 1.0289,
+      "step": 14698
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.879771418366892e-05,
+      "loss": 1.1144,
+      "step": 14699
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.878878521362561e-05,
+      "loss": 1.0674,
+      "step": 14700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.87798562435823e-05,
+      "loss": 1.1408,
+      "step": 14701
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.8770927273539e-05,
+      "loss": 1.0622,
+      "step": 14702
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.87619983034957e-05,
+      "loss": 0.9378,
+      "step": 14703
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.875306933345239e-05,
+      "loss": 0.9886,
+      "step": 14704
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.874414036340908e-05,
+      "loss": 0.9595,
+      "step": 14705
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.873521139336578e-05,
+      "loss": 1.0384,
+      "step": 14706
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.872628242332247e-05,
+      "loss": 0.8929,
+      "step": 14707
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.871735345327916e-05,
+      "loss": 0.9925,
+      "step": 14708
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.870842448323587e-05,
+      "loss": 1.0054,
+      "step": 14709
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.869949551319256e-05,
+      "loss": 1.0031,
+      "step": 14710
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.869056654314925e-05,
+      "loss": 1.0515,
+      "step": 14711
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.868163757310595e-05,
+      "loss": 1.0538,
+      "step": 14712
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.867270860306264e-05,
+      "loss": 0.9388,
+      "step": 14713
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.866377963301933e-05,
+      "loss": 0.992,
+      "step": 14714
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.865485066297604e-05,
+      "loss": 0.9747,
+      "step": 14715
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.864592169293273e-05,
+      "loss": 1.0706,
+      "step": 14716
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.863699272288942e-05,
+      "loss": 1.0923,
+      "step": 14717
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.862806375284612e-05,
+      "loss": 0.941,
+      "step": 14718
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.861913478280281e-05,
+      "loss": 0.9854,
+      "step": 14719
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.86102058127595e-05,
+      "loss": 0.9767,
+      "step": 14720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.86012768427162e-05,
+      "loss": 1.0117,
+      "step": 14721
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.859234787267288e-05,
+      "loss": 1.1413,
+      "step": 14722
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.858341890262957e-05,
+      "loss": 0.9853,
+      "step": 14723
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.857448993258628e-05,
+      "loss": 1.0852,
+      "step": 14724
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.856556096254297e-05,
+      "loss": 1.1268,
+      "step": 14725
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.855663199249966e-05,
+      "loss": 1.0558,
+      "step": 14726
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.854770302245636e-05,
+      "loss": 0.9427,
+      "step": 14727
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.853877405241305e-05,
+      "loss": 1.1814,
+      "step": 14728
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.852984508236974e-05,
+      "loss": 1.0077,
+      "step": 14729
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.852091611232645e-05,
+      "loss": 1.1725,
+      "step": 14730
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.851198714228314e-05,
+      "loss": 0.9303,
+      "step": 14731
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.850305817223983e-05,
+      "loss": 0.9598,
+      "step": 14732
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.849412920219653e-05,
+      "loss": 1.2396,
+      "step": 14733
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.848520023215322e-05,
+      "loss": 1.0657,
+      "step": 14734
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.847627126210991e-05,
+      "loss": 1.014,
+      "step": 14735
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.846734229206662e-05,
+      "loss": 1.1619,
+      "step": 14736
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.845841332202331e-05,
+      "loss": 1.0279,
+      "step": 14737
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.844948435198e-05,
+      "loss": 1.0189,
+      "step": 14738
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.84405553819367e-05,
+      "loss": 1.0978,
+      "step": 14739
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.84316264118934e-05,
+      "loss": 1.0528,
+      "step": 14740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.842269744185008e-05,
+      "loss": 1.1594,
+      "step": 14741
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.841376847180679e-05,
+      "loss": 1.0469,
+      "step": 14742
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.840483950176348e-05,
+      "loss": 1.1056,
+      "step": 14743
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.839591053172017e-05,
+      "loss": 0.9584,
+      "step": 14744
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.838698156167686e-05,
+      "loss": 1.1271,
+      "step": 14745
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.837805259163355e-05,
+      "loss": 0.9896,
+      "step": 14746
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.836912362159024e-05,
+      "loss": 1.0808,
+      "step": 14747
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.836019465154695e-05,
+      "loss": 1.0522,
+      "step": 14748
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.835126568150364e-05,
+      "loss": 0.9228,
+      "step": 14749
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.834233671146033e-05,
+      "loss": 1.0107,
+      "step": 14750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.833340774141703e-05,
+      "loss": 1.0758,
+      "step": 14751
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.832447877137372e-05,
+      "loss": 1.0862,
+      "step": 14752
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.831554980133042e-05,
+      "loss": 1.0525,
+      "step": 14753
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.830662083128712e-05,
+      "loss": 1.0099,
+      "step": 14754
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.82976918612438e-05,
+      "loss": 1.1099,
+      "step": 14755
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.828876289120051e-05,
+      "loss": 1.0115,
+      "step": 14756
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.82798339211572e-05,
+      "loss": 0.9873,
+      "step": 14757
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.827090495111389e-05,
+      "loss": 1.0605,
+      "step": 14758
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.82619759810706e-05,
+      "loss": 1.0071,
+      "step": 14759
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.825304701102729e-05,
+      "loss": 0.947,
+      "step": 14760
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.824411804098398e-05,
+      "loss": 0.9813,
+      "step": 14761
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.823518907094068e-05,
+      "loss": 1.0936,
+      "step": 14762
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.822626010089737e-05,
+      "loss": 0.9506,
+      "step": 14763
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.821733113085406e-05,
+      "loss": 1.094,
+      "step": 14764
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.820840216081076e-05,
+      "loss": 1.0716,
+      "step": 14765
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.819947319076746e-05,
+      "loss": 0.8937,
+      "step": 14766
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.819054422072415e-05,
+      "loss": 0.9424,
+      "step": 14767
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.818161525068084e-05,
+      "loss": 1.1779,
+      "step": 14768
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.817268628063753e-05,
+      "loss": 0.9722,
+      "step": 14769
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.816375731059422e-05,
+      "loss": 1.0796,
+      "step": 14770
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.815482834055092e-05,
+      "loss": 0.9051,
+      "step": 14771
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.814589937050761e-05,
+      "loss": 1.1003,
+      "step": 14772
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.81369704004643e-05,
+      "loss": 1.0578,
+      "step": 14773
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.8128041430421e-05,
+      "loss": 1.1193,
+      "step": 14774
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.81191124603777e-05,
+      "loss": 0.9996,
+      "step": 14775
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.811018349033439e-05,
+      "loss": 1.0383,
+      "step": 14776
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.810125452029109e-05,
+      "loss": 1.0314,
+      "step": 14777
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.809232555024778e-05,
+      "loss": 0.9829,
+      "step": 14778
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.808339658020447e-05,
+      "loss": 0.9974,
+      "step": 14779
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.807446761016118e-05,
+      "loss": 0.9857,
+      "step": 14780
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.806553864011787e-05,
+      "loss": 0.9872,
+      "step": 14781
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.805660967007456e-05,
+      "loss": 0.9977,
+      "step": 14782
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.804768070003126e-05,
+      "loss": 0.9126,
+      "step": 14783
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.803875172998795e-05,
+      "loss": 0.8366,
+      "step": 14784
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.802982275994464e-05,
+      "loss": 0.8834,
+      "step": 14785
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.802089378990135e-05,
+      "loss": 1.0092,
+      "step": 14786
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.801196481985804e-05,
+      "loss": 1.0237,
+      "step": 14787
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.800303584981473e-05,
+      "loss": 0.9584,
+      "step": 14788
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.799410687977142e-05,
+      "loss": 1.0345,
+      "step": 14789
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.798517790972812e-05,
+      "loss": 0.9522,
+      "step": 14790
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.797624893968481e-05,
+      "loss": 1.0817,
+      "step": 14791
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.79673199696415e-05,
+      "loss": 0.9954,
+      "step": 14792
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.795839099959819e-05,
+      "loss": 0.9714,
+      "step": 14793
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.794946202955488e-05,
+      "loss": 1.0453,
+      "step": 14794
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.794053305951159e-05,
+      "loss": 1.0761,
+      "step": 14795
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.793160408946828e-05,
+      "loss": 1.0632,
+      "step": 14796
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.792267511942497e-05,
+      "loss": 0.9068,
+      "step": 14797
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.791374614938167e-05,
+      "loss": 1.1322,
+      "step": 14798
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.790481717933836e-05,
+      "loss": 1.1013,
+      "step": 14799
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.789588820929507e-05,
+      "loss": 1.3286,
+      "step": 14800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.788695923925176e-05,
+      "loss": 0.9806,
+      "step": 14801
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.787803026920845e-05,
+      "loss": 0.9936,
+      "step": 14802
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.786910129916515e-05,
+      "loss": 1.1254,
+      "step": 14803
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.786017232912184e-05,
+      "loss": 1.0422,
+      "step": 14804
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.785124335907853e-05,
+      "loss": 1.1551,
+      "step": 14805
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.784231438903524e-05,
+      "loss": 1.0735,
+      "step": 14806
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.783338541899193e-05,
+      "loss": 0.9889,
+      "step": 14807
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.782445644894862e-05,
+      "loss": 1.0174,
+      "step": 14808
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.781552747890532e-05,
+      "loss": 1.0737,
+      "step": 14809
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.780659850886201e-05,
+      "loss": 1.0896,
+      "step": 14810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.77976695388187e-05,
+      "loss": 1.0234,
+      "step": 14811
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.77887405687754e-05,
+      "loss": 0.9839,
+      "step": 14812
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.777981159873208e-05,
+      "loss": 0.9404,
+      "step": 14813
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.777088262868877e-05,
+      "loss": 0.9331,
+      "step": 14814
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.776195365864548e-05,
+      "loss": 0.9779,
+      "step": 14815
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.775302468860217e-05,
+      "loss": 1.1083,
+      "step": 14816
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.774409571855886e-05,
+      "loss": 0.9851,
+      "step": 14817
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.773516674851556e-05,
+      "loss": 1.0054,
+      "step": 14818
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.772623777847225e-05,
+      "loss": 1.0652,
+      "step": 14819
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.771730880842894e-05,
+      "loss": 1.1258,
+      "step": 14820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.770837983838565e-05,
+      "loss": 0.8683,
+      "step": 14821
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.769945086834234e-05,
+      "loss": 0.9848,
+      "step": 14822
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.769052189829903e-05,
+      "loss": 1.0518,
+      "step": 14823
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.768159292825573e-05,
+      "loss": 1.0921,
+      "step": 14824
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.767266395821242e-05,
+      "loss": 1.0359,
+      "step": 14825
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.766373498816911e-05,
+      "loss": 1.061,
+      "step": 14826
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.765480601812582e-05,
+      "loss": 1.0457,
+      "step": 14827
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.764587704808251e-05,
+      "loss": 1.1321,
+      "step": 14828
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.76369480780392e-05,
+      "loss": 1.057,
+      "step": 14829
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.76280191079959e-05,
+      "loss": 1.0035,
+      "step": 14830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.76190901379526e-05,
+      "loss": 0.9856,
+      "step": 14831
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.761016116790928e-05,
+      "loss": 1.1524,
+      "step": 14832
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.760123219786599e-05,
+      "loss": 1.0178,
+      "step": 14833
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.759230322782268e-05,
+      "loss": 0.9496,
+      "step": 14834
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.758337425777937e-05,
+      "loss": 1.1682,
+      "step": 14835
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.757444528773606e-05,
+      "loss": 0.9265,
+      "step": 14836
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.756551631769275e-05,
+      "loss": 0.9531,
+      "step": 14837
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.755658734764944e-05,
+      "loss": 1.2619,
+      "step": 14838
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.754765837760614e-05,
+      "loss": 1.0744,
+      "step": 14839
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.753872940756284e-05,
+      "loss": 1.0058,
+      "step": 14840
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.752980043751953e-05,
+      "loss": 1.0421,
+      "step": 14841
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.752087146747623e-05,
+      "loss": 1.0956,
+      "step": 14842
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.751194249743292e-05,
+      "loss": 1.0155,
+      "step": 14843
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.750301352738961e-05,
+      "loss": 0.9378,
+      "step": 14844
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.749408455734631e-05,
+      "loss": 1.0611,
+      "step": 14845
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.7485155587303e-05,
+      "loss": 0.9635,
+      "step": 14846
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.74762266172597e-05,
+      "loss": 1.039,
+      "step": 14847
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.74672976472164e-05,
+      "loss": 0.946,
+      "step": 14848
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.745836867717309e-05,
+      "loss": 1.0545,
+      "step": 14849
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.74494397071298e-05,
+      "loss": 0.8989,
+      "step": 14850
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.744051073708648e-05,
+      "loss": 1.1231,
+      "step": 14851
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.743158176704318e-05,
+      "loss": 0.9322,
+      "step": 14852
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.742265279699988e-05,
+      "loss": 1.0789,
+      "step": 14853
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.741372382695657e-05,
+      "loss": 1.17,
+      "step": 14854
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.740479485691326e-05,
+      "loss": 1.073,
+      "step": 14855
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.739586588686996e-05,
+      "loss": 1.073,
+      "step": 14856
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.738693691682665e-05,
+      "loss": 1.0715,
+      "step": 14857
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.737800794678335e-05,
+      "loss": 1.0057,
+      "step": 14858
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.736907897674004e-05,
+      "loss": 0.9342,
+      "step": 14859
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.736015000669673e-05,
+      "loss": 1.0286,
+      "step": 14860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.735122103665342e-05,
+      "loss": 1.0126,
+      "step": 14861
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.734229206661012e-05,
+      "loss": 0.9548,
+      "step": 14862
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.733336309656681e-05,
+      "loss": 1.085,
+      "step": 14863
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.73244341265235e-05,
+      "loss": 0.9651,
+      "step": 14864
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.73155051564802e-05,
+      "loss": 0.8497,
+      "step": 14865
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.73065761864369e-05,
+      "loss": 0.9724,
+      "step": 14866
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.729764721639359e-05,
+      "loss": 1.0461,
+      "step": 14867
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.728871824635029e-05,
+      "loss": 1.1274,
+      "step": 14868
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.727978927630698e-05,
+      "loss": 1.0086,
+      "step": 14869
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.727086030626367e-05,
+      "loss": 0.9544,
+      "step": 14870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.726193133622038e-05,
+      "loss": 1.0661,
+      "step": 14871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.725300236617707e-05,
+      "loss": 1.054,
+      "step": 14872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.724407339613376e-05,
+      "loss": 1.1556,
+      "step": 14873
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.723514442609046e-05,
+      "loss": 1.0755,
+      "step": 14874
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.722621545604715e-05,
+      "loss": 1.0143,
+      "step": 14875
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.721728648600384e-05,
+      "loss": 1.1056,
+      "step": 14876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.720835751596055e-05,
+      "loss": 0.9651,
+      "step": 14877
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.719942854591724e-05,
+      "loss": 1.1009,
+      "step": 14878
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.719049957587393e-05,
+      "loss": 1.0668,
+      "step": 14879
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.718157060583062e-05,
+      "loss": 1.0074,
+      "step": 14880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.717264163578731e-05,
+      "loss": 1.0771,
+      "step": 14881
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.716371266574401e-05,
+      "loss": 1.1143,
+      "step": 14882
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.71547836957007e-05,
+      "loss": 1.0468,
+      "step": 14883
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.714585472565739e-05,
+      "loss": 1.056,
+      "step": 14884
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.713692575561408e-05,
+      "loss": 1.0652,
+      "step": 14885
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.712799678557079e-05,
+      "loss": 0.8986,
+      "step": 14886
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.711906781552748e-05,
+      "loss": 1.0332,
+      "step": 14887
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.711013884548417e-05,
+      "loss": 0.9875,
+      "step": 14888
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.710120987544087e-05,
+      "loss": 1.0232,
+      "step": 14889
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.709228090539756e-05,
+      "loss": 1.1175,
+      "step": 14890
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.708335193535425e-05,
+      "loss": 1.0957,
+      "step": 14891
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.707442296531096e-05,
+      "loss": 1.0277,
+      "step": 14892
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.706549399526765e-05,
+      "loss": 1.0277,
+      "step": 14893
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.705656502522434e-05,
+      "loss": 0.9561,
+      "step": 14894
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.704763605518104e-05,
+      "loss": 1.1407,
+      "step": 14895
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.703870708513773e-05,
+      "loss": 1.1145,
+      "step": 14896
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.702977811509444e-05,
+      "loss": 1.0748,
+      "step": 14897
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.702084914505113e-05,
+      "loss": 1.069,
+      "step": 14898
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.701192017500782e-05,
+      "loss": 0.9704,
+      "step": 14899
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.700299120496452e-05,
+      "loss": 1.0166,
+      "step": 14900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.699406223492121e-05,
+      "loss": 1.0598,
+      "step": 14901
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.69851332648779e-05,
+      "loss": 1.0651,
+      "step": 14902
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.697620429483459e-05,
+      "loss": 1.0471,
+      "step": 14903
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.696727532479128e-05,
+      "loss": 0.9394,
+      "step": 14904
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.695834635474797e-05,
+      "loss": 1.0967,
+      "step": 14905
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.694941738470468e-05,
+      "loss": 1.0074,
+      "step": 14906
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.694048841466137e-05,
+      "loss": 1.0448,
+      "step": 14907
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.693155944461806e-05,
+      "loss": 0.9866,
+      "step": 14908
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.692263047457476e-05,
+      "loss": 1.0398,
+      "step": 14909
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.691370150453145e-05,
+      "loss": 1.0238,
+      "step": 14910
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.690477253448814e-05,
+      "loss": 1.0738,
+      "step": 14911
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.689584356444485e-05,
+      "loss": 1.0593,
+      "step": 14912
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.688691459440154e-05,
+      "loss": 1.0253,
+      "step": 14913
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.687798562435823e-05,
+      "loss": 1.1676,
+      "step": 14914
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.686905665431493e-05,
+      "loss": 0.9332,
+      "step": 14915
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.686012768427162e-05,
+      "loss": 1.0594,
+      "step": 14916
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.685119871422831e-05,
+      "loss": 1.2149,
+      "step": 14917
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.684226974418502e-05,
+      "loss": 1.4277,
+      "step": 14918
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.683334077414171e-05,
+      "loss": 1.1274,
+      "step": 14919
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.68244118040984e-05,
+      "loss": 1.1065,
+      "step": 14920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.68154828340551e-05,
+      "loss": 1.0605,
+      "step": 14921
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.68065538640118e-05,
+      "loss": 1.0298,
+      "step": 14922
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.679762489396848e-05,
+      "loss": 1.1518,
+      "step": 14923
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.678869592392519e-05,
+      "loss": 0.9221,
+      "step": 14924
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.677976695388188e-05,
+      "loss": 1.0326,
+      "step": 14925
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.677083798383857e-05,
+      "loss": 1.1832,
+      "step": 14926
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.676190901379526e-05,
+      "loss": 1.1654,
+      "step": 14927
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.675298004375195e-05,
+      "loss": 1.543,
+      "step": 14928
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.674405107370864e-05,
+      "loss": 1.0086,
+      "step": 14929
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.673512210366534e-05,
+      "loss": 1.0137,
+      "step": 14930
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.672619313362204e-05,
+      "loss": 1.1733,
+      "step": 14931
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.671726416357873e-05,
+      "loss": 1.1186,
+      "step": 14932
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.670833519353543e-05,
+      "loss": 1.0617,
+      "step": 14933
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.669940622349212e-05,
+      "loss": 1.0184,
+      "step": 14934
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.669047725344881e-05,
+      "loss": 0.9466,
+      "step": 14935
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.668154828340551e-05,
+      "loss": 0.905,
+      "step": 14936
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.66726193133622e-05,
+      "loss": 1.004,
+      "step": 14937
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.66636903433189e-05,
+      "loss": 1.0633,
+      "step": 14938
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.66547613732756e-05,
+      "loss": 1.082,
+      "step": 14939
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.664583240323229e-05,
+      "loss": 1.1042,
+      "step": 14940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.663690343318898e-05,
+      "loss": 1.0039,
+      "step": 14941
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.662797446314568e-05,
+      "loss": 1.088,
+      "step": 14942
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.661904549310238e-05,
+      "loss": 0.9779,
+      "step": 14943
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.661011652305907e-05,
+      "loss": 1.0476,
+      "step": 14944
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.660118755301577e-05,
+      "loss": 1.0117,
+      "step": 14945
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.659225858297246e-05,
+      "loss": 1.092,
+      "step": 14946
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.658332961292916e-05,
+      "loss": 0.9748,
+      "step": 14947
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.657440064288585e-05,
+      "loss": 0.934,
+      "step": 14948
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.656547167284254e-05,
+      "loss": 1.1025,
+      "step": 14949
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.655654270279924e-05,
+      "loss": 1.0399,
+      "step": 14950
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.654761373275593e-05,
+      "loss": 1.1327,
+      "step": 14951
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.653868476271262e-05,
+      "loss": 0.9747,
+      "step": 14952
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.652975579266932e-05,
+      "loss": 1.0863,
+      "step": 14953
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.652082682262601e-05,
+      "loss": 1.0506,
+      "step": 14954
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.65118978525827e-05,
+      "loss": 1.017,
+      "step": 14955
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.65029688825394e-05,
+      "loss": 1.1247,
+      "step": 14956
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.64940399124961e-05,
+      "loss": 1.1167,
+      "step": 14957
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.648511094245279e-05,
+      "loss": 1.0046,
+      "step": 14958
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.647618197240949e-05,
+      "loss": 1.0395,
+      "step": 14959
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.646725300236618e-05,
+      "loss": 1.021,
+      "step": 14960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.645832403232287e-05,
+      "loss": 1.0648,
+      "step": 14961
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.644939506227958e-05,
+      "loss": 0.9882,
+      "step": 14962
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.644046609223627e-05,
+      "loss": 0.8977,
+      "step": 14963
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.643153712219296e-05,
+      "loss": 1.0083,
+      "step": 14964
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.642260815214966e-05,
+      "loss": 0.9085,
+      "step": 14965
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.641367918210635e-05,
+      "loss": 1.0558,
+      "step": 14966
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.640475021206304e-05,
+      "loss": 0.9465,
+      "step": 14967
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.639582124201975e-05,
+      "loss": 1.08,
+      "step": 14968
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.638689227197644e-05,
+      "loss": 1.1062,
+      "step": 14969
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.637796330193313e-05,
+      "loss": 0.9897,
+      "step": 14970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.636903433188982e-05,
+      "loss": 0.9155,
+      "step": 14971
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.636010536184651e-05,
+      "loss": 1.0471,
+      "step": 14972
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.635117639180321e-05,
+      "loss": 0.8158,
+      "step": 14973
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.63422474217599e-05,
+      "loss": 1.0054,
+      "step": 14974
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.633331845171659e-05,
+      "loss": 0.9896,
+      "step": 14975
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.632438948167328e-05,
+      "loss": 1.2041,
+      "step": 14976
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.631546051162999e-05,
+      "loss": 1.0916,
+      "step": 14977
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.630653154158668e-05,
+      "loss": 0.9359,
+      "step": 14978
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.629760257154337e-05,
+      "loss": 1.0253,
+      "step": 14979
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.628867360150007e-05,
+      "loss": 1.2292,
+      "step": 14980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.627974463145676e-05,
+      "loss": 1.1079,
+      "step": 14981
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.627081566141345e-05,
+      "loss": 1.0232,
+      "step": 14982
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.626188669137016e-05,
+      "loss": 1.0582,
+      "step": 14983
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.625295772132685e-05,
+      "loss": 0.9043,
+      "step": 14984
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.624402875128354e-05,
+      "loss": 1.0898,
+      "step": 14985
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.623509978124024e-05,
+      "loss": 1.096,
+      "step": 14986
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.622617081119693e-05,
+      "loss": 0.9146,
+      "step": 14987
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.621724184115362e-05,
+      "loss": 1.0451,
+      "step": 14988
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.620831287111033e-05,
+      "loss": 0.9704,
+      "step": 14989
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.619938390106702e-05,
+      "loss": 1.047,
+      "step": 14990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.619045493102371e-05,
+      "loss": 0.9552,
+      "step": 14991
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.618152596098041e-05,
+      "loss": 0.9933,
+      "step": 14992
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.61725969909371e-05,
+      "loss": 0.9861,
+      "step": 14993
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.616366802089379e-05,
+      "loss": 1.1529,
+      "step": 14994
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.615473905085048e-05,
+      "loss": 0.9566,
+      "step": 14995
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.614581008080717e-05,
+      "loss": 1.005,
+      "step": 14996
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.613688111076388e-05,
+      "loss": 1.0274,
+      "step": 14997
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.612795214072057e-05,
+      "loss": 1.0067,
+      "step": 14998
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.611902317067726e-05,
+      "loss": 1.1476,
+      "step": 14999
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.611009420063396e-05,
+      "loss": 1.0746,
+      "step": 15000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.610116523059065e-05,
+      "loss": 0.9707,
+      "step": 15001
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.609223626054734e-05,
+      "loss": 0.9457,
+      "step": 15002
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.608330729050405e-05,
+      "loss": 1.0851,
+      "step": 15003
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.607437832046074e-05,
+      "loss": 1.0049,
+      "step": 15004
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.606544935041743e-05,
+      "loss": 1.1548,
+      "step": 15005
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.605652038037413e-05,
+      "loss": 0.9655,
+      "step": 15006
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.604759141033082e-05,
+      "loss": 0.989,
+      "step": 15007
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.603866244028751e-05,
+      "loss": 1.0307,
+      "step": 15008
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.602973347024422e-05,
+      "loss": 0.9856,
+      "step": 15009
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.602080450020091e-05,
+      "loss": 0.9233,
+      "step": 15010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.60118755301576e-05,
+      "loss": 1.0124,
+      "step": 15011
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.60029465601143e-05,
+      "loss": 0.8673,
+      "step": 15012
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.5994017590071e-05,
+      "loss": 1.194,
+      "step": 15013
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.598508862002768e-05,
+      "loss": 0.9908,
+      "step": 15014
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.597615964998439e-05,
+      "loss": 0.8801,
+      "step": 15015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.596723067994108e-05,
+      "loss": 1.1222,
+      "step": 15016
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.595830170989777e-05,
+      "loss": 0.9671,
+      "step": 15017
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.594937273985446e-05,
+      "loss": 1.008,
+      "step": 15018
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.594044376981115e-05,
+      "loss": 1.0395,
+      "step": 15019
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.593151479976784e-05,
+      "loss": 1.0558,
+      "step": 15020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.592258582972454e-05,
+      "loss": 0.9714,
+      "step": 15021
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.591365685968123e-05,
+      "loss": 1.2212,
+      "step": 15022
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.590472788963793e-05,
+      "loss": 1.0258,
+      "step": 15023
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.589579891959463e-05,
+      "loss": 0.981,
+      "step": 15024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.588686994955132e-05,
+      "loss": 0.9263,
+      "step": 15025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.587794097950801e-05,
+      "loss": 1.0239,
+      "step": 15026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.586901200946471e-05,
+      "loss": 1.0562,
+      "step": 15027
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.58600830394214e-05,
+      "loss": 1.0127,
+      "step": 15028
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.58511540693781e-05,
+      "loss": 0.9944,
+      "step": 15029
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.58422250993348e-05,
+      "loss": 1.0325,
+      "step": 15030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.583329612929149e-05,
+      "loss": 0.9924,
+      "step": 15031
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.582436715924818e-05,
+      "loss": 1.0935,
+      "step": 15032
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.581543818920488e-05,
+      "loss": 1.0709,
+      "step": 15033
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.580650921916157e-05,
+      "loss": 1.0872,
+      "step": 15034
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.579758024911827e-05,
+      "loss": 1.1414,
+      "step": 15035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.578865127907497e-05,
+      "loss": 1.0657,
+      "step": 15036
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.577972230903166e-05,
+      "loss": 1.1459,
+      "step": 15037
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.577079333898835e-05,
+      "loss": 1.0636,
+      "step": 15038
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.576186436894505e-05,
+      "loss": 1.0359,
+      "step": 15039
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.575293539890174e-05,
+      "loss": 1.0659,
+      "step": 15040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.574400642885844e-05,
+      "loss": 1.0253,
+      "step": 15041
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.573507745881513e-05,
+      "loss": 0.9686,
+      "step": 15042
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.572614848877182e-05,
+      "loss": 0.993,
+      "step": 15043
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.571721951872852e-05,
+      "loss": 1.0343,
+      "step": 15044
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.570829054868521e-05,
+      "loss": 1.0641,
+      "step": 15045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.56993615786419e-05,
+      "loss": 1.0679,
+      "step": 15046
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.56904326085986e-05,
+      "loss": 1.085,
+      "step": 15047
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.56815036385553e-05,
+      "loss": 0.9471,
+      "step": 15048
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.567257466851199e-05,
+      "loss": 0.9491,
+      "step": 15049
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.566364569846869e-05,
+      "loss": 0.984,
+      "step": 15050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.565471672842538e-05,
+      "loss": 1.0384,
+      "step": 15051
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.564578775838207e-05,
+      "loss": 1.0037,
+      "step": 15052
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.563685878833878e-05,
+      "loss": 1.103,
+      "step": 15053
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.562792981829547e-05,
+      "loss": 1.1391,
+      "step": 15054
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.561900084825216e-05,
+      "loss": 1.0203,
+      "step": 15055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.561007187820886e-05,
+      "loss": 1.0387,
+      "step": 15056
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.560114290816555e-05,
+      "loss": 1.062,
+      "step": 15057
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.559221393812224e-05,
+      "loss": 1.047,
+      "step": 15058
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.558328496807895e-05,
+      "loss": 1.1104,
+      "step": 15059
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.557435599803564e-05,
+      "loss": 1.1222,
+      "step": 15060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.556542702799233e-05,
+      "loss": 1.0154,
+      "step": 15061
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.555649805794902e-05,
+      "loss": 0.9953,
+      "step": 15062
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.554756908790571e-05,
+      "loss": 0.9812,
+      "step": 15063
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.553864011786241e-05,
+      "loss": 0.9815,
+      "step": 15064
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.55297111478191e-05,
+      "loss": 1.0026,
+      "step": 15065
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.552078217777579e-05,
+      "loss": 0.9254,
+      "step": 15066
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.551185320773248e-05,
+      "loss": 1.1663,
+      "step": 15067
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.550292423768919e-05,
+      "loss": 1.0578,
+      "step": 15068
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.549399526764588e-05,
+      "loss": 1.0677,
+      "step": 15069
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.548506629760257e-05,
+      "loss": 0.9958,
+      "step": 15070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.547613732755927e-05,
+      "loss": 1.0692,
+      "step": 15071
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.546720835751596e-05,
+      "loss": 1.1071,
+      "step": 15072
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.545827938747265e-05,
+      "loss": 1.0824,
+      "step": 15073
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.544935041742936e-05,
+      "loss": 0.9217,
+      "step": 15074
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.544042144738605e-05,
+      "loss": 1.0356,
+      "step": 15075
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.543149247734274e-05,
+      "loss": 1.0524,
+      "step": 15076
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.542256350729944e-05,
+      "loss": 1.1222,
+      "step": 15077
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.541363453725613e-05,
+      "loss": 1.108,
+      "step": 15078
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.540470556721282e-05,
+      "loss": 0.914,
+      "step": 15079
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.539577659716953e-05,
+      "loss": 1.2257,
+      "step": 15080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.538684762712622e-05,
+      "loss": 0.9651,
+      "step": 15081
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.537791865708291e-05,
+      "loss": 1.0043,
+      "step": 15082
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.536898968703961e-05,
+      "loss": 1.0829,
+      "step": 15083
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.53600607169963e-05,
+      "loss": 1.0164,
+      "step": 15084
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.535113174695299e-05,
+      "loss": 0.9535,
+      "step": 15085
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.534220277690968e-05,
+      "loss": 0.9109,
+      "step": 15086
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.533327380686637e-05,
+      "loss": 1.0947,
+      "step": 15087
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.532434483682306e-05,
+      "loss": 0.9953,
+      "step": 15088
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.531541586677977e-05,
+      "loss": 1.0386,
+      "step": 15089
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.530648689673646e-05,
+      "loss": 1.0924,
+      "step": 15090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.529755792669315e-05,
+      "loss": 1.052,
+      "step": 15091
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.528862895664985e-05,
+      "loss": 0.9425,
+      "step": 15092
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.527969998660654e-05,
+      "loss": 1.121,
+      "step": 15093
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.527077101656325e-05,
+      "loss": 1.0578,
+      "step": 15094
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.526184204651994e-05,
+      "loss": 1.0015,
+      "step": 15095
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.525291307647663e-05,
+      "loss": 0.9926,
+      "step": 15096
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.524398410643333e-05,
+      "loss": 1.155,
+      "step": 15097
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.523505513639002e-05,
+      "loss": 1.2994,
+      "step": 15098
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.522612616634671e-05,
+      "loss": 0.9327,
+      "step": 15099
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.521719719630342e-05,
+      "loss": 1.1371,
+      "step": 15100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.520826822626011e-05,
+      "loss": 1.1601,
+      "step": 15101
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.51993392562168e-05,
+      "loss": 0.8817,
+      "step": 15102
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.51904102861735e-05,
+      "loss": 1.2552,
+      "step": 15103
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.518148131613019e-05,
+      "loss": 0.9647,
+      "step": 15104
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.517255234608688e-05,
+      "loss": 0.9326,
+      "step": 15105
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.516362337604359e-05,
+      "loss": 1.0522,
+      "step": 15106
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.515469440600028e-05,
+      "loss": 0.9154,
+      "step": 15107
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.514576543595697e-05,
+      "loss": 1.0812,
+      "step": 15108
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.513683646591366e-05,
+      "loss": 1.1034,
+      "step": 15109
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.512790749587035e-05,
+      "loss": 1.0559,
+      "step": 15110
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.511897852582704e-05,
+      "loss": 0.9845,
+      "step": 15111
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.511004955578374e-05,
+      "loss": 1.0171,
+      "step": 15112
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.510112058574043e-05,
+      "loss": 1.2133,
+      "step": 15113
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.509219161569712e-05,
+      "loss": 1.148,
+      "step": 15114
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.508326264565383e-05,
+      "loss": 0.9566,
+      "step": 15115
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.507433367561052e-05,
+      "loss": 1.1029,
+      "step": 15116
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.506540470556721e-05,
+      "loss": 1.03,
+      "step": 15117
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.505647573552391e-05,
+      "loss": 1.0327,
+      "step": 15118
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.50475467654806e-05,
+      "loss": 0.9261,
+      "step": 15119
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.50386177954373e-05,
+      "loss": 1.1673,
+      "step": 15120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.5029688825394e-05,
+      "loss": 0.9664,
+      "step": 15121
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.502075985535069e-05,
+      "loss": 1.0399,
+      "step": 15122
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.501183088530738e-05,
+      "loss": 1.0054,
+      "step": 15123
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.500290191526408e-05,
+      "loss": 0.9854,
+      "step": 15124
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.499397294522077e-05,
+      "loss": 0.9849,
+      "step": 15125
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.498504397517746e-05,
+      "loss": 0.9387,
+      "step": 15126
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.497611500513417e-05,
+      "loss": 1.2125,
+      "step": 15127
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.496718603509086e-05,
+      "loss": 0.9709,
+      "step": 15128
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.495825706504755e-05,
+      "loss": 0.9801,
+      "step": 15129
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.494932809500425e-05,
+      "loss": 0.9935,
+      "step": 15130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.494039912496094e-05,
+      "loss": 1.0641,
+      "step": 15131
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.493147015491763e-05,
+      "loss": 1.076,
+      "step": 15132
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.492254118487433e-05,
+      "loss": 1.0375,
+      "step": 15133
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.491361221483102e-05,
+      "loss": 1.114,
+      "step": 15134
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.49046832447877e-05,
+      "loss": 0.9703,
+      "step": 15135
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.489575427474441e-05,
+      "loss": 0.9726,
+      "step": 15136
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.48868253047011e-05,
+      "loss": 1.0865,
+      "step": 15137
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.487789633465779e-05,
+      "loss": 1.0275,
+      "step": 15138
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.48689673646145e-05,
+      "loss": 1.0396,
+      "step": 15139
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.486003839457119e-05,
+      "loss": 0.9599,
+      "step": 15140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.485110942452788e-05,
+      "loss": 1.0739,
+      "step": 15141
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.484218045448458e-05,
+      "loss": 0.9962,
+      "step": 15142
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.483325148444127e-05,
+      "loss": 1.0513,
+      "step": 15143
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.482432251439797e-05,
+      "loss": 0.9815,
+      "step": 15144
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.481539354435467e-05,
+      "loss": 0.9717,
+      "step": 15145
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.480646457431136e-05,
+      "loss": 1.0306,
+      "step": 15146
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.479753560426806e-05,
+      "loss": 0.9523,
+      "step": 15147
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.478860663422475e-05,
+      "loss": 1.0961,
+      "step": 15148
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.477967766418144e-05,
+      "loss": 1.034,
+      "step": 15149
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.477074869413814e-05,
+      "loss": 1.0043,
+      "step": 15150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.476181972409484e-05,
+      "loss": 1.0827,
+      "step": 15151
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.475289075405153e-05,
+      "loss": 0.9707,
+      "step": 15152
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.474396178400822e-05,
+      "loss": 0.92,
+      "step": 15153
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.47350328139649e-05,
+      "loss": 0.9856,
+      "step": 15154
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.47261038439216e-05,
+      "loss": 1.0477,
+      "step": 15155
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.47171748738783e-05,
+      "loss": 1.1119,
+      "step": 15156
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.470824590383499e-05,
+      "loss": 1.0485,
+      "step": 15157
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.469931693379168e-05,
+      "loss": 1.0331,
+      "step": 15158
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.469038796374839e-05,
+      "loss": 0.9677,
+      "step": 15159
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.468145899370508e-05,
+      "loss": 1.0611,
+      "step": 15160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.467253002366177e-05,
+      "loss": 0.9744,
+      "step": 15161
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.466360105361847e-05,
+      "loss": 1.1022,
+      "step": 15162
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.465467208357516e-05,
+      "loss": 1.0629,
+      "step": 15163
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.464574311353185e-05,
+      "loss": 1.1002,
+      "step": 15164
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.463681414348856e-05,
+      "loss": 0.9847,
+      "step": 15165
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.462788517344525e-05,
+      "loss": 0.9939,
+      "step": 15166
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.461895620340194e-05,
+      "loss": 1.0403,
+      "step": 15167
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.461002723335864e-05,
+      "loss": 0.9191,
+      "step": 15168
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.460109826331533e-05,
+      "loss": 1.0214,
+      "step": 15169
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.459216929327202e-05,
+      "loss": 1.076,
+      "step": 15170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.458324032322873e-05,
+      "loss": 1.074,
+      "step": 15171
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.457431135318542e-05,
+      "loss": 1.0481,
+      "step": 15172
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.456538238314211e-05,
+      "loss": 1.0887,
+      "step": 15173
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.455645341309881e-05,
+      "loss": 1.0812,
+      "step": 15174
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.45475244430555e-05,
+      "loss": 1.1259,
+      "step": 15175
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.453859547301219e-05,
+      "loss": 1.0309,
+      "step": 15176
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.452966650296888e-05,
+      "loss": 0.9974,
+      "step": 15177
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.452073753292557e-05,
+      "loss": 1.0896,
+      "step": 15178
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.451180856288226e-05,
+      "loss": 1.0339,
+      "step": 15179
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.450287959283897e-05,
+      "loss": 1.0011,
+      "step": 15180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.449395062279566e-05,
+      "loss": 1.0524,
+      "step": 15181
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.448502165275235e-05,
+      "loss": 0.9719,
+      "step": 15182
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.447609268270905e-05,
+      "loss": 0.959,
+      "step": 15183
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.446716371266574e-05,
+      "loss": 0.9655,
+      "step": 15184
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.445823474262243e-05,
+      "loss": 1.0185,
+      "step": 15185
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.444930577257914e-05,
+      "loss": 0.9833,
+      "step": 15186
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.444037680253583e-05,
+      "loss": 0.9272,
+      "step": 15187
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.443144783249252e-05,
+      "loss": 0.9903,
+      "step": 15188
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.442251886244922e-05,
+      "loss": 1.3573,
+      "step": 15189
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.441358989240591e-05,
+      "loss": 0.9853,
+      "step": 15190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.440466092236262e-05,
+      "loss": 0.9143,
+      "step": 15191
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.439573195231931e-05,
+      "loss": 1.0516,
+      "step": 15192
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.4386802982276e-05,
+      "loss": 0.9623,
+      "step": 15193
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.43778740122327e-05,
+      "loss": 1.1038,
+      "step": 15194
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.436894504218939e-05,
+      "loss": 1.0441,
+      "step": 15195
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.436001607214608e-05,
+      "loss": 0.9648,
+      "step": 15196
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.435108710210279e-05,
+      "loss": 1.0311,
+      "step": 15197
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.434215813205948e-05,
+      "loss": 1.0943,
+      "step": 15198
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.433322916201617e-05,
+      "loss": 0.989,
+      "step": 15199
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.432430019197286e-05,
+      "loss": 0.9118,
+      "step": 15200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.431537122192955e-05,
+      "loss": 0.9663,
+      "step": 15201
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.430644225188624e-05,
+      "loss": 1.1194,
+      "step": 15202
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.429751328184294e-05,
+      "loss": 0.9165,
+      "step": 15203
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.428858431179963e-05,
+      "loss": 0.9809,
+      "step": 15204
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.427965534175632e-05,
+      "loss": 0.9505,
+      "step": 15205
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.427072637171303e-05,
+      "loss": 0.8278,
+      "step": 15206
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.426179740166972e-05,
+      "loss": 1.1255,
+      "step": 15207
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.425286843162641e-05,
+      "loss": 1.1278,
+      "step": 15208
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.424393946158311e-05,
+      "loss": 1.0147,
+      "step": 15209
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.42350104915398e-05,
+      "loss": 1.0253,
+      "step": 15210
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.42260815214965e-05,
+      "loss": 1.0109,
+      "step": 15211
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.42171525514532e-05,
+      "loss": 0.9566,
+      "step": 15212
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.420822358140989e-05,
+      "loss": 0.9786,
+      "step": 15213
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.419929461136658e-05,
+      "loss": 1.0207,
+      "step": 15214
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.419036564132328e-05,
+      "loss": 1.0356,
+      "step": 15215
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.418143667127997e-05,
+      "loss": 0.9817,
+      "step": 15216
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.417250770123666e-05,
+      "loss": 0.9756,
+      "step": 15217
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.416357873119337e-05,
+      "loss": 0.9503,
+      "step": 15218
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.415464976115006e-05,
+      "loss": 0.9661,
+      "step": 15219
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.414572079110675e-05,
+      "loss": 1.0553,
+      "step": 15220
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.413679182106345e-05,
+      "loss": 1.0013,
+      "step": 15221
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.412786285102014e-05,
+      "loss": 1.1255,
+      "step": 15222
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.411893388097683e-05,
+      "loss": 1.0707,
+      "step": 15223
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.411000491093352e-05,
+      "loss": 1.0866,
+      "step": 15224
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.410107594089022e-05,
+      "loss": 0.9115,
+      "step": 15225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.40921469708469e-05,
+      "loss": 1.0344,
+      "step": 15226
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.408321800080361e-05,
+      "loss": 1.031,
+      "step": 15227
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.40742890307603e-05,
+      "loss": 1.0436,
+      "step": 15228
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.406536006071699e-05,
+      "loss": 0.9577,
+      "step": 15229
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.40564310906737e-05,
+      "loss": 1.0653,
+      "step": 15230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.404750212063039e-05,
+      "loss": 0.9634,
+      "step": 15231
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.403857315058708e-05,
+      "loss": 1.0484,
+      "step": 15232
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.402964418054378e-05,
+      "loss": 0.949,
+      "step": 15233
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.402071521050047e-05,
+      "loss": 0.9894,
+      "step": 15234
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.401178624045716e-05,
+      "loss": 1.0073,
+      "step": 15235
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.400285727041386e-05,
+      "loss": 0.9644,
+      "step": 15236
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.399392830037056e-05,
+      "loss": 1.2028,
+      "step": 15237
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.398499933032725e-05,
+      "loss": 0.9977,
+      "step": 15238
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.397607036028395e-05,
+      "loss": 1.0193,
+      "step": 15239
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.396714139024064e-05,
+      "loss": 1.0242,
+      "step": 15240
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.395821242019734e-05,
+      "loss": 1.0378,
+      "step": 15241
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.394928345015403e-05,
+      "loss": 0.9642,
+      "step": 15242
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.394035448011073e-05,
+      "loss": 1.0305,
+      "step": 15243
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.393142551006742e-05,
+      "loss": 0.9639,
+      "step": 15244
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.39224965400241e-05,
+      "loss": 0.9073,
+      "step": 15245
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.39135675699808e-05,
+      "loss": 1.1142,
+      "step": 15246
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.39046385999375e-05,
+      "loss": 0.9851,
+      "step": 15247
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.389570962989419e-05,
+      "loss": 0.9598,
+      "step": 15248
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.388678065985088e-05,
+      "loss": 0.9537,
+      "step": 15249
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.387785168980759e-05,
+      "loss": 0.9855,
+      "step": 15250
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.386892271976428e-05,
+      "loss": 0.9177,
+      "step": 15251
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.385999374972097e-05,
+      "loss": 0.9865,
+      "step": 15252
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.385106477967767e-05,
+      "loss": 0.9531,
+      "step": 15253
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.384213580963436e-05,
+      "loss": 0.9517,
+      "step": 15254
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.383320683959105e-05,
+      "loss": 0.9811,
+      "step": 15255
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.382427786954776e-05,
+      "loss": 0.981,
+      "step": 15256
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.381534889950445e-05,
+      "loss": 1.0562,
+      "step": 15257
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.380641992946114e-05,
+      "loss": 1.0145,
+      "step": 15258
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.379749095941784e-05,
+      "loss": 0.965,
+      "step": 15259
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.378856198937453e-05,
+      "loss": 1.0237,
+      "step": 15260
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.377963301933122e-05,
+      "loss": 1.0579,
+      "step": 15261
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.377070404928793e-05,
+      "loss": 0.9636,
+      "step": 15262
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.376177507924462e-05,
+      "loss": 0.9387,
+      "step": 15263
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.37528461092013e-05,
+      "loss": 1.0929,
+      "step": 15264
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.374391713915801e-05,
+      "loss": 1.047,
+      "step": 15265
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.37349881691147e-05,
+      "loss": 1.0585,
+      "step": 15266
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.372605919907139e-05,
+      "loss": 0.9904,
+      "step": 15267
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.371713022902808e-05,
+      "loss": 0.9632,
+      "step": 15268
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.370820125898477e-05,
+      "loss": 1.2296,
+      "step": 15269
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.369927228894146e-05,
+      "loss": 1.0459,
+      "step": 15270
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.369034331889817e-05,
+      "loss": 1.1272,
+      "step": 15271
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.368141434885486e-05,
+      "loss": 1.0566,
+      "step": 15272
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.367248537881155e-05,
+      "loss": 0.9813,
+      "step": 15273
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.366355640876825e-05,
+      "loss": 1.0267,
+      "step": 15274
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.365462743872494e-05,
+      "loss": 0.9704,
+      "step": 15275
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.364569846868163e-05,
+      "loss": 1.1146,
+      "step": 15276
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.363676949863834e-05,
+      "loss": 0.976,
+      "step": 15277
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.362784052859503e-05,
+      "loss": 1.0091,
+      "step": 15278
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.361891155855172e-05,
+      "loss": 0.9941,
+      "step": 15279
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.360998258850842e-05,
+      "loss": 0.9711,
+      "step": 15280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.360105361846511e-05,
+      "loss": 0.9484,
+      "step": 15281
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.35921246484218e-05,
+      "loss": 1.0754,
+      "step": 15282
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.358319567837851e-05,
+      "loss": 1.1123,
+      "step": 15283
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.35742667083352e-05,
+      "loss": 1.0197,
+      "step": 15284
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.356533773829189e-05,
+      "loss": 1.0073,
+      "step": 15285
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.355640876824859e-05,
+      "loss": 1.0486,
+      "step": 15286
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.354747979820528e-05,
+      "loss": 0.9868,
+      "step": 15287
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.353855082816197e-05,
+      "loss": 1.0331,
+      "step": 15288
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.352962185811868e-05,
+      "loss": 1.1157,
+      "step": 15289
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.352069288807537e-05,
+      "loss": 1.159,
+      "step": 15290
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.351176391803206e-05,
+      "loss": 1.0892,
+      "step": 15291
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.350283494798875e-05,
+      "loss": 0.9475,
+      "step": 15292
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.349390597794544e-05,
+      "loss": 0.9945,
+      "step": 15293
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.348497700790214e-05,
+      "loss": 0.9892,
+      "step": 15294
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.347604803785883e-05,
+      "loss": 1.2754,
+      "step": 15295
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.346711906781552e-05,
+      "loss": 1.0058,
+      "step": 15296
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.345819009777223e-05,
+      "loss": 1.0391,
+      "step": 15297
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.344926112772892e-05,
+      "loss": 0.9344,
+      "step": 15298
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.344033215768561e-05,
+      "loss": 1.0516,
+      "step": 15299
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.343140318764231e-05,
+      "loss": 0.9977,
+      "step": 15300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3422474217599e-05,
+      "loss": 0.9925,
+      "step": 15301
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.34135452475557e-05,
+      "loss": 0.9487,
+      "step": 15302
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.34046162775124e-05,
+      "loss": 1.1114,
+      "step": 15303
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.339568730746909e-05,
+      "loss": 1.0525,
+      "step": 15304
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.338675833742578e-05,
+      "loss": 1.0228,
+      "step": 15305
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.337782936738248e-05,
+      "loss": 1.1914,
+      "step": 15306
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.336890039733917e-05,
+      "loss": 0.8871,
+      "step": 15307
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.335997142729586e-05,
+      "loss": 1.1557,
+      "step": 15308
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.335104245725257e-05,
+      "loss": 1.0168,
+      "step": 15309
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.334211348720926e-05,
+      "loss": 1.033,
+      "step": 15310
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.333318451716595e-05,
+      "loss": 0.9918,
+      "step": 15311
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.332425554712264e-05,
+      "loss": 1.1339,
+      "step": 15312
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.331532657707934e-05,
+      "loss": 1.0563,
+      "step": 15313
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.330639760703603e-05,
+      "loss": 0.9535,
+      "step": 15314
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.329746863699272e-05,
+      "loss": 1.0078,
+      "step": 15315
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.328853966694941e-05,
+      "loss": 1.0366,
+      "step": 15316
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.32796106969061e-05,
+      "loss": 1.152,
+      "step": 15317
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.327068172686281e-05,
+      "loss": 1.0084,
+      "step": 15318
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.32617527568195e-05,
+      "loss": 0.9898,
+      "step": 15319
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.325282378677619e-05,
+      "loss": 1.1149,
+      "step": 15320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.32438948167329e-05,
+      "loss": 1.0543,
+      "step": 15321
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.323496584668958e-05,
+      "loss": 0.9374,
+      "step": 15322
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.322603687664628e-05,
+      "loss": 1.1099,
+      "step": 15323
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.321710790660298e-05,
+      "loss": 1.0056,
+      "step": 15324
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.320817893655967e-05,
+      "loss": 0.8961,
+      "step": 15325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.319924996651636e-05,
+      "loss": 0.8273,
+      "step": 15326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.319032099647306e-05,
+      "loss": 1.0191,
+      "step": 15327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.318139202642975e-05,
+      "loss": 1.1085,
+      "step": 15328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.317246305638645e-05,
+      "loss": 0.9804,
+      "step": 15329
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.316353408634315e-05,
+      "loss": 1.1116,
+      "step": 15330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.315460511629984e-05,
+      "loss": 1.0616,
+      "step": 15331
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.314567614625653e-05,
+      "loss": 0.9363,
+      "step": 15332
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.313674717621323e-05,
+      "loss": 1.077,
+      "step": 15333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.312781820616992e-05,
+      "loss": 0.9671,
+      "step": 15334
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.311888923612662e-05,
+      "loss": 1.1966,
+      "step": 15335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.31099602660833e-05,
+      "loss": 1.0406,
+      "step": 15336
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.310103129604e-05,
+      "loss": 1.0642,
+      "step": 15337
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.30921023259967e-05,
+      "loss": 1.0904,
+      "step": 15338
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.308317335595339e-05,
+      "loss": 1.122,
+      "step": 15339
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.307424438591008e-05,
+      "loss": 1.0432,
+      "step": 15340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.306531541586679e-05,
+      "loss": 1.1584,
+      "step": 15341
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.305638644582348e-05,
+      "loss": 0.9322,
+      "step": 15342
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.304745747578017e-05,
+      "loss": 1.0128,
+      "step": 15343
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.303852850573687e-05,
+      "loss": 1.0953,
+      "step": 15344
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.302959953569356e-05,
+      "loss": 1.1549,
+      "step": 15345
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.302067056565025e-05,
+      "loss": 0.9852,
+      "step": 15346
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.301174159560696e-05,
+      "loss": 0.9887,
+      "step": 15347
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.300281262556365e-05,
+      "loss": 0.988,
+      "step": 15348
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.299388365552034e-05,
+      "loss": 1.016,
+      "step": 15349
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.298495468547704e-05,
+      "loss": 1.0371,
+      "step": 15350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.297602571543373e-05,
+      "loss": 0.9777,
+      "step": 15351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.296709674539042e-05,
+      "loss": 1.1534,
+      "step": 15352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.295816777534713e-05,
+      "loss": 1.0547,
+      "step": 15353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.294923880530382e-05,
+      "loss": 1.0215,
+      "step": 15354
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.29403098352605e-05,
+      "loss": 0.9953,
+      "step": 15355
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.293138086521721e-05,
+      "loss": 1.0577,
+      "step": 15356
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.29224518951739e-05,
+      "loss": 0.9652,
+      "step": 15357
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.291352292513059e-05,
+      "loss": 0.9965,
+      "step": 15358
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.290459395508728e-05,
+      "loss": 0.9003,
+      "step": 15359
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.289566498504397e-05,
+      "loss": 0.9458,
+      "step": 15360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.288673601500066e-05,
+      "loss": 1.0342,
+      "step": 15361
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.287780704495737e-05,
+      "loss": 0.9045,
+      "step": 15362
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.286887807491406e-05,
+      "loss": 0.9795,
+      "step": 15363
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.285994910487075e-05,
+      "loss": 1.0081,
+      "step": 15364
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.285102013482745e-05,
+      "loss": 1.0034,
+      "step": 15365
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.284209116478414e-05,
+      "loss": 1.0812,
+      "step": 15366
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.283316219474083e-05,
+      "loss": 1.0177,
+      "step": 15367
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.282423322469754e-05,
+      "loss": 1.0708,
+      "step": 15368
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.281530425465423e-05,
+      "loss": 1.1618,
+      "step": 15369
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.280637528461092e-05,
+      "loss": 0.964,
+      "step": 15370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.279744631456762e-05,
+      "loss": 0.9786,
+      "step": 15371
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.278851734452431e-05,
+      "loss": 1.0801,
+      "step": 15372
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.2779588374481e-05,
+      "loss": 1.1264,
+      "step": 15373
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.27706594044377e-05,
+      "loss": 1.0789,
+      "step": 15374
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.27617304343944e-05,
+      "loss": 1.1342,
+      "step": 15375
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.275280146435109e-05,
+      "loss": 0.9838,
+      "step": 15376
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.274387249430779e-05,
+      "loss": 1.0543,
+      "step": 15377
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.273494352426448e-05,
+      "loss": 0.9423,
+      "step": 15378
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.272601455422117e-05,
+      "loss": 0.9368,
+      "step": 15379
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.271708558417788e-05,
+      "loss": 1.0034,
+      "step": 15380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.270815661413457e-05,
+      "loss": 1.0133,
+      "step": 15381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.269922764409126e-05,
+      "loss": 0.8961,
+      "step": 15382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.269029867404795e-05,
+      "loss": 0.954,
+      "step": 15383
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.268136970400464e-05,
+      "loss": 0.9916,
+      "step": 15384
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.267244073396133e-05,
+      "loss": 0.9073,
+      "step": 15385
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.266351176391803e-05,
+      "loss": 1.0102,
+      "step": 15386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.265458279387472e-05,
+      "loss": 1.0414,
+      "step": 15387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.264565382383143e-05,
+      "loss": 0.9702,
+      "step": 15388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.263672485378812e-05,
+      "loss": 0.9099,
+      "step": 15389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.262779588374481e-05,
+      "loss": 0.9942,
+      "step": 15390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.261886691370151e-05,
+      "loss": 1.1101,
+      "step": 15391
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.26099379436582e-05,
+      "loss": 1.0867,
+      "step": 15392
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.26010089736149e-05,
+      "loss": 0.8468,
+      "step": 15393
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.25920800035716e-05,
+      "loss": 1.0849,
+      "step": 15394
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.258315103352829e-05,
+      "loss": 1.0306,
+      "step": 15395
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.257422206348498e-05,
+      "loss": 0.9496,
+      "step": 15396
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.256529309344168e-05,
+      "loss": 1.0737,
+      "step": 15397
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.255636412339837e-05,
+      "loss": 1.0628,
+      "step": 15398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.254743515335506e-05,
+      "loss": 0.9898,
+      "step": 15399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.253850618331177e-05,
+      "loss": 0.957,
+      "step": 15400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.252957721326846e-05,
+      "loss": 0.8801,
+      "step": 15401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.252064824322515e-05,
+      "loss": 0.9978,
+      "step": 15402
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.251171927318184e-05,
+      "loss": 1.036,
+      "step": 15403
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.250279030313854e-05,
+      "loss": 0.9978,
+      "step": 15404
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.249386133309523e-05,
+      "loss": 1.0811,
+      "step": 15405
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.248493236305192e-05,
+      "loss": 0.9872,
+      "step": 15406
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.247600339300861e-05,
+      "loss": 1.0989,
+      "step": 15407
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.24670744229653e-05,
+      "loss": 1.0356,
+      "step": 15408
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.245814545292201e-05,
+      "loss": 1.0642,
+      "step": 15409
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.24492164828787e-05,
+      "loss": 0.9842,
+      "step": 15410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.244028751283539e-05,
+      "loss": 1.0399,
+      "step": 15411
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.24313585427921e-05,
+      "loss": 1.141,
+      "step": 15412
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.242242957274878e-05,
+      "loss": 0.9148,
+      "step": 15413
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.241350060270548e-05,
+      "loss": 1.0616,
+      "step": 15414
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.240457163266218e-05,
+      "loss": 1.0567,
+      "step": 15415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.239564266261887e-05,
+      "loss": 1.1045,
+      "step": 15416
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.238671369257556e-05,
+      "loss": 1.0604,
+      "step": 15417
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.237778472253226e-05,
+      "loss": 0.9737,
+      "step": 15418
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.236885575248895e-05,
+      "loss": 1.6109,
+      "step": 15419
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.235992678244565e-05,
+      "loss": 1.0091,
+      "step": 15420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.235099781240235e-05,
+      "loss": 1.0714,
+      "step": 15421
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.234206884235904e-05,
+      "loss": 1.1297,
+      "step": 15422
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.233313987231573e-05,
+      "loss": 0.8754,
+      "step": 15423
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.232421090227243e-05,
+      "loss": 1.1535,
+      "step": 15424
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.231528193222912e-05,
+      "loss": 1.3063,
+      "step": 15425
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.230635296218582e-05,
+      "loss": 0.9921,
+      "step": 15426
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.22974239921425e-05,
+      "loss": 0.9477,
+      "step": 15427
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.22884950220992e-05,
+      "loss": 1.1875,
+      "step": 15428
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.227956605205589e-05,
+      "loss": 0.9422,
+      "step": 15429
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.227063708201259e-05,
+      "loss": 1.0994,
+      "step": 15430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.226170811196928e-05,
+      "loss": 1.0512,
+      "step": 15431
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.225277914192597e-05,
+      "loss": 0.9691,
+      "step": 15432
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.224385017188268e-05,
+      "loss": 1.1316,
+      "step": 15433
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.223492120183937e-05,
+      "loss": 1.1114,
+      "step": 15434
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.222599223179607e-05,
+      "loss": 0.9559,
+      "step": 15435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.221706326175276e-05,
+      "loss": 0.9578,
+      "step": 15436
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.220813429170945e-05,
+      "loss": 1.0744,
+      "step": 15437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.219920532166616e-05,
+      "loss": 0.977,
+      "step": 15438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.219027635162285e-05,
+      "loss": 0.8777,
+      "step": 15439
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.218134738157954e-05,
+      "loss": 1.0808,
+      "step": 15440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.217241841153624e-05,
+      "loss": 1.0364,
+      "step": 15441
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.216348944149293e-05,
+      "loss": 1.0462,
+      "step": 15442
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.215456047144962e-05,
+      "loss": 1.1329,
+      "step": 15443
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.214563150140632e-05,
+      "loss": 1.0479,
+      "step": 15444
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.213670253136302e-05,
+      "loss": 1.0047,
+      "step": 15445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.21277735613197e-05,
+      "loss": 0.9141,
+      "step": 15446
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.211884459127641e-05,
+      "loss": 1.2218,
+      "step": 15447
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.21099156212331e-05,
+      "loss": 1.1077,
+      "step": 15448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.210098665118979e-05,
+      "loss": 0.9965,
+      "step": 15449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.209205768114648e-05,
+      "loss": 0.8766,
+      "step": 15450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.208312871110317e-05,
+      "loss": 1.0575,
+      "step": 15451
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.207419974105986e-05,
+      "loss": 0.9644,
+      "step": 15452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.206527077101657e-05,
+      "loss": 1.065,
+      "step": 15453
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.205634180097326e-05,
+      "loss": 1.0498,
+      "step": 15454
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.204741283092995e-05,
+      "loss": 1.0174,
+      "step": 15455
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.203848386088665e-05,
+      "loss": 0.9693,
+      "step": 15456
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.202955489084334e-05,
+      "loss": 1.0735,
+      "step": 15457
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.202062592080003e-05,
+      "loss": 0.9954,
+      "step": 15458
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.201169695075674e-05,
+      "loss": 0.963,
+      "step": 15459
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.200276798071343e-05,
+      "loss": 1.0425,
+      "step": 15460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.199383901067012e-05,
+      "loss": 1.0428,
+      "step": 15461
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.198491004062682e-05,
+      "loss": 1.0106,
+      "step": 15462
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.197598107058351e-05,
+      "loss": 0.9825,
+      "step": 15463
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.19670521005402e-05,
+      "loss": 1.2119,
+      "step": 15464
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.19581231304969e-05,
+      "loss": 0.8986,
+      "step": 15465
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.19491941604536e-05,
+      "loss": 1.0077,
+      "step": 15466
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.194026519041029e-05,
+      "loss": 0.9995,
+      "step": 15467
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.193133622036699e-05,
+      "loss": 0.9145,
+      "step": 15468
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.192240725032368e-05,
+      "loss": 0.9641,
+      "step": 15469
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.191347828028037e-05,
+      "loss": 0.9607,
+      "step": 15470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.190454931023708e-05,
+      "loss": 1.0319,
+      "step": 15471
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.189562034019377e-05,
+      "loss": 0.9368,
+      "step": 15472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.188669137015046e-05,
+      "loss": 1.0857,
+      "step": 15473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.187776240010715e-05,
+      "loss": 1.1173,
+      "step": 15474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.186883343006384e-05,
+      "loss": 0.9419,
+      "step": 15475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.185990446002053e-05,
+      "loss": 1.066,
+      "step": 15476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.185097548997723e-05,
+      "loss": 1.2018,
+      "step": 15477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.184204651993392e-05,
+      "loss": 0.9989,
+      "step": 15478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.183311754989061e-05,
+      "loss": 0.9563,
+      "step": 15479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.182418857984732e-05,
+      "loss": 1.0249,
+      "step": 15480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.181525960980401e-05,
+      "loss": 1.019,
+      "step": 15481
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.18063306397607e-05,
+      "loss": 1.105,
+      "step": 15482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.17974016697174e-05,
+      "loss": 1.0451,
+      "step": 15483
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.17884726996741e-05,
+      "loss": 1.1129,
+      "step": 15484
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.17795437296308e-05,
+      "loss": 1.046,
+      "step": 15485
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.177061475958749e-05,
+      "loss": 0.9897,
+      "step": 15486
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.176168578954418e-05,
+      "loss": 1.0396,
+      "step": 15487
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.175275681950088e-05,
+      "loss": 0.9655,
+      "step": 15488
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.174382784945757e-05,
+      "loss": 1.0335,
+      "step": 15489
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.173489887941426e-05,
+      "loss": 1.0696,
+      "step": 15490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.172596990937097e-05,
+      "loss": 0.9855,
+      "step": 15491
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.171704093932766e-05,
+      "loss": 1.0951,
+      "step": 15492
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.170811196928435e-05,
+      "loss": 0.9781,
+      "step": 15493
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.169918299924104e-05,
+      "loss": 1.0195,
+      "step": 15494
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.169025402919773e-05,
+      "loss": 1.0226,
+      "step": 15495
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.168132505915443e-05,
+      "loss": 1.0155,
+      "step": 15496
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.167239608911112e-05,
+      "loss": 1.1272,
+      "step": 15497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.166346711906781e-05,
+      "loss": 1.1616,
+      "step": 15498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.16545381490245e-05,
+      "loss": 1.0716,
+      "step": 15499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.164560917898121e-05,
+      "loss": 1.0547,
+      "step": 15500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.16366802089379e-05,
+      "loss": 1.0243,
+      "step": 15501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.162775123889459e-05,
+      "loss": 1.0204,
+      "step": 15502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.16188222688513e-05,
+      "loss": 1.0776,
+      "step": 15503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.160989329880798e-05,
+      "loss": 0.9261,
+      "step": 15504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.160096432876467e-05,
+      "loss": 1.0314,
+      "step": 15505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.159203535872138e-05,
+      "loss": 1.1718,
+      "step": 15506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.158310638867807e-05,
+      "loss": 1.0374,
+      "step": 15507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.157417741863476e-05,
+      "loss": 0.9575,
+      "step": 15508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.156524844859146e-05,
+      "loss": 1.1192,
+      "step": 15509
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.155631947854815e-05,
+      "loss": 0.9331,
+      "step": 15510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.154739050850484e-05,
+      "loss": 1.0349,
+      "step": 15511
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 1.0374,
+      "step": 15512
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.152953256841824e-05,
+      "loss": 0.9698,
+      "step": 15513
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.152060359837493e-05,
+      "loss": 0.9347,
+      "step": 15514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.151167462833163e-05,
+      "loss": 0.9893,
+      "step": 15515
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.150274565828832e-05,
+      "loss": 1.0362,
+      "step": 15516
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.149381668824501e-05,
+      "loss": 1.1191,
+      "step": 15517
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.14848877182017e-05,
+      "loss": 1.0154,
+      "step": 15518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.14759587481584e-05,
+      "loss": 1.0827,
+      "step": 15519
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.146702977811509e-05,
+      "loss": 1.1136,
+      "step": 15520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.145810080807179e-05,
+      "loss": 1.0323,
+      "step": 15521
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.144917183802848e-05,
+      "loss": 0.9155,
+      "step": 15522
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.144024286798517e-05,
+      "loss": 1.0829,
+      "step": 15523
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.143131389794188e-05,
+      "loss": 0.9953,
+      "step": 15524
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.142238492789857e-05,
+      "loss": 0.9477,
+      "step": 15525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.141345595785526e-05,
+      "loss": 0.9878,
+      "step": 15526
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.140452698781196e-05,
+      "loss": 0.9443,
+      "step": 15527
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.139559801776865e-05,
+      "loss": 1.0598,
+      "step": 15528
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.138666904772534e-05,
+      "loss": 1.0761,
+      "step": 15529
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.137774007768205e-05,
+      "loss": 0.9515,
+      "step": 15530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.136881110763874e-05,
+      "loss": 1.1538,
+      "step": 15531
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.135988213759543e-05,
+      "loss": 0.976,
+      "step": 15532
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.135095316755213e-05,
+      "loss": 1.0114,
+      "step": 15533
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.134202419750882e-05,
+      "loss": 1.1883,
+      "step": 15534
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.133309522746552e-05,
+      "loss": 1.0626,
+      "step": 15535
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.132416625742222e-05,
+      "loss": 1.0667,
+      "step": 15536
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.13152372873789e-05,
+      "loss": 1.1279,
+      "step": 15537
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.130630831733561e-05,
+      "loss": 0.9305,
+      "step": 15538
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.12973793472923e-05,
+      "loss": 1.1185,
+      "step": 15539
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.128845037724899e-05,
+      "loss": 1.0407,
+      "step": 15540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.127952140720568e-05,
+      "loss": 1.0604,
+      "step": 15541
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.127059243716237e-05,
+      "loss": 1.0061,
+      "step": 15542
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.126166346711906e-05,
+      "loss": 1.1694,
+      "step": 15543
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.125273449707577e-05,
+      "loss": 1.0544,
+      "step": 15544
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.124380552703246e-05,
+      "loss": 1.0337,
+      "step": 15545
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.123487655698915e-05,
+      "loss": 0.8635,
+      "step": 15546
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.122594758694585e-05,
+      "loss": 1.0436,
+      "step": 15547
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.121701861690254e-05,
+      "loss": 1.189,
+      "step": 15548
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.120808964685923e-05,
+      "loss": 0.9908,
+      "step": 15549
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.119916067681594e-05,
+      "loss": 1.1609,
+      "step": 15550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.119023170677263e-05,
+      "loss": 0.9876,
+      "step": 15551
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.118130273672932e-05,
+      "loss": 0.9759,
+      "step": 15552
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.117237376668602e-05,
+      "loss": 0.9645,
+      "step": 15553
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.116344479664271e-05,
+      "loss": 0.9316,
+      "step": 15554
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.11545158265994e-05,
+      "loss": 0.9844,
+      "step": 15555
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.11455868565561e-05,
+      "loss": 0.988,
+      "step": 15556
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.11366578865128e-05,
+      "loss": 0.9895,
+      "step": 15557
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.112772891646949e-05,
+      "loss": 1.1001,
+      "step": 15558
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.111879994642619e-05,
+      "loss": 0.9941,
+      "step": 15559
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.110987097638288e-05,
+      "loss": 1.0727,
+      "step": 15560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.110094200633957e-05,
+      "loss": 1.0015,
+      "step": 15561
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.109201303629628e-05,
+      "loss": 1.0378,
+      "step": 15562
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.108308406625297e-05,
+      "loss": 1.0611,
+      "step": 15563
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.107415509620966e-05,
+      "loss": 1.0269,
+      "step": 15564
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.106522612616635e-05,
+      "loss": 0.8388,
+      "step": 15565
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.105629715612304e-05,
+      "loss": 1.029,
+      "step": 15566
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.104736818607973e-05,
+      "loss": 1.0035,
+      "step": 15567
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.103843921603643e-05,
+      "loss": 0.9436,
+      "step": 15568
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.102951024599313e-05,
+      "loss": 1.0569,
+      "step": 15569
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.102058127594982e-05,
+      "loss": 0.9181,
+      "step": 15570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.101165230590652e-05,
+      "loss": 1.0231,
+      "step": 15571
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.100272333586321e-05,
+      "loss": 0.9908,
+      "step": 15572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.09937943658199e-05,
+      "loss": 0.9081,
+      "step": 15573
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.09848653957766e-05,
+      "loss": 0.9805,
+      "step": 15574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.097593642573329e-05,
+      "loss": 1.0098,
+      "step": 15575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0967007455689983e-05,
+      "loss": 1.0298,
+      "step": 15576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.095807848564669e-05,
+      "loss": 1.0529,
+      "step": 15577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.094914951560338e-05,
+      "loss": 0.9461,
+      "step": 15578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.094022054556007e-05,
+      "loss": 1.0253,
+      "step": 15579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0931291575516766e-05,
+      "loss": 0.9943,
+      "step": 15580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.092236260547346e-05,
+      "loss": 1.0409,
+      "step": 15581
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.091343363543016e-05,
+      "loss": 1.0994,
+      "step": 15582
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.090450466538685e-05,
+      "loss": 1.0981,
+      "step": 15583
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.089557569534354e-05,
+      "loss": 1.0059,
+      "step": 15584
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0886646725300245e-05,
+      "loss": 1.1013,
+      "step": 15585
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0877717755256936e-05,
+      "loss": 1.111,
+      "step": 15586
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0868788785213626e-05,
+      "loss": 1.0904,
+      "step": 15587
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.085985981517033e-05,
+      "loss": 1.1166,
+      "step": 15588
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.085093084512702e-05,
+      "loss": 1.1088,
+      "step": 15589
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.084200187508371e-05,
+      "loss": 1.0455,
+      "step": 15590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.083307290504041e-05,
+      "loss": 1.1387,
+      "step": 15591
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.08241439349971e-05,
+      "loss": 1.0609,
+      "step": 15592
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0815214964953796e-05,
+      "loss": 1.1312,
+      "step": 15593
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.080628599491049e-05,
+      "loss": 1.0358,
+      "step": 15594
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0797357024867184e-05,
+      "loss": 1.0377,
+      "step": 15595
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0788428054823874e-05,
+      "loss": 1.0202,
+      "step": 15596
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.077949908478058e-05,
+      "loss": 0.9626,
+      "step": 15597
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.077057011473727e-05,
+      "loss": 1.0729,
+      "step": 15598
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.076164114469396e-05,
+      "loss": 1.2222,
+      "step": 15599
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.075271217465066e-05,
+      "loss": 0.9864,
+      "step": 15600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0743783204607354e-05,
+      "loss": 0.958,
+      "step": 15601
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0734854234564044e-05,
+      "loss": 1.0042,
+      "step": 15602
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.072592526452074e-05,
+      "loss": 0.9739,
+      "step": 15603
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.071699629447743e-05,
+      "loss": 1.0141,
+      "step": 15604
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.070806732443412e-05,
+      "loss": 0.9843,
+      "step": 15605
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0699138354390827e-05,
+      "loss": 1.1065,
+      "step": 15606
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.069020938434752e-05,
+      "loss": 1.0449,
+      "step": 15607
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.068128041430421e-05,
+      "loss": 0.9873,
+      "step": 15608
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.067235144426091e-05,
+      "loss": 1.0217,
+      "step": 15609
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.06634224742176e-05,
+      "loss": 1.0733,
+      "step": 15610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.065449350417429e-05,
+      "loss": 1.0648,
+      "step": 15611
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0645564534130997e-05,
+      "loss": 0.9284,
+      "step": 15612
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.063663556408769e-05,
+      "loss": 1.0092,
+      "step": 15613
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.062770659404438e-05,
+      "loss": 1.0239,
+      "step": 15614
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0618777624001075e-05,
+      "loss": 1.08,
+      "step": 15615
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0609848653957765e-05,
+      "loss": 1.2793,
+      "step": 15616
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0600919683914456e-05,
+      "loss": 1.0116,
+      "step": 15617
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.059199071387116e-05,
+      "loss": 1.0821,
+      "step": 15618
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.058306174382785e-05,
+      "loss": 1.0231,
+      "step": 15619
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.057413277378454e-05,
+      "loss": 0.9715,
+      "step": 15620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0565203803741245e-05,
+      "loss": 0.9884,
+      "step": 15621
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0556274833697935e-05,
+      "loss": 1.0223,
+      "step": 15622
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0547345863654626e-05,
+      "loss": 1.0493,
+      "step": 15623
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.053841689361133e-05,
+      "loss": 1.0904,
+      "step": 15624
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.052948792356802e-05,
+      "loss": 1.0014,
+      "step": 15625
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.052055895352471e-05,
+      "loss": 1.1309,
+      "step": 15626
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.051162998348141e-05,
+      "loss": 1.0362,
+      "step": 15627
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.05027010134381e-05,
+      "loss": 0.8984,
+      "step": 15628
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.049377204339479e-05,
+      "loss": 1.0313,
+      "step": 15629
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.048484307335149e-05,
+      "loss": 1.0211,
+      "step": 15630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0475914103308183e-05,
+      "loss": 0.9851,
+      "step": 15631
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.046698513326489e-05,
+      "loss": 1.1184,
+      "step": 15632
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.045805616322158e-05,
+      "loss": 0.9796,
+      "step": 15633
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.044912719317827e-05,
+      "loss": 0.9282,
+      "step": 15634
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.044019822313497e-05,
+      "loss": 0.9281,
+      "step": 15635
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.043126925309166e-05,
+      "loss": 1.1862,
+      "step": 15636
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0422340283048353e-05,
+      "loss": 1.0293,
+      "step": 15637
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.041341131300505e-05,
+      "loss": 1.1401,
+      "step": 15638
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.040448234296174e-05,
+      "loss": 0.9905,
+      "step": 15639
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.039555337291843e-05,
+      "loss": 1.0205,
+      "step": 15640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0386624402875136e-05,
+      "loss": 0.9722,
+      "step": 15641
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0377695432831826e-05,
+      "loss": 1.0172,
+      "step": 15642
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.036876646278852e-05,
+      "loss": 1.085,
+      "step": 15643
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.035983749274522e-05,
+      "loss": 1.1074,
+      "step": 15644
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.035090852270191e-05,
+      "loss": 0.8784,
+      "step": 15645
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.03419795526586e-05,
+      "loss": 0.9899,
+      "step": 15646
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0333050582615306e-05,
+      "loss": 1.1047,
+      "step": 15647
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0324121612571996e-05,
+      "loss": 0.9967,
+      "step": 15648
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.031519264252869e-05,
+      "loss": 1.0343,
+      "step": 15649
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0306263672485384e-05,
+      "loss": 1.1288,
+      "step": 15650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0297334702442074e-05,
+      "loss": 1.1165,
+      "step": 15651
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0288405732398765e-05,
+      "loss": 0.9606,
+      "step": 15652
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.027947676235547e-05,
+      "loss": 1.0584,
+      "step": 15653
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.027054779231216e-05,
+      "loss": 1.0052,
+      "step": 15654
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.026161882226885e-05,
+      "loss": 1.0224,
+      "step": 15655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0252689852225554e-05,
+      "loss": 1.0379,
+      "step": 15656
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0243760882182244e-05,
+      "loss": 1.1135,
+      "step": 15657
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0234831912138935e-05,
+      "loss": 1.1055,
+      "step": 15658
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.022590294209564e-05,
+      "loss": 0.9524,
+      "step": 15659
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.021697397205233e-05,
+      "loss": 0.9642,
+      "step": 15660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.020804500200902e-05,
+      "loss": 1.1721,
+      "step": 15661
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.019911603196572e-05,
+      "loss": 1.1018,
+      "step": 15662
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.019018706192241e-05,
+      "loss": 1.0296,
+      "step": 15663
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.01812580918791e-05,
+      "loss": 1.1716,
+      "step": 15664
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.01723291218358e-05,
+      "loss": 1.0436,
+      "step": 15665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.016340015179249e-05,
+      "loss": 1.0269,
+      "step": 15666
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.015447118174918e-05,
+      "loss": 0.9835,
+      "step": 15667
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.014554221170589e-05,
+      "loss": 1.0448,
+      "step": 15668
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.013661324166258e-05,
+      "loss": 1.0201,
+      "step": 15669
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.012768427161927e-05,
+      "loss": 0.9827,
+      "step": 15670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0118755301575965e-05,
+      "loss": 1.1138,
+      "step": 15671
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.010982633153266e-05,
+      "loss": 1.0626,
+      "step": 15672
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.010089736148935e-05,
+      "loss": 0.8201,
+      "step": 15673
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.009196839144605e-05,
+      "loss": 1.0125,
+      "step": 15674
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.008303942140274e-05,
+      "loss": 0.8967,
+      "step": 15675
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.007411045135943e-05,
+      "loss": 1.0566,
+      "step": 15676
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0065181481316135e-05,
+      "loss": 1.0354,
+      "step": 15677
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0056252511272826e-05,
+      "loss": 0.9717,
+      "step": 15678
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0047323541229516e-05,
+      "loss": 1.0536,
+      "step": 15679
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.003839457118622e-05,
+      "loss": 1.0043,
+      "step": 15680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.002946560114291e-05,
+      "loss": 1.03,
+      "step": 15681
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.002053663109961e-05,
+      "loss": 1.0421,
+      "step": 15682
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.00116076610563e-05,
+      "loss": 1.0192,
+      "step": 15683
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.000267869101299e-05,
+      "loss": 1.1309,
+      "step": 15684
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.999374972096969e-05,
+      "loss": 1.1198,
+      "step": 15685
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9984820750926383e-05,
+      "loss": 0.8923,
+      "step": 15686
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9975891780883074e-05,
+      "loss": 1.0293,
+      "step": 15687
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.996696281083978e-05,
+      "loss": 1.0503,
+      "step": 15688
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.995803384079647e-05,
+      "loss": 0.9861,
+      "step": 15689
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.994910487075316e-05,
+      "loss": 0.9666,
+      "step": 15690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.994017590070986e-05,
+      "loss": 0.9317,
+      "step": 15691
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9931246930666553e-05,
+      "loss": 0.8729,
+      "step": 15692
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9922317960623244e-05,
+      "loss": 1.0296,
+      "step": 15693
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.991338899057994e-05,
+      "loss": 1.03,
+      "step": 15694
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.990446002053663e-05,
+      "loss": 1.1616,
+      "step": 15695
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.989553105049332e-05,
+      "loss": 0.9448,
+      "step": 15696
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9886602080450026e-05,
+      "loss": 0.9819,
+      "step": 15697
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.987767311040672e-05,
+      "loss": 1.0602,
+      "step": 15698
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.986874414036341e-05,
+      "loss": 1.0131,
+      "step": 15699
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.985981517032011e-05,
+      "loss": 0.9887,
+      "step": 15700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.98508862002768e-05,
+      "loss": 1.0166,
+      "step": 15701
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.984195723023349e-05,
+      "loss": 1.0002,
+      "step": 15702
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9833028260190196e-05,
+      "loss": 1.1624,
+      "step": 15703
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.982409929014689e-05,
+      "loss": 0.8822,
+      "step": 15704
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.981517032010358e-05,
+      "loss": 0.9741,
+      "step": 15705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9806241350060274e-05,
+      "loss": 0.9613,
+      "step": 15706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9797312380016965e-05,
+      "loss": 0.9845,
+      "step": 15707
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9788383409973655e-05,
+      "loss": 1.0518,
+      "step": 15708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.977945443993036e-05,
+      "loss": 1.0593,
+      "step": 15709
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.977052546988705e-05,
+      "loss": 1.1106,
+      "step": 15710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.976159649984374e-05,
+      "loss": 1.0447,
+      "step": 15711
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9752667529800444e-05,
+      "loss": 0.9819,
+      "step": 15712
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9743738559757135e-05,
+      "loss": 0.9649,
+      "step": 15713
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9734809589713825e-05,
+      "loss": 1.0084,
+      "step": 15714
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.972588061967053e-05,
+      "loss": 1.0827,
+      "step": 15715
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.971695164962722e-05,
+      "loss": 0.9794,
+      "step": 15716
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.970802267958391e-05,
+      "loss": 1.0006,
+      "step": 15717
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.969909370954061e-05,
+      "loss": 1.0971,
+      "step": 15718
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.96901647394973e-05,
+      "loss": 1.0027,
+      "step": 15719
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.968123576945399e-05,
+      "loss": 1.0388,
+      "step": 15720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.967230679941069e-05,
+      "loss": 1.0241,
+      "step": 15721
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.966337782936738e-05,
+      "loss": 1.0476,
+      "step": 15722
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9654448859324074e-05,
+      "loss": 1.0523,
+      "step": 15723
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.964551988928078e-05,
+      "loss": 1.0405,
+      "step": 15724
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.963659091923747e-05,
+      "loss": 1.0659,
+      "step": 15725
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.962766194919416e-05,
+      "loss": 0.9722,
+      "step": 15726
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.961873297915086e-05,
+      "loss": 0.9821,
+      "step": 15727
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.960980400910755e-05,
+      "loss": 1.1284,
+      "step": 15728
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.960087503906425e-05,
+      "loss": 1.0459,
+      "step": 15729
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.959194606902094e-05,
+      "loss": 1.0459,
+      "step": 15730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.958301709897763e-05,
+      "loss": 1.2194,
+      "step": 15731
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9574088128934335e-05,
+      "loss": 1.1827,
+      "step": 15732
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9565159158891026e-05,
+      "loss": 0.9701,
+      "step": 15733
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9556230188847716e-05,
+      "loss": 1.0371,
+      "step": 15734
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.954730121880442e-05,
+      "loss": 0.9691,
+      "step": 15735
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.953837224876111e-05,
+      "loss": 0.9489,
+      "step": 15736
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.95294432787178e-05,
+      "loss": 0.9959,
+      "step": 15737
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9520514308674505e-05,
+      "loss": 1.0947,
+      "step": 15738
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9511585338631196e-05,
+      "loss": 1.0152,
+      "step": 15739
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9502656368587886e-05,
+      "loss": 1.1172,
+      "step": 15740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9493727398544584e-05,
+      "loss": 1.1663,
+      "step": 15741
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9484798428501274e-05,
+      "loss": 1.0857,
+      "step": 15742
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9475869458457964e-05,
+      "loss": 0.9926,
+      "step": 15743
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.946694048841467e-05,
+      "loss": 0.9353,
+      "step": 15744
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.945801151837136e-05,
+      "loss": 1.0671,
+      "step": 15745
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.944908254832805e-05,
+      "loss": 1.0258,
+      "step": 15746
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9440153578284753e-05,
+      "loss": 1.0566,
+      "step": 15747
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9431224608241444e-05,
+      "loss": 0.9675,
+      "step": 15748
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9422295638198134e-05,
+      "loss": 1.116,
+      "step": 15749
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.941336666815484e-05,
+      "loss": 0.9994,
+      "step": 15750
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.940443769811153e-05,
+      "loss": 0.9563,
+      "step": 15751
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.939550872806822e-05,
+      "loss": 1.0305,
+      "step": 15752
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.938657975802492e-05,
+      "loss": 0.9811,
+      "step": 15753
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.937765078798161e-05,
+      "loss": 1.0428,
+      "step": 15754
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.93687218179383e-05,
+      "loss": 0.9784,
+      "step": 15755
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9359792847895e-05,
+      "loss": 0.9829,
+      "step": 15756
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.935086387785169e-05,
+      "loss": 1.1188,
+      "step": 15757
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.934193490780838e-05,
+      "loss": 1.0165,
+      "step": 15758
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.933300593776509e-05,
+      "loss": 1.1889,
+      "step": 15759
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.932407696772178e-05,
+      "loss": 1.0994,
+      "step": 15760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.931514799767847e-05,
+      "loss": 0.9638,
+      "step": 15761
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9306219027635165e-05,
+      "loss": 1.1253,
+      "step": 15762
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.929729005759186e-05,
+      "loss": 0.9867,
+      "step": 15763
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.928836108754855e-05,
+      "loss": 1.1021,
+      "step": 15764
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.927943211750525e-05,
+      "loss": 0.9949,
+      "step": 15765
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.927050314746194e-05,
+      "loss": 1.0624,
+      "step": 15766
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.926157417741863e-05,
+      "loss": 0.9839,
+      "step": 15767
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9252645207375335e-05,
+      "loss": 1.041,
+      "step": 15768
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9243716237332025e-05,
+      "loss": 0.9315,
+      "step": 15769
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9234787267288716e-05,
+      "loss": 1.163,
+      "step": 15770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.922585829724542e-05,
+      "loss": 0.9863,
+      "step": 15771
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.921692932720211e-05,
+      "loss": 1.2544,
+      "step": 15772
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.92080003571588e-05,
+      "loss": 0.9296,
+      "step": 15773
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.91990713871155e-05,
+      "loss": 1.0101,
+      "step": 15774
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.919014241707219e-05,
+      "loss": 1.1744,
+      "step": 15775
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9181213447028886e-05,
+      "loss": 0.9541,
+      "step": 15776
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.917228447698558e-05,
+      "loss": 1.071,
+      "step": 15777
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9163355506942274e-05,
+      "loss": 1.0071,
+      "step": 15778
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.915442653689898e-05,
+      "loss": 1.0843,
+      "step": 15779
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.914549756685567e-05,
+      "loss": 1.0012,
+      "step": 15780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.913656859681236e-05,
+      "loss": 0.9976,
+      "step": 15781
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.912763962676906e-05,
+      "loss": 0.9952,
+      "step": 15782
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.911871065672575e-05,
+      "loss": 1.0632,
+      "step": 15783
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9109781686682444e-05,
+      "loss": 0.9365,
+      "step": 15784
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.910085271663914e-05,
+      "loss": 1.0003,
+      "step": 15785
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.909192374659583e-05,
+      "loss": 0.8582,
+      "step": 15786
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.908299477655252e-05,
+      "loss": 0.9768,
+      "step": 15787
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9074065806509226e-05,
+      "loss": 0.9607,
+      "step": 15788
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9065136836465916e-05,
+      "loss": 1.1313,
+      "step": 15789
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.905620786642261e-05,
+      "loss": 1.1397,
+      "step": 15790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.904727889637931e-05,
+      "loss": 0.9271,
+      "step": 15791
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9038349926336e-05,
+      "loss": 1.0237,
+      "step": 15792
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.902942095629269e-05,
+      "loss": 1.0169,
+      "step": 15793
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9020491986249396e-05,
+      "loss": 1.0944,
+      "step": 15794
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9011563016206086e-05,
+      "loss": 1.026,
+      "step": 15795
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.900263404616278e-05,
+      "loss": 0.9936,
+      "step": 15796
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8993705076119474e-05,
+      "loss": 0.9926,
+      "step": 15797
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8984776106076165e-05,
+      "loss": 1.0727,
+      "step": 15798
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8975847136032855e-05,
+      "loss": 1.0459,
+      "step": 15799
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.896691816598956e-05,
+      "loss": 0.9813,
+      "step": 15800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.895798919594625e-05,
+      "loss": 1.1024,
+      "step": 15801
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.894906022590294e-05,
+      "loss": 1.0304,
+      "step": 15802
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8940131255859644e-05,
+      "loss": 1.0197,
+      "step": 15803
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8931202285816334e-05,
+      "loss": 1.0948,
+      "step": 15804
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8922273315773025e-05,
+      "loss": 1.0694,
+      "step": 15805
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.891334434572973e-05,
+      "loss": 1.0675,
+      "step": 15806
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.890441537568642e-05,
+      "loss": 0.9021,
+      "step": 15807
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.889548640564311e-05,
+      "loss": 0.9913,
+      "step": 15808
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.888655743559981e-05,
+      "loss": 1.0235,
+      "step": 15809
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.88776284655565e-05,
+      "loss": 1.2053,
+      "step": 15810
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.886869949551319e-05,
+      "loss": 0.979,
+      "step": 15811
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.885977052546989e-05,
+      "loss": 1.0926,
+      "step": 15812
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.885084155542658e-05,
+      "loss": 1.0061,
+      "step": 15813
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.884191258538327e-05,
+      "loss": 1.0558,
+      "step": 15814
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.883298361533998e-05,
+      "loss": 1.1264,
+      "step": 15815
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.882405464529667e-05,
+      "loss": 1.0898,
+      "step": 15816
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.881512567525336e-05,
+      "loss": 1.0981,
+      "step": 15817
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.880619670521006e-05,
+      "loss": 1.0337,
+      "step": 15818
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.879726773516675e-05,
+      "loss": 1.183,
+      "step": 15819
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.878833876512344e-05,
+      "loss": 0.9398,
+      "step": 15820
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.877940979508014e-05,
+      "loss": 1.0041,
+      "step": 15821
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.877048082503683e-05,
+      "loss": 1.0679,
+      "step": 15822
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.876155185499352e-05,
+      "loss": 1.0545,
+      "step": 15823
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8752622884950225e-05,
+      "loss": 1.0676,
+      "step": 15824
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8743693914906916e-05,
+      "loss": 0.9527,
+      "step": 15825
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8734764944863606e-05,
+      "loss": 0.9504,
+      "step": 15826
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.872583597482031e-05,
+      "loss": 1.0395,
+      "step": 15827
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8716907004777e-05,
+      "loss": 0.9219,
+      "step": 15828
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8707978034733705e-05,
+      "loss": 1.0975,
+      "step": 15829
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8699049064690395e-05,
+      "loss": 0.9127,
+      "step": 15830
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8690120094647086e-05,
+      "loss": 1.0457,
+      "step": 15831
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.868119112460378e-05,
+      "loss": 0.9893,
+      "step": 15832
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8672262154560474e-05,
+      "loss": 1.0251,
+      "step": 15833
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8663333184517164e-05,
+      "loss": 0.9116,
+      "step": 15834
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.865440421447387e-05,
+      "loss": 1.2473,
+      "step": 15835
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.864547524443056e-05,
+      "loss": 1.0988,
+      "step": 15836
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.863654627438725e-05,
+      "loss": 0.9986,
+      "step": 15837
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.862761730434395e-05,
+      "loss": 1.0237,
+      "step": 15838
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8618688334300644e-05,
+      "loss": 1.0987,
+      "step": 15839
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8609759364257334e-05,
+      "loss": 1.0217,
+      "step": 15840
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.860083039421403e-05,
+      "loss": 1.0157,
+      "step": 15841
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.859190142417073e-05,
+      "loss": 1.0267,
+      "step": 15842
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.858297245412742e-05,
+      "loss": 0.9191,
+      "step": 15843
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8574043484084116e-05,
+      "loss": 1.1668,
+      "step": 15844
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.856511451404081e-05,
+      "loss": 1.0417,
+      "step": 15845
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.85561855439975e-05,
+      "loss": 1.0844,
+      "step": 15846
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.85472565739542e-05,
+      "loss": 1.0574,
+      "step": 15847
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.853832760391089e-05,
+      "loss": 1.0759,
+      "step": 15848
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.852939863386758e-05,
+      "loss": 0.9808,
+      "step": 15849
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8520469663824286e-05,
+      "loss": 1.0636,
+      "step": 15850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.851154069378098e-05,
+      "loss": 1.0608,
+      "step": 15851
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.850261172373767e-05,
+      "loss": 0.9605,
+      "step": 15852
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8493682753694365e-05,
+      "loss": 0.9746,
+      "step": 15853
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8484753783651055e-05,
+      "loss": 1.0012,
+      "step": 15854
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.847582481360775e-05,
+      "loss": 0.9553,
+      "step": 15855
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.846689584356445e-05,
+      "loss": 1.1859,
+      "step": 15856
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.845796687352114e-05,
+      "loss": 1.0084,
+      "step": 15857
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.844903790347783e-05,
+      "loss": 1.022,
+      "step": 15858
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8440108933434535e-05,
+      "loss": 1.0639,
+      "step": 15859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8431179963391225e-05,
+      "loss": 1.1569,
+      "step": 15860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8422250993347915e-05,
+      "loss": 0.9306,
+      "step": 15861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.841332202330462e-05,
+      "loss": 1.0802,
+      "step": 15862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.840439305326131e-05,
+      "loss": 0.8575,
+      "step": 15863
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8395464083218e-05,
+      "loss": 1.0661,
+      "step": 15864
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.83865351131747e-05,
+      "loss": 0.9656,
+      "step": 15865
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.837760614313139e-05,
+      "loss": 1.1116,
+      "step": 15866
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8368677173088085e-05,
+      "loss": 1.0517,
+      "step": 15867
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.835974820304478e-05,
+      "loss": 0.9827,
+      "step": 15868
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.835081923300147e-05,
+      "loss": 0.9869,
+      "step": 15869
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.8341890262958164e-05,
+      "loss": 1.0618,
+      "step": 15870
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.833296129291487e-05,
+      "loss": 1.0929,
+      "step": 15871
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.832403232287156e-05,
+      "loss": 1.1522,
+      "step": 15872
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.831510335282825e-05,
+      "loss": 0.921,
+      "step": 15873
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.830617438278495e-05,
+      "loss": 1.0533,
+      "step": 15874
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.829724541274164e-05,
+      "loss": 0.9787,
+      "step": 15875
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.828831644269834e-05,
+      "loss": 1.0723,
+      "step": 15876
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.827938747265503e-05,
+      "loss": 1.1019,
+      "step": 15877
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.827045850261172e-05,
+      "loss": 1.0499,
+      "step": 15878
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8261529532568425e-05,
+      "loss": 1.0602,
+      "step": 15879
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8252600562525116e-05,
+      "loss": 1.0247,
+      "step": 15880
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8243671592481806e-05,
+      "loss": 1.1288,
+      "step": 15881
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.823474262243851e-05,
+      "loss": 1.0974,
+      "step": 15882
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.82258136523952e-05,
+      "loss": 1.1454,
+      "step": 15883
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.821688468235189e-05,
+      "loss": 1.146,
+      "step": 15884
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8207955712308595e-05,
+      "loss": 1.0873,
+      "step": 15885
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8199026742265286e-05,
+      "loss": 1.0141,
+      "step": 15886
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8190097772221976e-05,
+      "loss": 0.8999,
+      "step": 15887
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8181168802178674e-05,
+      "loss": 0.8758,
+      "step": 15888
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8172239832135364e-05,
+      "loss": 1.0755,
+      "step": 15889
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8163310862092055e-05,
+      "loss": 1.022,
+      "step": 15890
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.815438189204876e-05,
+      "loss": 1.0644,
+      "step": 15891
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.814545292200545e-05,
+      "loss": 1.0609,
+      "step": 15892
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.813652395196214e-05,
+      "loss": 0.9792,
+      "step": 15893
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8127594981918844e-05,
+      "loss": 1.1407,
+      "step": 15894
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8118666011875534e-05,
+      "loss": 1.0493,
+      "step": 15895
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8109737041832225e-05,
+      "loss": 1.0751,
+      "step": 15896
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.810080807178893e-05,
+      "loss": 1.1827,
+      "step": 15897
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.809187910174562e-05,
+      "loss": 1.0812,
+      "step": 15898
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.808295013170231e-05,
+      "loss": 1.0419,
+      "step": 15899
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.807402116165901e-05,
+      "loss": 1.1002,
+      "step": 15900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.80650921916157e-05,
+      "loss": 1.0073,
+      "step": 15901
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.805616322157239e-05,
+      "loss": 0.9745,
+      "step": 15902
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.804723425152909e-05,
+      "loss": 1.1006,
+      "step": 15903
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.803830528148578e-05,
+      "loss": 1.154,
+      "step": 15904
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.802937631144247e-05,
+      "loss": 0.9911,
+      "step": 15905
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.802044734139918e-05,
+      "loss": 1.1267,
+      "step": 15906
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.801151837135587e-05,
+      "loss": 1.0737,
+      "step": 15907
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.800258940131256e-05,
+      "loss": 0.8706,
+      "step": 15908
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.799366043126926e-05,
+      "loss": 1.0885,
+      "step": 15909
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.798473146122595e-05,
+      "loss": 1.0936,
+      "step": 15910
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.797580249118264e-05,
+      "loss": 1.1463,
+      "step": 15911
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.796687352113934e-05,
+      "loss": 1.0833,
+      "step": 15912
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.795794455109603e-05,
+      "loss": 1.0425,
+      "step": 15913
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.794901558105272e-05,
+      "loss": 0.9916,
+      "step": 15914
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7940086611009425e-05,
+      "loss": 1.0973,
+      "step": 15915
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7931157640966116e-05,
+      "loss": 1.4841,
+      "step": 15916
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7922228670922806e-05,
+      "loss": 1.0471,
+      "step": 15917
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.791329970087951e-05,
+      "loss": 1.0376,
+      "step": 15918
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.79043707308362e-05,
+      "loss": 1.0702,
+      "step": 15919
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.789544176079289e-05,
+      "loss": 0.987,
+      "step": 15920
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7886512790749595e-05,
+      "loss": 0.9134,
+      "step": 15921
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7877583820706286e-05,
+      "loss": 0.9945,
+      "step": 15922
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7868654850662976e-05,
+      "loss": 1.0113,
+      "step": 15923
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.785972588061967e-05,
+      "loss": 1.0579,
+      "step": 15924
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7850796910576364e-05,
+      "loss": 1.1074,
+      "step": 15925
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.784186794053307e-05,
+      "loss": 1.0465,
+      "step": 15926
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.783293897048976e-05,
+      "loss": 1.1446,
+      "step": 15927
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.782401000044645e-05,
+      "loss": 0.9918,
+      "step": 15928
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.781508103040315e-05,
+      "loss": 1.0189,
+      "step": 15929
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.780615206035984e-05,
+      "loss": 0.9395,
+      "step": 15930
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7797223090316534e-05,
+      "loss": 1.0402,
+      "step": 15931
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.778829412027323e-05,
+      "loss": 1.0327,
+      "step": 15932
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.777936515022993e-05,
+      "loss": 1.0451,
+      "step": 15933
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.777043618018662e-05,
+      "loss": 1.1053,
+      "step": 15934
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7761507210143316e-05,
+      "loss": 0.9865,
+      "step": 15935
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7752578240100006e-05,
+      "loss": 1.0724,
+      "step": 15936
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.77436492700567e-05,
+      "loss": 0.9821,
+      "step": 15937
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.77347203000134e-05,
+      "loss": 1.0863,
+      "step": 15938
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.772579132997009e-05,
+      "loss": 1.0843,
+      "step": 15939
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.771686235992678e-05,
+      "loss": 1.0652,
+      "step": 15940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7707933389883486e-05,
+      "loss": 1.127,
+      "step": 15941
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7699004419840176e-05,
+      "loss": 1.1085,
+      "step": 15942
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.769007544979687e-05,
+      "loss": 1.0352,
+      "step": 15943
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7681146479753564e-05,
+      "loss": 1.04,
+      "step": 15944
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7672217509710255e-05,
+      "loss": 1.0479,
+      "step": 15945
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.766328853966695e-05,
+      "loss": 1.0162,
+      "step": 15946
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.765435956962365e-05,
+      "loss": 1.0766,
+      "step": 15947
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.764543059958034e-05,
+      "loss": 1.0777,
+      "step": 15948
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.763650162953703e-05,
+      "loss": 0.958,
+      "step": 15949
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7627572659493734e-05,
+      "loss": 0.9949,
+      "step": 15950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7618643689450425e-05,
+      "loss": 1.1571,
+      "step": 15951
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7609714719407115e-05,
+      "loss": 1.1007,
+      "step": 15952
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.760078574936382e-05,
+      "loss": 0.9196,
+      "step": 15953
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.759185677932051e-05,
+      "loss": 1.0414,
+      "step": 15954
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.75829278092772e-05,
+      "loss": 0.9282,
+      "step": 15955
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.75739988392339e-05,
+      "loss": 0.9507,
+      "step": 15956
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.756506986919059e-05,
+      "loss": 1.0244,
+      "step": 15957
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.755614089914728e-05,
+      "loss": 1.0078,
+      "step": 15958
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.754721192910398e-05,
+      "loss": 1.0585,
+      "step": 15959
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.753828295906067e-05,
+      "loss": 1.0757,
+      "step": 15960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.752935398901736e-05,
+      "loss": 1.0068,
+      "step": 15961
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.752042501897407e-05,
+      "loss": 0.9467,
+      "step": 15962
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.751149604893076e-05,
+      "loss": 1.1686,
+      "step": 15963
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.750256707888745e-05,
+      "loss": 0.9971,
+      "step": 15964
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.749363810884415e-05,
+      "loss": 1.098,
+      "step": 15965
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.748470913880084e-05,
+      "loss": 1.1121,
+      "step": 15966
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.747578016875753e-05,
+      "loss": 1.1004,
+      "step": 15967
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.746685119871423e-05,
+      "loss": 1.0001,
+      "step": 15968
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.745792222867092e-05,
+      "loss": 0.902,
+      "step": 15969
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.744899325862761e-05,
+      "loss": 1.1891,
+      "step": 15970
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7440064288584316e-05,
+      "loss": 0.972,
+      "step": 15971
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7431135318541006e-05,
+      "loss": 1.0246,
+      "step": 15972
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.742220634849771e-05,
+      "loss": 0.9105,
+      "step": 15973
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.74132773784544e-05,
+      "loss": 0.9758,
+      "step": 15974
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.740434840841109e-05,
+      "loss": 0.988,
+      "step": 15975
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7395419438367795e-05,
+      "loss": 1.1153,
+      "step": 15976
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7386490468324486e-05,
+      "loss": 0.9825,
+      "step": 15977
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7377561498281176e-05,
+      "loss": 1.0955,
+      "step": 15978
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.736863252823787e-05,
+      "loss": 1.1061,
+      "step": 15979
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7359703558194564e-05,
+      "loss": 0.9362,
+      "step": 15980
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7350774588151254e-05,
+      "loss": 1.0349,
+      "step": 15981
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.734184561810796e-05,
+      "loss": 1.3686,
+      "step": 15982
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.733291664806465e-05,
+      "loss": 1.0386,
+      "step": 15983
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.732398767802134e-05,
+      "loss": 1.121,
+      "step": 15984
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.731505870797804e-05,
+      "loss": 1.0317,
+      "step": 15985
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7306129737934734e-05,
+      "loss": 0.9443,
+      "step": 15986
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7297200767891424e-05,
+      "loss": 1.1408,
+      "step": 15987
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.728827179784813e-05,
+      "loss": 1.0034,
+      "step": 15988
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.727934282780482e-05,
+      "loss": 1.0286,
+      "step": 15989
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.727041385776151e-05,
+      "loss": 0.9917,
+      "step": 15990
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7261484887718206e-05,
+      "loss": 0.9502,
+      "step": 15991
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.72525559176749e-05,
+      "loss": 0.9889,
+      "step": 15992
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.724362694763159e-05,
+      "loss": 1.0111,
+      "step": 15993
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.723469797758829e-05,
+      "loss": 1.0519,
+      "step": 15994
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.722576900754498e-05,
+      "loss": 1.079,
+      "step": 15995
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.721684003750167e-05,
+      "loss": 1.001,
+      "step": 15996
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7207911067458376e-05,
+      "loss": 0.9782,
+      "step": 15997
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.719898209741507e-05,
+      "loss": 0.828,
+      "step": 15998
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.719005312737176e-05,
+      "loss": 1.1537,
+      "step": 15999
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.718112415732846e-05,
+      "loss": 0.9865,
+      "step": 16000
+    },
+    {
+      "epoch": 2.14,
+      "eval_loss": 1.0270307064056396,
+      "eval_runtime": 3048.7359,
+      "eval_samples_per_second": 6.168,
+      "eval_steps_per_second": 0.771,
+      "step": 16000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.717219518728515e-05,
+      "loss": 1.0658,
+      "step": 16001
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.716326621724184e-05,
+      "loss": 0.9856,
+      "step": 16002
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.715433724719854e-05,
+      "loss": 1.0617,
+      "step": 16003
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.714540827715523e-05,
+      "loss": 1.0906,
+      "step": 16004
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.713647930711192e-05,
+      "loss": 1.047,
+      "step": 16005
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7127550337068625e-05,
+      "loss": 0.9611,
+      "step": 16006
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7118621367025315e-05,
+      "loss": 1.0839,
+      "step": 16007
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7109692396982006e-05,
+      "loss": 1.0156,
+      "step": 16008
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.710076342693871e-05,
+      "loss": 0.9511,
+      "step": 16009
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.70918344568954e-05,
+      "loss": 1.0085,
+      "step": 16010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.708290548685209e-05,
+      "loss": 1.0346,
+      "step": 16011
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7073976516808795e-05,
+      "loss": 1.1061,
+      "step": 16012
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7065047546765485e-05,
+      "loss": 0.9732,
+      "step": 16013
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7056118576722176e-05,
+      "loss": 1.0291,
+      "step": 16014
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.704718960667887e-05,
+      "loss": 1.2074,
+      "step": 16015
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.703826063663556e-05,
+      "loss": 0.9086,
+      "step": 16016
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7029331666592254e-05,
+      "loss": 1.0671,
+      "step": 16017
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.702040269654896e-05,
+      "loss": 0.8674,
+      "step": 16018
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.701147372650565e-05,
+      "loss": 1.0816,
+      "step": 16019
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.700254475646234e-05,
+      "loss": 1.1119,
+      "step": 16020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.699361578641904e-05,
+      "loss": 1.0814,
+      "step": 16021
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.698468681637573e-05,
+      "loss": 1.0531,
+      "step": 16022
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.697575784633243e-05,
+      "loss": 1.1052,
+      "step": 16023
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.696682887628912e-05,
+      "loss": 0.8841,
+      "step": 16024
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.695789990624582e-05,
+      "loss": 0.9898,
+      "step": 16025
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6948970936202516e-05,
+      "loss": 1.1623,
+      "step": 16026
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6940041966159206e-05,
+      "loss": 1.0334,
+      "step": 16027
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6931112996115897e-05,
+      "loss": 1.0864,
+      "step": 16028
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.69221840260726e-05,
+      "loss": 0.9606,
+      "step": 16029
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.691325505602929e-05,
+      "loss": 0.9836,
+      "step": 16030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.690432608598598e-05,
+      "loss": 1.0255,
+      "step": 16031
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6895397115942686e-05,
+      "loss": 0.9968,
+      "step": 16032
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6886468145899376e-05,
+      "loss": 1.1136,
+      "step": 16033
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6877539175856067e-05,
+      "loss": 1.2374,
+      "step": 16034
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6868610205812764e-05,
+      "loss": 1.0957,
+      "step": 16035
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6859681235769454e-05,
+      "loss": 1.0151,
+      "step": 16036
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.685075226572615e-05,
+      "loss": 1.0493,
+      "step": 16037
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.684182329568285e-05,
+      "loss": 1.1206,
+      "step": 16038
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.683289432563954e-05,
+      "loss": 0.9764,
+      "step": 16039
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.682396535559623e-05,
+      "loss": 1.0916,
+      "step": 16040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6815036385552934e-05,
+      "loss": 1.0467,
+      "step": 16041
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6806107415509624e-05,
+      "loss": 1.1265,
+      "step": 16042
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6797178445466315e-05,
+      "loss": 1.0546,
+      "step": 16043
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.678824947542302e-05,
+      "loss": 0.9398,
+      "step": 16044
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.677932050537971e-05,
+      "loss": 0.8579,
+      "step": 16045
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.67703915353364e-05,
+      "loss": 0.8969,
+      "step": 16046
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.67614625652931e-05,
+      "loss": 0.8604,
+      "step": 16047
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.675253359524979e-05,
+      "loss": 1.1477,
+      "step": 16048
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.674360462520648e-05,
+      "loss": 0.9215,
+      "step": 16049
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.673467565516318e-05,
+      "loss": 0.9349,
+      "step": 16050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.672574668511987e-05,
+      "loss": 1.0905,
+      "step": 16051
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.671681771507656e-05,
+      "loss": 0.9054,
+      "step": 16052
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.670788874503327e-05,
+      "loss": 0.9685,
+      "step": 16053
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.669895977498996e-05,
+      "loss": 1.1658,
+      "step": 16054
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.669003080494665e-05,
+      "loss": 0.9134,
+      "step": 16055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.668110183490335e-05,
+      "loss": 1.0434,
+      "step": 16056
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.667217286486004e-05,
+      "loss": 1.0168,
+      "step": 16057
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.666324389481673e-05,
+      "loss": 0.9994,
+      "step": 16058
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.665431492477343e-05,
+      "loss": 1.1791,
+      "step": 16059
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.664538595473012e-05,
+      "loss": 0.9934,
+      "step": 16060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.663645698468681e-05,
+      "loss": 1.1398,
+      "step": 16061
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6627528014643515e-05,
+      "loss": 1.1378,
+      "step": 16062
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6618599044600206e-05,
+      "loss": 0.9704,
+      "step": 16063
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6609670074556896e-05,
+      "loss": 1.0212,
+      "step": 16064
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.66007411045136e-05,
+      "loss": 0.9633,
+      "step": 16065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.659181213447029e-05,
+      "loss": 1.0271,
+      "step": 16066
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.658288316442698e-05,
+      "loss": 1.0359,
+      "step": 16067
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6573954194383685e-05,
+      "loss": 0.9564,
+      "step": 16068
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6565025224340376e-05,
+      "loss": 0.9408,
+      "step": 16069
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6556096254297066e-05,
+      "loss": 0.9631,
+      "step": 16070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6547167284253763e-05,
+      "loss": 1.0329,
+      "step": 16071
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6538238314210454e-05,
+      "loss": 0.8451,
+      "step": 16072
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.652930934416716e-05,
+      "loss": 1.0296,
+      "step": 16073
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.652038037412385e-05,
+      "loss": 1.0023,
+      "step": 16074
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.651145140408054e-05,
+      "loss": 0.8583,
+      "step": 16075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.650252243403724e-05,
+      "loss": 1.0333,
+      "step": 16076
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.649359346399393e-05,
+      "loss": 1.0414,
+      "step": 16077
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6484664493950624e-05,
+      "loss": 1.0801,
+      "step": 16078
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.647573552390733e-05,
+      "loss": 0.9953,
+      "step": 16079
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.646680655386402e-05,
+      "loss": 1.0695,
+      "step": 16080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.645787758382071e-05,
+      "loss": 1.0776,
+      "step": 16081
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6448948613777406e-05,
+      "loss": 0.9882,
+      "step": 16082
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6440019643734097e-05,
+      "loss": 1.1549,
+      "step": 16083
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.643109067369079e-05,
+      "loss": 0.9652,
+      "step": 16084
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.642216170364749e-05,
+      "loss": 1.0541,
+      "step": 16085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.641323273360418e-05,
+      "loss": 1.1779,
+      "step": 16086
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.640430376356087e-05,
+      "loss": 0.9533,
+      "step": 16087
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6395374793517576e-05,
+      "loss": 1.1122,
+      "step": 16088
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6386445823474267e-05,
+      "loss": 1.0057,
+      "step": 16089
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.637751685343096e-05,
+      "loss": 0.9258,
+      "step": 16090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.636858788338766e-05,
+      "loss": 1.0639,
+      "step": 16091
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.635965891334435e-05,
+      "loss": 1.0973,
+      "step": 16092
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.635072994330104e-05,
+      "loss": 1.09,
+      "step": 16093
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.634180097325774e-05,
+      "loss": 0.9636,
+      "step": 16094
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.633287200321443e-05,
+      "loss": 1.0696,
+      "step": 16095
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.632394303317112e-05,
+      "loss": 1.0509,
+      "step": 16096
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6315014063127824e-05,
+      "loss": 0.9715,
+      "step": 16097
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6306085093084515e-05,
+      "loss": 1.1698,
+      "step": 16098
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6297156123041205e-05,
+      "loss": 1.0166,
+      "step": 16099
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.628822715299791e-05,
+      "loss": 1.2914,
+      "step": 16100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.62792981829546e-05,
+      "loss": 0.9806,
+      "step": 16101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.627036921291129e-05,
+      "loss": 1.0339,
+      "step": 16102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6261440242867994e-05,
+      "loss": 0.9436,
+      "step": 16103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6252511272824685e-05,
+      "loss": 1.0404,
+      "step": 16104
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6243582302781375e-05,
+      "loss": 1.0829,
+      "step": 16105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.623465333273807e-05,
+      "loss": 0.9546,
+      "step": 16106
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.622572436269476e-05,
+      "loss": 1.072,
+      "step": 16107
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6216795392651453e-05,
+      "loss": 0.9864,
+      "step": 16108
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.620786642260816e-05,
+      "loss": 1.0519,
+      "step": 16109
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.619893745256485e-05,
+      "loss": 1.147,
+      "step": 16110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.619000848252154e-05,
+      "loss": 0.9028,
+      "step": 16111
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.618107951247824e-05,
+      "loss": 0.966,
+      "step": 16112
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.617215054243493e-05,
+      "loss": 1.0578,
+      "step": 16113
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6163221572391623e-05,
+      "loss": 0.96,
+      "step": 16114
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.615429260234832e-05,
+      "loss": 1.0188,
+      "step": 16115
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.614536363230502e-05,
+      "loss": 0.9616,
+      "step": 16116
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.613643466226171e-05,
+      "loss": 1.0681,
+      "step": 16117
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6127505692218406e-05,
+      "loss": 1.3226,
+      "step": 16118
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6118576722175096e-05,
+      "loss": 1.0245,
+      "step": 16119
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.61096477521318e-05,
+      "loss": 1.0546,
+      "step": 16120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.610071878208849e-05,
+      "loss": 1.0369,
+      "step": 16121
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.609178981204518e-05,
+      "loss": 1.0396,
+      "step": 16122
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6082860842001885e-05,
+      "loss": 1.0464,
+      "step": 16123
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6073931871958576e-05,
+      "loss": 0.9447,
+      "step": 16124
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6065002901915266e-05,
+      "loss": 1.0151,
+      "step": 16125
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6056073931871963e-05,
+      "loss": 0.8832,
+      "step": 16126
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6047144961828654e-05,
+      "loss": 1.1821,
+      "step": 16127
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6038215991785344e-05,
+      "loss": 0.9478,
+      "step": 16128
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.602928702174205e-05,
+      "loss": 0.9684,
+      "step": 16129
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.602035805169874e-05,
+      "loss": 1.1776,
+      "step": 16130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.601142908165543e-05,
+      "loss": 1.0257,
+      "step": 16131
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.6002500111612133e-05,
+      "loss": 0.9744,
+      "step": 16132
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5993571141568824e-05,
+      "loss": 1.0883,
+      "step": 16133
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5984642171525514e-05,
+      "loss": 0.9802,
+      "step": 16134
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.597571320148222e-05,
+      "loss": 0.974,
+      "step": 16135
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.596678423143891e-05,
+      "loss": 1.0114,
+      "step": 16136
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.59578552613956e-05,
+      "loss": 1.0149,
+      "step": 16137
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.59489262913523e-05,
+      "loss": 0.9776,
+      "step": 16138
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.593999732130899e-05,
+      "loss": 0.9374,
+      "step": 16139
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.593106835126568e-05,
+      "loss": 0.9908,
+      "step": 16140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.592213938122238e-05,
+      "loss": 1.0766,
+      "step": 16141
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.591321041117907e-05,
+      "loss": 0.9673,
+      "step": 16142
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.590428144113576e-05,
+      "loss": 1.1235,
+      "step": 16143
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.589535247109247e-05,
+      "loss": 1.0059,
+      "step": 16144
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.588642350104916e-05,
+      "loss": 1.0322,
+      "step": 16145
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.587749453100585e-05,
+      "loss": 1.0792,
+      "step": 16146
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.586856556096255e-05,
+      "loss": 0.9676,
+      "step": 16147
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.585963659091924e-05,
+      "loss": 1.0374,
+      "step": 16148
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.585070762087593e-05,
+      "loss": 1.0612,
+      "step": 16149
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.584177865083263e-05,
+      "loss": 0.9224,
+      "step": 16150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.583284968078932e-05,
+      "loss": 1.0298,
+      "step": 16151
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.582392071074601e-05,
+      "loss": 1.0413,
+      "step": 16152
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5814991740702715e-05,
+      "loss": 0.9786,
+      "step": 16153
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5806062770659405e-05,
+      "loss": 0.9612,
+      "step": 16154
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5797133800616096e-05,
+      "loss": 0.8932,
+      "step": 16155
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.57882048305728e-05,
+      "loss": 0.9721,
+      "step": 16156
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.577927586052949e-05,
+      "loss": 1.0134,
+      "step": 16157
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.577034689048618e-05,
+      "loss": 1.099,
+      "step": 16158
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5761417920442885e-05,
+      "loss": 1.0735,
+      "step": 16159
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5752488950399575e-05,
+      "loss": 0.9844,
+      "step": 16160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5743559980356266e-05,
+      "loss": 0.9646,
+      "step": 16161
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.573463101031296e-05,
+      "loss": 1.0719,
+      "step": 16162
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5725702040269654e-05,
+      "loss": 0.9066,
+      "step": 16163
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5716773070226344e-05,
+      "loss": 1.0566,
+      "step": 16164
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.570784410018305e-05,
+      "loss": 1.2085,
+      "step": 16165
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.569891513013974e-05,
+      "loss": 0.8509,
+      "step": 16166
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.568998616009643e-05,
+      "loss": 1.0323,
+      "step": 16167
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.568105719005313e-05,
+      "loss": 1.0745,
+      "step": 16168
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5672128220009823e-05,
+      "loss": 1.0508,
+      "step": 16169
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.566319924996653e-05,
+      "loss": 1.0353,
+      "step": 16170
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.565427027992322e-05,
+      "loss": 0.9879,
+      "step": 16171
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.564534130987991e-05,
+      "loss": 1.1595,
+      "step": 16172
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5636412339836606e-05,
+      "loss": 1.0067,
+      "step": 16173
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5627483369793296e-05,
+      "loss": 1.0307,
+      "step": 16174
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.561855439974999e-05,
+      "loss": 1.0865,
+      "step": 16175
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.560962542970669e-05,
+      "loss": 0.9764,
+      "step": 16176
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.560069645966338e-05,
+      "loss": 1.0551,
+      "step": 16177
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.559176748962007e-05,
+      "loss": 0.9844,
+      "step": 16178
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5582838519576776e-05,
+      "loss": 1.054,
+      "step": 16179
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5573909549533466e-05,
+      "loss": 0.9914,
+      "step": 16180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.556498057949016e-05,
+      "loss": 1.0346,
+      "step": 16181
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.555605160944686e-05,
+      "loss": 1.0728,
+      "step": 16182
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.554712263940355e-05,
+      "loss": 0.9875,
+      "step": 16183
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.553819366936024e-05,
+      "loss": 0.9332,
+      "step": 16184
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.552926469931694e-05,
+      "loss": 0.9323,
+      "step": 16185
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.552033572927363e-05,
+      "loss": 0.9043,
+      "step": 16186
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.551140675923032e-05,
+      "loss": 1.064,
+      "step": 16187
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5502477789187024e-05,
+      "loss": 1.0855,
+      "step": 16188
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5493548819143714e-05,
+      "loss": 0.8788,
+      "step": 16189
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5484619849100405e-05,
+      "loss": 0.9342,
+      "step": 16190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.547569087905711e-05,
+      "loss": 1.1555,
+      "step": 16191
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.54667619090138e-05,
+      "loss": 0.9372,
+      "step": 16192
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.545783293897049e-05,
+      "loss": 1.0771,
+      "step": 16193
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.544890396892719e-05,
+      "loss": 1.0734,
+      "step": 16194
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5439974998883884e-05,
+      "loss": 0.8433,
+      "step": 16195
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5431046028840575e-05,
+      "loss": 0.8987,
+      "step": 16196
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.542211705879727e-05,
+      "loss": 0.9153,
+      "step": 16197
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.541318808875396e-05,
+      "loss": 0.9295,
+      "step": 16198
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.540425911871065e-05,
+      "loss": 0.9895,
+      "step": 16199
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.539533014866736e-05,
+      "loss": 1.0332,
+      "step": 16200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.538640117862405e-05,
+      "loss": 1.0085,
+      "step": 16201
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.537747220858074e-05,
+      "loss": 0.9623,
+      "step": 16202
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.536854323853744e-05,
+      "loss": 0.9749,
+      "step": 16203
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.535961426849413e-05,
+      "loss": 1.0074,
+      "step": 16204
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.535068529845082e-05,
+      "loss": 0.9442,
+      "step": 16205
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.534175632840752e-05,
+      "loss": 1.0854,
+      "step": 16206
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.533282735836422e-05,
+      "loss": 0.9318,
+      "step": 16207
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.532389838832091e-05,
+      "loss": 0.8549,
+      "step": 16208
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5314969418277605e-05,
+      "loss": 0.9723,
+      "step": 16209
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5306040448234296e-05,
+      "loss": 1.2813,
+      "step": 16210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5297111478190986e-05,
+      "loss": 1.0187,
+      "step": 16211
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.528818250814769e-05,
+      "loss": 1.0354,
+      "step": 16212
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.527925353810438e-05,
+      "loss": 0.9467,
+      "step": 16213
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.527032456806107e-05,
+      "loss": 1.0376,
+      "step": 16214
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5261395598017775e-05,
+      "loss": 1.0953,
+      "step": 16215
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5252466627974466e-05,
+      "loss": 0.9721,
+      "step": 16216
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5243537657931156e-05,
+      "loss": 1.0418,
+      "step": 16217
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5234608687887854e-05,
+      "loss": 0.9637,
+      "step": 16218
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5225679717844544e-05,
+      "loss": 1.0943,
+      "step": 16219
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.521675074780125e-05,
+      "loss": 1.0039,
+      "step": 16220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.520782177775794e-05,
+      "loss": 0.7745,
+      "step": 16221
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.519889280771463e-05,
+      "loss": 0.9535,
+      "step": 16222
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.518996383767133e-05,
+      "loss": 1.0899,
+      "step": 16223
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5181034867628024e-05,
+      "loss": 0.9926,
+      "step": 16224
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5172105897584714e-05,
+      "loss": 1.0019,
+      "step": 16225
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.516317692754142e-05,
+      "loss": 1.0145,
+      "step": 16226
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.515424795749811e-05,
+      "loss": 1.0207,
+      "step": 16227
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.51453189874548e-05,
+      "loss": 1.0306,
+      "step": 16228
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5136390017411496e-05,
+      "loss": 0.8793,
+      "step": 16229
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.512746104736819e-05,
+      "loss": 0.9782,
+      "step": 16230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.511853207732488e-05,
+      "loss": 1.1009,
+      "step": 16231
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.510960310728158e-05,
+      "loss": 1.0241,
+      "step": 16232
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.510067413723827e-05,
+      "loss": 0.9752,
+      "step": 16233
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.509174516719496e-05,
+      "loss": 1.0808,
+      "step": 16234
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5082816197151666e-05,
+      "loss": 1.0976,
+      "step": 16235
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.507388722710836e-05,
+      "loss": 0.9461,
+      "step": 16236
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.506495825706505e-05,
+      "loss": 1.1451,
+      "step": 16237
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.505602928702175e-05,
+      "loss": 1.0033,
+      "step": 16238
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.504710031697844e-05,
+      "loss": 0.9249,
+      "step": 16239
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.503817134693513e-05,
+      "loss": 1.049,
+      "step": 16240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.502924237689183e-05,
+      "loss": 0.9999,
+      "step": 16241
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.502031340684852e-05,
+      "loss": 1.0001,
+      "step": 16242
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.501138443680521e-05,
+      "loss": 1.0608,
+      "step": 16243
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.5002455466761914e-05,
+      "loss": 0.8983,
+      "step": 16244
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4993526496718605e-05,
+      "loss": 1.018,
+      "step": 16245
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4984597526675295e-05,
+      "loss": 1.1362,
+      "step": 16246
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4975668556632e-05,
+      "loss": 0.9676,
+      "step": 16247
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.496673958658869e-05,
+      "loss": 1.0354,
+      "step": 16248
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.495781061654538e-05,
+      "loss": 1.0027,
+      "step": 16249
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4948881646502084e-05,
+      "loss": 1.1202,
+      "step": 16250
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4939952676458775e-05,
+      "loss": 1.0104,
+      "step": 16251
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4931023706415465e-05,
+      "loss": 0.9861,
+      "step": 16252
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.492209473637216e-05,
+      "loss": 0.9879,
+      "step": 16253
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.491316576632885e-05,
+      "loss": 1.0373,
+      "step": 16254
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4904236796285544e-05,
+      "loss": 1.0543,
+      "step": 16255
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.489530782624225e-05,
+      "loss": 1.0802,
+      "step": 16256
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.488637885619894e-05,
+      "loss": 1.051,
+      "step": 16257
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.487744988615563e-05,
+      "loss": 1.1019,
+      "step": 16258
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.486852091611233e-05,
+      "loss": 1.0406,
+      "step": 16259
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.485959194606902e-05,
+      "loss": 1.0143,
+      "step": 16260
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4850662976025714e-05,
+      "loss": 1.1546,
+      "step": 16261
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.484173400598242e-05,
+      "loss": 1.123,
+      "step": 16262
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.483280503593911e-05,
+      "loss": 1.0688,
+      "step": 16263
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.48238760658958e-05,
+      "loss": 1.2985,
+      "step": 16264
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4814947095852496e-05,
+      "loss": 1.0533,
+      "step": 16265
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4806018125809186e-05,
+      "loss": 1.06,
+      "step": 16266
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.479708915576589e-05,
+      "loss": 1.1144,
+      "step": 16267
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.478816018572258e-05,
+      "loss": 1.0846,
+      "step": 16268
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.477923121567927e-05,
+      "loss": 0.954,
+      "step": 16269
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4770302245635975e-05,
+      "loss": 0.9526,
+      "step": 16270
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4761373275592666e-05,
+      "loss": 0.9472,
+      "step": 16271
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4752444305549356e-05,
+      "loss": 1.0319,
+      "step": 16272
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.474351533550606e-05,
+      "loss": 0.999,
+      "step": 16273
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.473458636546275e-05,
+      "loss": 1.0564,
+      "step": 16274
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.472565739541944e-05,
+      "loss": 1.043,
+      "step": 16275
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.471672842537614e-05,
+      "loss": 1.107,
+      "step": 16276
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.470779945533283e-05,
+      "loss": 1.066,
+      "step": 16277
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.469887048528952e-05,
+      "loss": 1.0177,
+      "step": 16278
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4689941515246224e-05,
+      "loss": 1.0995,
+      "step": 16279
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4681012545202914e-05,
+      "loss": 0.9732,
+      "step": 16280
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4672083575159605e-05,
+      "loss": 1.0164,
+      "step": 16281
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.466315460511631e-05,
+      "loss": 0.9436,
+      "step": 16282
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4654225635073e-05,
+      "loss": 1.013,
+      "step": 16283
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.464529666502969e-05,
+      "loss": 1.1049,
+      "step": 16284
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.463636769498639e-05,
+      "loss": 1.0815,
+      "step": 16285
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4627438724943084e-05,
+      "loss": 0.9875,
+      "step": 16286
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4618509754899775e-05,
+      "loss": 0.9471,
+      "step": 16287
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.460958078485647e-05,
+      "loss": 1.162,
+      "step": 16288
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.460065181481316e-05,
+      "loss": 1.1107,
+      "step": 16289
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.459172284476985e-05,
+      "loss": 0.9698,
+      "step": 16290
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.458279387472656e-05,
+      "loss": 1.1102,
+      "step": 16291
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.457386490468325e-05,
+      "loss": 1.1282,
+      "step": 16292
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.456493593463994e-05,
+      "loss": 0.9609,
+      "step": 16293
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.455600696459664e-05,
+      "loss": 0.9214,
+      "step": 16294
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.454707799455333e-05,
+      "loss": 0.9991,
+      "step": 16295
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.453814902451002e-05,
+      "loss": 1.0098,
+      "step": 16296
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.452922005446672e-05,
+      "loss": 1.1122,
+      "step": 16297
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.452029108442341e-05,
+      "loss": 0.9826,
+      "step": 16298
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.451136211438011e-05,
+      "loss": 1.0719,
+      "step": 16299
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4502433144336805e-05,
+      "loss": 1.1292,
+      "step": 16300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4493504174293495e-05,
+      "loss": 1.0813,
+      "step": 16301
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4484575204250186e-05,
+      "loss": 0.9523,
+      "step": 16302
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.447564623420689e-05,
+      "loss": 1.1459,
+      "step": 16303
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.446671726416358e-05,
+      "loss": 0.9851,
+      "step": 16304
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.445778829412027e-05,
+      "loss": 1.0605,
+      "step": 16305
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4448859324076975e-05,
+      "loss": 0.8066,
+      "step": 16306
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4439930354033665e-05,
+      "loss": 0.9397,
+      "step": 16307
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4431001383990356e-05,
+      "loss": 1.1415,
+      "step": 16308
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.442207241394705e-05,
+      "loss": 1.032,
+      "step": 16309
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4413143443903744e-05,
+      "loss": 1.1555,
+      "step": 16310
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.440421447386044e-05,
+      "loss": 1.109,
+      "step": 16311
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.439528550381714e-05,
+      "loss": 1.0469,
+      "step": 16312
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.438635653377383e-05,
+      "loss": 1.0245,
+      "step": 16313
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.437742756373052e-05,
+      "loss": 1.2178,
+      "step": 16314
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.436849859368722e-05,
+      "loss": 1.0013,
+      "step": 16315
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4359569623643914e-05,
+      "loss": 1.1367,
+      "step": 16316
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.435064065360062e-05,
+      "loss": 1.1371,
+      "step": 16317
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.434171168355731e-05,
+      "loss": 1.0788,
+      "step": 16318
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4332782713514e-05,
+      "loss": 0.9837,
+      "step": 16319
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4323853743470696e-05,
+      "loss": 1.0473,
+      "step": 16320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4314924773427386e-05,
+      "loss": 0.9871,
+      "step": 16321
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.430599580338408e-05,
+      "loss": 1.0278,
+      "step": 16322
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.429706683334078e-05,
+      "loss": 1.0331,
+      "step": 16323
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.428813786329747e-05,
+      "loss": 1.0575,
+      "step": 16324
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.427920889325416e-05,
+      "loss": 0.9514,
+      "step": 16325
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4270279923210866e-05,
+      "loss": 1.0704,
+      "step": 16326
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4261350953167556e-05,
+      "loss": 0.9978,
+      "step": 16327
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.425242198312425e-05,
+      "loss": 1.0641,
+      "step": 16328
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.424349301308095e-05,
+      "loss": 1.0113,
+      "step": 16329
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.423456404303764e-05,
+      "loss": 1.2096,
+      "step": 16330
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.422563507299433e-05,
+      "loss": 1.0418,
+      "step": 16331
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.421670610295103e-05,
+      "loss": 1.0944,
+      "step": 16332
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.420777713290772e-05,
+      "loss": 0.9753,
+      "step": 16333
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.419884816286441e-05,
+      "loss": 1.0628,
+      "step": 16334
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4189919192821114e-05,
+      "loss": 0.9017,
+      "step": 16335
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4180990222777805e-05,
+      "loss": 1.2424,
+      "step": 16336
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4172061252734495e-05,
+      "loss": 1.1369,
+      "step": 16337
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.41631322826912e-05,
+      "loss": 0.9473,
+      "step": 16338
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.415420331264789e-05,
+      "loss": 0.9956,
+      "step": 16339
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.414527434260458e-05,
+      "loss": 0.9557,
+      "step": 16340
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4136345372561284e-05,
+      "loss": 1.0748,
+      "step": 16341
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4127416402517975e-05,
+      "loss": 1.0251,
+      "step": 16342
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4118487432474665e-05,
+      "loss": 0.9691,
+      "step": 16343
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.410955846243136e-05,
+      "loss": 1.4792,
+      "step": 16344
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.410062949238805e-05,
+      "loss": 1.1432,
+      "step": 16345
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.409170052234474e-05,
+      "loss": 1.0558,
+      "step": 16346
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.408277155230145e-05,
+      "loss": 1.0261,
+      "step": 16347
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.407384258225814e-05,
+      "loss": 0.9854,
+      "step": 16348
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.406491361221483e-05,
+      "loss": 0.9584,
+      "step": 16349
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.405598464217153e-05,
+      "loss": 0.9876,
+      "step": 16350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.404705567212822e-05,
+      "loss": 0.9476,
+      "step": 16351
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.403812670208491e-05,
+      "loss": 1.0868,
+      "step": 16352
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.402919773204162e-05,
+      "loss": 0.8853,
+      "step": 16353
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.402026876199831e-05,
+      "loss": 0.82,
+      "step": 16354
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4011339791955e-05,
+      "loss": 1.0457,
+      "step": 16355
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.4002410821911695e-05,
+      "loss": 1.0528,
+      "step": 16356
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3993481851868386e-05,
+      "loss": 1.0941,
+      "step": 16357
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3984552881825076e-05,
+      "loss": 1.072,
+      "step": 16358
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.397562391178178e-05,
+      "loss": 1.081,
+      "step": 16359
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.396669494173847e-05,
+      "loss": 0.9624,
+      "step": 16360
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.395776597169516e-05,
+      "loss": 1.0823,
+      "step": 16361
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3948837001651865e-05,
+      "loss": 1.0157,
+      "step": 16362
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3939908031608556e-05,
+      "loss": 0.9402,
+      "step": 16363
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3930979061565246e-05,
+      "loss": 0.9247,
+      "step": 16364
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.392205009152195e-05,
+      "loss": 0.9442,
+      "step": 16365
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.391312112147864e-05,
+      "loss": 1.0164,
+      "step": 16366
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.390419215143534e-05,
+      "loss": 1.1285,
+      "step": 16367
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.389526318139203e-05,
+      "loss": 1.002,
+      "step": 16368
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.388633421134872e-05,
+      "loss": 0.9542,
+      "step": 16369
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.387740524130542e-05,
+      "loss": 1.0566,
+      "step": 16370
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3868476271262114e-05,
+      "loss": 0.9958,
+      "step": 16371
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3859547301218804e-05,
+      "loss": 0.9807,
+      "step": 16372
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.385061833117551e-05,
+      "loss": 0.9162,
+      "step": 16373
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.38416893611322e-05,
+      "loss": 1.1183,
+      "step": 16374
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.383276039108889e-05,
+      "loss": 0.968,
+      "step": 16375
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3823831421045586e-05,
+      "loss": 0.9981,
+      "step": 16376
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3814902451002284e-05,
+      "loss": 1.0871,
+      "step": 16377
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3805973480958974e-05,
+      "loss": 1.0153,
+      "step": 16378
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.379704451091567e-05,
+      "loss": 0.9649,
+      "step": 16379
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.378811554087236e-05,
+      "loss": 0.9101,
+      "step": 16380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.377918657082905e-05,
+      "loss": 1.0113,
+      "step": 16381
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3770257600785756e-05,
+      "loss": 1.0258,
+      "step": 16382
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.376132863074245e-05,
+      "loss": 1.0249,
+      "step": 16383
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.375239966069914e-05,
+      "loss": 0.9662,
+      "step": 16384
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.374347069065584e-05,
+      "loss": 0.9532,
+      "step": 16385
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.373454172061253e-05,
+      "loss": 1.1213,
+      "step": 16386
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.372561275056922e-05,
+      "loss": 0.9558,
+      "step": 16387
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.371668378052592e-05,
+      "loss": 1.1072,
+      "step": 16388
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.370775481048261e-05,
+      "loss": 0.9897,
+      "step": 16389
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.369882584043931e-05,
+      "loss": 1.0334,
+      "step": 16390
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3689896870396005e-05,
+      "loss": 0.9647,
+      "step": 16391
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3680967900352695e-05,
+      "loss": 1.103,
+      "step": 16392
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3672038930309386e-05,
+      "loss": 0.9687,
+      "step": 16393
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.366310996026609e-05,
+      "loss": 1.0021,
+      "step": 16394
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.365418099022278e-05,
+      "loss": 1.0601,
+      "step": 16395
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.364525202017947e-05,
+      "loss": 0.9733,
+      "step": 16396
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3636323050136175e-05,
+      "loss": 1.1386,
+      "step": 16397
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3627394080092865e-05,
+      "loss": 1.0439,
+      "step": 16398
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3618465110049556e-05,
+      "loss": 1.0944,
+      "step": 16399
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.360953614000625e-05,
+      "loss": 1.038,
+      "step": 16400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.360060716996294e-05,
+      "loss": 1.0398,
+      "step": 16401
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3591678199919634e-05,
+      "loss": 1.078,
+      "step": 16402
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.358274922987634e-05,
+      "loss": 1.0159,
+      "step": 16403
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.357382025983303e-05,
+      "loss": 1.035,
+      "step": 16404
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.356489128978972e-05,
+      "loss": 1.0279,
+      "step": 16405
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.355596231974642e-05,
+      "loss": 0.9767,
+      "step": 16406
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.354703334970311e-05,
+      "loss": 1.1391,
+      "step": 16407
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3538104379659804e-05,
+      "loss": 1.061,
+      "step": 16408
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.352917540961651e-05,
+      "loss": 1.035,
+      "step": 16409
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.35202464395732e-05,
+      "loss": 0.9776,
+      "step": 16410
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.351131746952989e-05,
+      "loss": 0.9242,
+      "step": 16411
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3502388499486586e-05,
+      "loss": 1.064,
+      "step": 16412
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3493459529443276e-05,
+      "loss": 1.012,
+      "step": 16413
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.348453055939998e-05,
+      "loss": 1.0907,
+      "step": 16414
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.347560158935667e-05,
+      "loss": 1.1326,
+      "step": 16415
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.346667261931336e-05,
+      "loss": 1.0476,
+      "step": 16416
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3457743649270066e-05,
+      "loss": 0.9993,
+      "step": 16417
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3448814679226756e-05,
+      "loss": 0.9393,
+      "step": 16418
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3439885709183446e-05,
+      "loss": 0.9964,
+      "step": 16419
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.343095673914015e-05,
+      "loss": 0.9434,
+      "step": 16420
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.342202776909684e-05,
+      "loss": 0.93,
+      "step": 16421
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.341309879905353e-05,
+      "loss": 1.1555,
+      "step": 16422
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.340416982901023e-05,
+      "loss": 1.0967,
+      "step": 16423
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.339524085896692e-05,
+      "loss": 0.9287,
+      "step": 16424
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.338631188892361e-05,
+      "loss": 1.1337,
+      "step": 16425
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3377382918880314e-05,
+      "loss": 0.9804,
+      "step": 16426
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3368453948837004e-05,
+      "loss": 1.0374,
+      "step": 16427
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3359524978793695e-05,
+      "loss": 1.0766,
+      "step": 16428
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.33505960087504e-05,
+      "loss": 0.9564,
+      "step": 16429
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.334166703870709e-05,
+      "loss": 1.0048,
+      "step": 16430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.333273806866378e-05,
+      "loss": 1.0831,
+      "step": 16431
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3323809098620484e-05,
+      "loss": 1.0149,
+      "step": 16432
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3314880128577174e-05,
+      "loss": 1.0971,
+      "step": 16433
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3305951158533865e-05,
+      "loss": 0.9908,
+      "step": 16434
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.329702218849056e-05,
+      "loss": 1.0505,
+      "step": 16435
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.328809321844725e-05,
+      "loss": 0.9508,
+      "step": 16436
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.327916424840394e-05,
+      "loss": 0.9768,
+      "step": 16437
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.327023527836065e-05,
+      "loss": 1.093,
+      "step": 16438
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.326130630831734e-05,
+      "loss": 0.8939,
+      "step": 16439
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.325237733827403e-05,
+      "loss": 0.9887,
+      "step": 16440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.324344836823073e-05,
+      "loss": 1.0843,
+      "step": 16441
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.323451939818742e-05,
+      "loss": 1.0808,
+      "step": 16442
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.322559042814411e-05,
+      "loss": 1.0182,
+      "step": 16443
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.321666145810082e-05,
+      "loss": 1.0501,
+      "step": 16444
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.320773248805751e-05,
+      "loss": 1.0844,
+      "step": 16445
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.31988035180142e-05,
+      "loss": 0.9443,
+      "step": 16446
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3189874547970895e-05,
+      "loss": 1.007,
+      "step": 16447
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3180945577927586e-05,
+      "loss": 1.0299,
+      "step": 16448
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3172016607884276e-05,
+      "loss": 0.9391,
+      "step": 16449
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.316308763784098e-05,
+      "loss": 0.9959,
+      "step": 16450
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.315415866779767e-05,
+      "loss": 1.0146,
+      "step": 16451
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.314522969775436e-05,
+      "loss": 1.0619,
+      "step": 16452
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3136300727711065e-05,
+      "loss": 0.9724,
+      "step": 16453
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3127371757667756e-05,
+      "loss": 1.0353,
+      "step": 16454
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3118442787624446e-05,
+      "loss": 0.9686,
+      "step": 16455
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.310951381758115e-05,
+      "loss": 1.0201,
+      "step": 16456
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.310058484753784e-05,
+      "loss": 0.9481,
+      "step": 16457
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.309165587749453e-05,
+      "loss": 1.091,
+      "step": 16458
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.308272690745123e-05,
+      "loss": 0.9694,
+      "step": 16459
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.307379793740792e-05,
+      "loss": 1.0119,
+      "step": 16460
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.306486896736461e-05,
+      "loss": 1.0514,
+      "step": 16461
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.305593999732131e-05,
+      "loss": 1.056,
+      "step": 16462
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3047011027278004e-05,
+      "loss": 1.0898,
+      "step": 16463
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.303808205723471e-05,
+      "loss": 1.0078,
+      "step": 16464
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.30291530871914e-05,
+      "loss": 0.9658,
+      "step": 16465
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.302022411714809e-05,
+      "loss": 0.9498,
+      "step": 16466
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3011295147104786e-05,
+      "loss": 1.0832,
+      "step": 16467
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3002366177061477e-05,
+      "loss": 1.0167,
+      "step": 16468
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2993437207018174e-05,
+      "loss": 1.1112,
+      "step": 16469
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.298450823697487e-05,
+      "loss": 1.1217,
+      "step": 16470
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.297557926693156e-05,
+      "loss": 1.0273,
+      "step": 16471
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.296665029688825e-05,
+      "loss": 1.0853,
+      "step": 16472
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2957721326844956e-05,
+      "loss": 1.0863,
+      "step": 16473
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2948792356801647e-05,
+      "loss": 1.0831,
+      "step": 16474
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.293986338675834e-05,
+      "loss": 0.9343,
+      "step": 16475
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.293093441671504e-05,
+      "loss": 1.0247,
+      "step": 16476
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.292200544667173e-05,
+      "loss": 1.2479,
+      "step": 16477
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.291307647662842e-05,
+      "loss": 1.003,
+      "step": 16478
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.290414750658512e-05,
+      "loss": 1.0944,
+      "step": 16479
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.289521853654181e-05,
+      "loss": 1.082,
+      "step": 16480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.288628956649851e-05,
+      "loss": 0.9258,
+      "step": 16481
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2877360596455204e-05,
+      "loss": 0.9544,
+      "step": 16482
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2868431626411895e-05,
+      "loss": 1.1031,
+      "step": 16483
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2859502656368585e-05,
+      "loss": 0.9828,
+      "step": 16484
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.285057368632529e-05,
+      "loss": 1.0072,
+      "step": 16485
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.284164471628198e-05,
+      "loss": 1.0128,
+      "step": 16486
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.283271574623867e-05,
+      "loss": 1.2636,
+      "step": 16487
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2823786776195374e-05,
+      "loss": 1.1454,
+      "step": 16488
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2814857806152065e-05,
+      "loss": 1.1504,
+      "step": 16489
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2805928836108755e-05,
+      "loss": 1.0936,
+      "step": 16490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.279699986606545e-05,
+      "loss": 1.0048,
+      "step": 16491
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.278807089602214e-05,
+      "loss": 0.9654,
+      "step": 16492
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2779141925978833e-05,
+      "loss": 1.0985,
+      "step": 16493
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.277021295593554e-05,
+      "loss": 1.0709,
+      "step": 16494
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.276128398589223e-05,
+      "loss": 1.0409,
+      "step": 16495
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.275235501584892e-05,
+      "loss": 1.0145,
+      "step": 16496
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.274342604580562e-05,
+      "loss": 1.0941,
+      "step": 16497
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.273449707576231e-05,
+      "loss": 1.0496,
+      "step": 16498
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2725568105719e-05,
+      "loss": 1.0135,
+      "step": 16499
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.271663913567571e-05,
+      "loss": 1.0226,
+      "step": 16500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.27077101656324e-05,
+      "loss": 1.0655,
+      "step": 16501
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.269878119558909e-05,
+      "loss": 1.1056,
+      "step": 16502
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2689852225545786e-05,
+      "loss": 1.0575,
+      "step": 16503
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2680923255502476e-05,
+      "loss": 1.0337,
+      "step": 16504
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2671994285459167e-05,
+      "loss": 0.9226,
+      "step": 16505
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.266306531541587e-05,
+      "loss": 1.0334,
+      "step": 16506
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.265413634537256e-05,
+      "loss": 1.0024,
+      "step": 16507
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.264520737532925e-05,
+      "loss": 1.0743,
+      "step": 16508
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2636278405285956e-05,
+      "loss": 0.9947,
+      "step": 16509
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2627349435242646e-05,
+      "loss": 1.0158,
+      "step": 16510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.261842046519935e-05,
+      "loss": 1.0305,
+      "step": 16511
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.260949149515604e-05,
+      "loss": 1.1169,
+      "step": 16512
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.260056252511273e-05,
+      "loss": 0.937,
+      "step": 16513
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.259163355506943e-05,
+      "loss": 1.0241,
+      "step": 16514
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.258270458502612e-05,
+      "loss": 0.9231,
+      "step": 16515
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.257377561498281e-05,
+      "loss": 0.8879,
+      "step": 16516
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.256484664493951e-05,
+      "loss": 1.1458,
+      "step": 16517
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2555917674896204e-05,
+      "loss": 1.1605,
+      "step": 16518
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2546988704852894e-05,
+      "loss": 0.9503,
+      "step": 16519
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.25380597348096e-05,
+      "loss": 0.9998,
+      "step": 16520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.252913076476629e-05,
+      "loss": 1.1208,
+      "step": 16521
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.252020179472298e-05,
+      "loss": 1.1095,
+      "step": 16522
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.251127282467968e-05,
+      "loss": 1.0483,
+      "step": 16523
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2502343854636374e-05,
+      "loss": 1.0531,
+      "step": 16524
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2493414884593064e-05,
+      "loss": 0.9723,
+      "step": 16525
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.248448591454976e-05,
+      "loss": 1.037,
+      "step": 16526
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.247555694450645e-05,
+      "loss": 0.9132,
+      "step": 16527
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.246662797446314e-05,
+      "loss": 0.9672,
+      "step": 16528
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2457699004419847e-05,
+      "loss": 0.9967,
+      "step": 16529
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.244877003437654e-05,
+      "loss": 1.0016,
+      "step": 16530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.243984106433323e-05,
+      "loss": 1.1082,
+      "step": 16531
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.243091209428993e-05,
+      "loss": 1.1459,
+      "step": 16532
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.242198312424662e-05,
+      "loss": 0.9905,
+      "step": 16533
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.241305415420331e-05,
+      "loss": 1.0676,
+      "step": 16534
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2404125184160017e-05,
+      "loss": 1.1409,
+      "step": 16535
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.239519621411671e-05,
+      "loss": 1.0353,
+      "step": 16536
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.23862672440734e-05,
+      "loss": 1.0505,
+      "step": 16537
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2377338274030095e-05,
+      "loss": 1.053,
+      "step": 16538
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2368409303986785e-05,
+      "loss": 1.0051,
+      "step": 16539
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2359480333943476e-05,
+      "loss": 1.0696,
+      "step": 16540
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.235055136390018e-05,
+      "loss": 1.0063,
+      "step": 16541
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.234162239385687e-05,
+      "loss": 0.9794,
+      "step": 16542
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.233269342381356e-05,
+      "loss": 1.1691,
+      "step": 16543
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2323764453770265e-05,
+      "loss": 0.8826,
+      "step": 16544
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2314835483726955e-05,
+      "loss": 0.9427,
+      "step": 16545
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2305906513683646e-05,
+      "loss": 1.0859,
+      "step": 16546
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.229697754364035e-05,
+      "loss": 1.0015,
+      "step": 16547
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.228804857359704e-05,
+      "loss": 1.0949,
+      "step": 16548
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.227911960355373e-05,
+      "loss": 1.0728,
+      "step": 16549
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.227019063351043e-05,
+      "loss": 1.0238,
+      "step": 16550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.226126166346712e-05,
+      "loss": 1.1206,
+      "step": 16551
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.225233269342381e-05,
+      "loss": 1.0433,
+      "step": 16552
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.224340372338051e-05,
+      "loss": 1.0543,
+      "step": 16553
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2234474753337203e-05,
+      "loss": 1.2547,
+      "step": 16554
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2225545783293894e-05,
+      "loss": 1.0267,
+      "step": 16555
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.22166168132506e-05,
+      "loss": 1.0507,
+      "step": 16556
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.220768784320729e-05,
+      "loss": 0.9824,
+      "step": 16557
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.219875887316398e-05,
+      "loss": 0.997,
+      "step": 16558
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2189829903120676e-05,
+      "loss": 0.9455,
+      "step": 16559
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2180900933077373e-05,
+      "loss": 1.0847,
+      "step": 16560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.217197196303407e-05,
+      "loss": 1.113,
+      "step": 16561
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.216304299299076e-05,
+      "loss": 1.0727,
+      "step": 16562
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.215411402294745e-05,
+      "loss": 1.0464,
+      "step": 16563
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2145185052904156e-05,
+      "loss": 0.9928,
+      "step": 16564
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2136256082860846e-05,
+      "loss": 1.0288,
+      "step": 16565
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2127327112817537e-05,
+      "loss": 1.0563,
+      "step": 16566
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.211839814277424e-05,
+      "loss": 1.0333,
+      "step": 16567
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.210946917273093e-05,
+      "loss": 0.8797,
+      "step": 16568
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.210054020268762e-05,
+      "loss": 1.2002,
+      "step": 16569
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.209161123264432e-05,
+      "loss": 0.9634,
+      "step": 16570
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.208268226260101e-05,
+      "loss": 1.0674,
+      "step": 16571
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.20737532925577e-05,
+      "loss": 1.0099,
+      "step": 16572
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2064824322514404e-05,
+      "loss": 1.0707,
+      "step": 16573
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2055895352471094e-05,
+      "loss": 0.9617,
+      "step": 16574
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2046966382427785e-05,
+      "loss": 1.0938,
+      "step": 16575
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.203803741238449e-05,
+      "loss": 1.1384,
+      "step": 16576
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.202910844234118e-05,
+      "loss": 1.0368,
+      "step": 16577
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.202017947229787e-05,
+      "loss": 1.0346,
+      "step": 16578
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2011250502254574e-05,
+      "loss": 1.0261,
+      "step": 16579
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.2002321532211264e-05,
+      "loss": 0.9814,
+      "step": 16580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1993392562167955e-05,
+      "loss": 0.8842,
+      "step": 16581
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.198446359212465e-05,
+      "loss": 0.9418,
+      "step": 16582
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.197553462208134e-05,
+      "loss": 1.1419,
+      "step": 16583
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.196660565203803e-05,
+      "loss": 0.9648,
+      "step": 16584
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.195767668199474e-05,
+      "loss": 1.0407,
+      "step": 16585
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.194874771195143e-05,
+      "loss": 1.1275,
+      "step": 16586
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.193981874190812e-05,
+      "loss": 1.0268,
+      "step": 16587
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.193088977186482e-05,
+      "loss": 1.0162,
+      "step": 16588
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.192196080182151e-05,
+      "loss": 1.0706,
+      "step": 16589
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.19130318317782e-05,
+      "loss": 1.1278,
+      "step": 16590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.190410286173491e-05,
+      "loss": 1.0622,
+      "step": 16591
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.18951738916916e-05,
+      "loss": 1.0222,
+      "step": 16592
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.188624492164829e-05,
+      "loss": 1.0329,
+      "step": 16593
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1877315951604985e-05,
+      "loss": 0.9676,
+      "step": 16594
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1868386981561676e-05,
+      "loss": 0.9687,
+      "step": 16595
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1859458011518366e-05,
+      "loss": 0.9672,
+      "step": 16596
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.185052904147507e-05,
+      "loss": 0.9976,
+      "step": 16597
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.184160007143176e-05,
+      "loss": 0.9625,
+      "step": 16598
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.183267110138845e-05,
+      "loss": 0.9251,
+      "step": 16599
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1823742131345155e-05,
+      "loss": 1.0313,
+      "step": 16600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1814813161301846e-05,
+      "loss": 1.0791,
+      "step": 16601
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1805884191258536e-05,
+      "loss": 1.0633,
+      "step": 16602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.179695522121524e-05,
+      "loss": 0.9504,
+      "step": 16603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.178802625117193e-05,
+      "loss": 1.0133,
+      "step": 16604
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.177909728112862e-05,
+      "loss": 1.0771,
+      "step": 16605
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.177016831108532e-05,
+      "loss": 1.0542,
+      "step": 16606
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.176123934104201e-05,
+      "loss": 1.0147,
+      "step": 16607
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.17523103709987e-05,
+      "loss": 1.0634,
+      "step": 16608
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1743381400955403e-05,
+      "loss": 1.1249,
+      "step": 16609
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1734452430912094e-05,
+      "loss": 0.9109,
+      "step": 16610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.17255234608688e-05,
+      "loss": 0.9548,
+      "step": 16611
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.171659449082549e-05,
+      "loss": 1.0153,
+      "step": 16612
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.170766552078218e-05,
+      "loss": 0.9526,
+      "step": 16613
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.169873655073888e-05,
+      "loss": 0.9117,
+      "step": 16614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1689807580695573e-05,
+      "loss": 1.0395,
+      "step": 16615
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.1680878610652264e-05,
+      "loss": 0.9054,
+      "step": 16616
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.167194964060896e-05,
+      "loss": 0.9969,
+      "step": 16617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.166302067056565e-05,
+      "loss": 0.969,
+      "step": 16618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.165409170052234e-05,
+      "loss": 1.0036,
+      "step": 16619
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1645162730479046e-05,
+      "loss": 0.9876,
+      "step": 16620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.163623376043574e-05,
+      "loss": 1.0473,
+      "step": 16621
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.162730479039243e-05,
+      "loss": 1.1907,
+      "step": 16622
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.161837582034913e-05,
+      "loss": 0.9773,
+      "step": 16623
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.160944685030582e-05,
+      "loss": 1.0241,
+      "step": 16624
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.160051788026251e-05,
+      "loss": 0.9741,
+      "step": 16625
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1591588910219216e-05,
+      "loss": 1.0515,
+      "step": 16626
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.158265994017591e-05,
+      "loss": 0.9464,
+      "step": 16627
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.15737309701326e-05,
+      "loss": 0.9758,
+      "step": 16628
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1564802000089294e-05,
+      "loss": 0.9985,
+      "step": 16629
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1555873030045985e-05,
+      "loss": 0.9982,
+      "step": 16630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1546944060002675e-05,
+      "loss": 1.1495,
+      "step": 16631
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.153801508995938e-05,
+      "loss": 1.0808,
+      "step": 16632
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.152908611991607e-05,
+      "loss": 0.9425,
+      "step": 16633
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.152015714987276e-05,
+      "loss": 1.2174,
+      "step": 16634
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1511228179829464e-05,
+      "loss": 0.965,
+      "step": 16635
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1502299209786155e-05,
+      "loss": 1.0047,
+      "step": 16636
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1493370239742845e-05,
+      "loss": 1.0549,
+      "step": 16637
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.148444126969954e-05,
+      "loss": 1.1409,
+      "step": 16638
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.147551229965624e-05,
+      "loss": 1.0365,
+      "step": 16639
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.146658332961293e-05,
+      "loss": 1.1509,
+      "step": 16640
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.145765435956963e-05,
+      "loss": 1.0779,
+      "step": 16641
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.144872538952632e-05,
+      "loss": 1.1129,
+      "step": 16642
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.143979641948301e-05,
+      "loss": 1.0103,
+      "step": 16643
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.143086744943971e-05,
+      "loss": 1.1625,
+      "step": 16644
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.14219384793964e-05,
+      "loss": 1.0546,
+      "step": 16645
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1413009509353094e-05,
+      "loss": 1.0164,
+      "step": 16646
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.14040805393098e-05,
+      "loss": 1.1061,
+      "step": 16647
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.139515156926649e-05,
+      "loss": 0.9545,
+      "step": 16648
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.138622259922318e-05,
+      "loss": 0.9162,
+      "step": 16649
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1377293629179876e-05,
+      "loss": 0.9447,
+      "step": 16650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.136836465913657e-05,
+      "loss": 1.0778,
+      "step": 16651
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1359435689093264e-05,
+      "loss": 1.0068,
+      "step": 16652
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.135050671904996e-05,
+      "loss": 1.1349,
+      "step": 16653
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.134157774900665e-05,
+      "loss": 0.995,
+      "step": 16654
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.133264877896334e-05,
+      "loss": 0.9278,
+      "step": 16655
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1323719808920046e-05,
+      "loss": 1.021,
+      "step": 16656
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1314790838876736e-05,
+      "loss": 0.9548,
+      "step": 16657
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.130586186883344e-05,
+      "loss": 1.0473,
+      "step": 16658
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.129693289879013e-05,
+      "loss": 0.9659,
+      "step": 16659
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.128800392874682e-05,
+      "loss": 0.9303,
+      "step": 16660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.127907495870352e-05,
+      "loss": 0.9537,
+      "step": 16661
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.127014598866021e-05,
+      "loss": 1.0486,
+      "step": 16662
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.12612170186169e-05,
+      "loss": 1.1064,
+      "step": 16663
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1252288048573603e-05,
+      "loss": 1.0461,
+      "step": 16664
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1243359078530294e-05,
+      "loss": 0.9521,
+      "step": 16665
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1234430108486984e-05,
+      "loss": 1.0905,
+      "step": 16666
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.122550113844369e-05,
+      "loss": 1.0338,
+      "step": 16667
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.121657216840038e-05,
+      "loss": 1.0595,
+      "step": 16668
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.120764319835707e-05,
+      "loss": 1.1053,
+      "step": 16669
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1198714228313773e-05,
+      "loss": 1.1163,
+      "step": 16670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1189785258270464e-05,
+      "loss": 0.9892,
+      "step": 16671
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1180856288227154e-05,
+      "loss": 0.9989,
+      "step": 16672
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.117192731818385e-05,
+      "loss": 1.0502,
+      "step": 16673
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.116299834814054e-05,
+      "loss": 1.2064,
+      "step": 16674
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.115406937809723e-05,
+      "loss": 1.0299,
+      "step": 16675
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.114514040805394e-05,
+      "loss": 1.0983,
+      "step": 16676
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.113621143801063e-05,
+      "loss": 1.004,
+      "step": 16677
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.112728246796732e-05,
+      "loss": 1.0528,
+      "step": 16678
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.111835349792402e-05,
+      "loss": 1.1865,
+      "step": 16679
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.110942452788071e-05,
+      "loss": 0.9855,
+      "step": 16680
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.11004955578374e-05,
+      "loss": 1.0324,
+      "step": 16681
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.109156658779411e-05,
+      "loss": 0.958,
+      "step": 16682
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.10826376177508e-05,
+      "loss": 0.9965,
+      "step": 16683
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.107370864770749e-05,
+      "loss": 0.9659,
+      "step": 16684
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1064779677664185e-05,
+      "loss": 0.8732,
+      "step": 16685
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1055850707620875e-05,
+      "loss": 1.0075,
+      "step": 16686
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1046921737577566e-05,
+      "loss": 1.082,
+      "step": 16687
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.103799276753427e-05,
+      "loss": 1.1291,
+      "step": 16688
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.102906379749096e-05,
+      "loss": 0.9158,
+      "step": 16689
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.102013482744765e-05,
+      "loss": 0.8838,
+      "step": 16690
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1011205857404355e-05,
+      "loss": 1.0013,
+      "step": 16691
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1002276887361045e-05,
+      "loss": 0.9592,
+      "step": 16692
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0993347917317736e-05,
+      "loss": 1.1252,
+      "step": 16693
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.098441894727444e-05,
+      "loss": 1.0331,
+      "step": 16694
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.097548997723113e-05,
+      "loss": 1.1135,
+      "step": 16695
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.096656100718782e-05,
+      "loss": 0.9118,
+      "step": 16696
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.095763203714452e-05,
+      "loss": 1.0239,
+      "step": 16697
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.094870306710121e-05,
+      "loss": 1.0391,
+      "step": 16698
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.09397740970579e-05,
+      "loss": 1.0048,
+      "step": 16699
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.09308451270146e-05,
+      "loss": 1.054,
+      "step": 16700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0921916156971294e-05,
+      "loss": 0.9281,
+      "step": 16701
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0912987186927984e-05,
+      "loss": 1.1538,
+      "step": 16702
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.090405821688469e-05,
+      "loss": 1.0981,
+      "step": 16703
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.089512924684138e-05,
+      "loss": 1.0989,
+      "step": 16704
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.088620027679807e-05,
+      "loss": 1.0815,
+      "step": 16705
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.087727130675477e-05,
+      "loss": 1.0906,
+      "step": 16706
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0868342336711464e-05,
+      "loss": 1.1204,
+      "step": 16707
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.085941336666816e-05,
+      "loss": 1.012,
+      "step": 16708
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.085048439662485e-05,
+      "loss": 0.9747,
+      "step": 16709
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.084155542658154e-05,
+      "loss": 0.9336,
+      "step": 16710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0832626456538246e-05,
+      "loss": 1.0178,
+      "step": 16711
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0823697486494936e-05,
+      "loss": 1.041,
+      "step": 16712
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.081476851645163e-05,
+      "loss": 0.9397,
+      "step": 16713
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.080583954640833e-05,
+      "loss": 1.1031,
+      "step": 16714
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.079691057636502e-05,
+      "loss": 1.0777,
+      "step": 16715
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.078798160632171e-05,
+      "loss": 1.0908,
+      "step": 16716
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0779052636278416e-05,
+      "loss": 1.0398,
+      "step": 16717
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0770123666235106e-05,
+      "loss": 1.0204,
+      "step": 16718
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.07611946961918e-05,
+      "loss": 0.9818,
+      "step": 16719
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0752265726148494e-05,
+      "loss": 1.1155,
+      "step": 16720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0743336756105184e-05,
+      "loss": 1.1317,
+      "step": 16721
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0734407786061875e-05,
+      "loss": 0.9685,
+      "step": 16722
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.072547881601858e-05,
+      "loss": 0.9526,
+      "step": 16723
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.071654984597527e-05,
+      "loss": 0.9881,
+      "step": 16724
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.070762087593196e-05,
+      "loss": 1.0757,
+      "step": 16725
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0698691905888664e-05,
+      "loss": 0.9739,
+      "step": 16726
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0689762935845354e-05,
+      "loss": 1.074,
+      "step": 16727
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0680833965802045e-05,
+      "loss": 1.1133,
+      "step": 16728
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.067190499575874e-05,
+      "loss": 0.9289,
+      "step": 16729
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.066297602571544e-05,
+      "loss": 0.9928,
+      "step": 16730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.065404705567213e-05,
+      "loss": 0.9949,
+      "step": 16731
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.064511808562883e-05,
+      "loss": 0.9769,
+      "step": 16732
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.063618911558552e-05,
+      "loss": 1.0366,
+      "step": 16733
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.062726014554221e-05,
+      "loss": 1.0391,
+      "step": 16734
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.061833117549891e-05,
+      "loss": 1.0848,
+      "step": 16735
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.06094022054556e-05,
+      "loss": 1.0596,
+      "step": 16736
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.060047323541229e-05,
+      "loss": 1.0199,
+      "step": 16737
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0591544265369e-05,
+      "loss": 1.1081,
+      "step": 16738
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.058261529532569e-05,
+      "loss": 1.0179,
+      "step": 16739
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.057368632528238e-05,
+      "loss": 0.9648,
+      "step": 16740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0564757355239075e-05,
+      "loss": 1.1105,
+      "step": 16741
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0555828385195766e-05,
+      "loss": 0.977,
+      "step": 16742
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.054689941515246e-05,
+      "loss": 0.9451,
+      "step": 16743
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.053797044510916e-05,
+      "loss": 0.9508,
+      "step": 16744
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.052904147506585e-05,
+      "loss": 1.1505,
+      "step": 16745
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.052011250502254e-05,
+      "loss": 0.9772,
+      "step": 16746
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0511183534979245e-05,
+      "loss": 1.1311,
+      "step": 16747
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0502254564935936e-05,
+      "loss": 1.0271,
+      "step": 16748
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0493325594892626e-05,
+      "loss": 1.1196,
+      "step": 16749
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.048439662484933e-05,
+      "loss": 1.0804,
+      "step": 16750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.047546765480602e-05,
+      "loss": 0.9715,
+      "step": 16751
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.046653868476271e-05,
+      "loss": 1.0008,
+      "step": 16752
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.045760971471941e-05,
+      "loss": 1.0783,
+      "step": 16753
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.04486807446761e-05,
+      "loss": 1.0985,
+      "step": 16754
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0439751774632796e-05,
+      "loss": 1.0709,
+      "step": 16755
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0430822804589494e-05,
+      "loss": 1.1122,
+      "step": 16756
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0421893834546184e-05,
+      "loss": 1.1185,
+      "step": 16757
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.041296486450289e-05,
+      "loss": 1.0194,
+      "step": 16758
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.040403589445958e-05,
+      "loss": 0.9214,
+      "step": 16759
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.039510692441627e-05,
+      "loss": 1.0735,
+      "step": 16760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.038617795437297e-05,
+      "loss": 0.9957,
+      "step": 16761
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0377248984329664e-05,
+      "loss": 1.0302,
+      "step": 16762
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0368320014286354e-05,
+      "loss": 1.1246,
+      "step": 16763
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.035939104424305e-05,
+      "loss": 1.0734,
+      "step": 16764
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.035046207419974e-05,
+      "loss": 0.9926,
+      "step": 16765
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.034153310415643e-05,
+      "loss": 1.143,
+      "step": 16766
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0332604134113136e-05,
+      "loss": 0.9572,
+      "step": 16767
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.032367516406983e-05,
+      "loss": 0.9063,
+      "step": 16768
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.031474619402652e-05,
+      "loss": 1.1059,
+      "step": 16769
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.030581722398322e-05,
+      "loss": 0.9447,
+      "step": 16770
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.029688825393991e-05,
+      "loss": 1.0306,
+      "step": 16771
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.02879592838966e-05,
+      "loss": 0.9901,
+      "step": 16772
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0279030313853306e-05,
+      "loss": 1.0119,
+      "step": 16773
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.027010134381e-05,
+      "loss": 1.1248,
+      "step": 16774
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.026117237376669e-05,
+      "loss": 0.9435,
+      "step": 16775
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0252243403723385e-05,
+      "loss": 1.0096,
+      "step": 16776
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0243314433680075e-05,
+      "loss": 1.0769,
+      "step": 16777
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0234385463636765e-05,
+      "loss": 0.9984,
+      "step": 16778
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.022545649359347e-05,
+      "loss": 1.0006,
+      "step": 16779
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.021652752355016e-05,
+      "loss": 1.0151,
+      "step": 16780
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.020759855350685e-05,
+      "loss": 1.0449,
+      "step": 16781
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0198669583463555e-05,
+      "loss": 0.875,
+      "step": 16782
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0189740613420245e-05,
+      "loss": 1.0508,
+      "step": 16783
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0180811643376935e-05,
+      "loss": 0.9618,
+      "step": 16784
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.017188267333364e-05,
+      "loss": 1.0796,
+      "step": 16785
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.016295370329033e-05,
+      "loss": 1.1654,
+      "step": 16786
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.015402473324702e-05,
+      "loss": 1.2208,
+      "step": 16787
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.014509576320372e-05,
+      "loss": 1.113,
+      "step": 16788
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.013616679316041e-05,
+      "loss": 1.1212,
+      "step": 16789
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.01272378231171e-05,
+      "loss": 1.151,
+      "step": 16790
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.01183088530738e-05,
+      "loss": 1.0475,
+      "step": 16791
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.010937988303049e-05,
+      "loss": 0.99,
+      "step": 16792
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0100450912987184e-05,
+      "loss": 1.0629,
+      "step": 16793
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.009152194294389e-05,
+      "loss": 0.9701,
+      "step": 16794
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.008259297290058e-05,
+      "loss": 1.0282,
+      "step": 16795
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.007366400285727e-05,
+      "loss": 1.1344,
+      "step": 16796
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.006473503281397e-05,
+      "loss": 1.0181,
+      "step": 16797
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.005580606277066e-05,
+      "loss": 0.9838,
+      "step": 16798
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0046877092727354e-05,
+      "loss": 0.9819,
+      "step": 16799
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.003794812268405e-05,
+      "loss": 1.1019,
+      "step": 16800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.002901915264074e-05,
+      "loss": 0.9609,
+      "step": 16801
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.002009018259743e-05,
+      "loss": 1.0698,
+      "step": 16802
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0011161212554136e-05,
+      "loss": 1.1166,
+      "step": 16803
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0002232242510826e-05,
+      "loss": 0.9925,
+      "step": 16804
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9993303272467524e-05,
+      "loss": 1.0285,
+      "step": 16805
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.998437430242422e-05,
+      "loss": 1.2074,
+      "step": 16806
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.997544533238091e-05,
+      "loss": 1.0324,
+      "step": 16807
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.996651636233761e-05,
+      "loss": 1.0767,
+      "step": 16808
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9957587392294306e-05,
+      "loss": 1.0383,
+      "step": 16809
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9948658422250996e-05,
+      "loss": 1.0365,
+      "step": 16810
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.993972945220769e-05,
+      "loss": 1.0596,
+      "step": 16811
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9930800482164384e-05,
+      "loss": 1.1323,
+      "step": 16812
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9921871512121075e-05,
+      "loss": 0.979,
+      "step": 16813
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.991294254207777e-05,
+      "loss": 0.9576,
+      "step": 16814
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.990401357203447e-05,
+      "loss": 0.9629,
+      "step": 16815
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.989508460199116e-05,
+      "loss": 1.1526,
+      "step": 16816
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.988615563194786e-05,
+      "loss": 1.117,
+      "step": 16817
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9877226661904554e-05,
+      "loss": 1.052,
+      "step": 16818
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.986829769186125e-05,
+      "loss": 1.013,
+      "step": 16819
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.985936872181794e-05,
+      "loss": 1.1284,
+      "step": 16820
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.985043975177464e-05,
+      "loss": 1.1133,
+      "step": 16821
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.984151078173133e-05,
+      "loss": 1.0821,
+      "step": 16822
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.983258181168802e-05,
+      "loss": 1.017,
+      "step": 16823
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.982365284164472e-05,
+      "loss": 1.1298,
+      "step": 16824
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9814723871601415e-05,
+      "loss": 1.0022,
+      "step": 16825
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9805794901558105e-05,
+      "loss": 1.0562,
+      "step": 16826
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.97968659315148e-05,
+      "loss": 0.9502,
+      "step": 16827
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.97879369614715e-05,
+      "loss": 1.2128,
+      "step": 16828
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.977900799142819e-05,
+      "loss": 0.9736,
+      "step": 16829
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.977007902138489e-05,
+      "loss": 1.107,
+      "step": 16830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9761150051341585e-05,
+      "loss": 1.0798,
+      "step": 16831
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9752221081298275e-05,
+      "loss": 0.8947,
+      "step": 16832
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9743292111254966e-05,
+      "loss": 1.0343,
+      "step": 16833
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.973436314121166e-05,
+      "loss": 1.0563,
+      "step": 16834
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.972543417116835e-05,
+      "loss": 1.0506,
+      "step": 16835
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.971650520112505e-05,
+      "loss": 1.0166,
+      "step": 16836
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.970757623108175e-05,
+      "loss": 0.9686,
+      "step": 16837
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.969864726103844e-05,
+      "loss": 1.0773,
+      "step": 16838
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.9689718290995136e-05,
+      "loss": 1.0146,
+      "step": 16839
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.968078932095183e-05,
+      "loss": 1.0122,
+      "step": 16840
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.967186035090853e-05,
+      "loss": 0.9501,
+      "step": 16841
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.966293138086522e-05,
+      "loss": 1.1004,
+      "step": 16842
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.965400241082192e-05,
+      "loss": 1.1387,
+      "step": 16843
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.964507344077861e-05,
+      "loss": 0.8643,
+      "step": 16844
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.96361444707353e-05,
+      "loss": 1.127,
+      "step": 16845
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9627215500691996e-05,
+      "loss": 0.9226,
+      "step": 16846
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.961828653064869e-05,
+      "loss": 0.9989,
+      "step": 16847
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9609357560605384e-05,
+      "loss": 1.065,
+      "step": 16848
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.960042859056208e-05,
+      "loss": 1.2025,
+      "step": 16849
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.959149962051878e-05,
+      "loss": 1.1337,
+      "step": 16850
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.958257065047547e-05,
+      "loss": 0.9517,
+      "step": 16851
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9573641680432166e-05,
+      "loss": 1.0328,
+      "step": 16852
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.956471271038886e-05,
+      "loss": 1.0491,
+      "step": 16853
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9555783740345554e-05,
+      "loss": 1.0498,
+      "step": 16854
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.954685477030225e-05,
+      "loss": 0.9905,
+      "step": 16855
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.953792580025894e-05,
+      "loss": 0.9866,
+      "step": 16856
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.952899683021563e-05,
+      "loss": 0.9874,
+      "step": 16857
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.952006786017233e-05,
+      "loss": 1.1579,
+      "step": 16858
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9511138890129026e-05,
+      "loss": 1.0826,
+      "step": 16859
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.950220992008572e-05,
+      "loss": 0.9604,
+      "step": 16860
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9493280950042414e-05,
+      "loss": 1.001,
+      "step": 16861
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.948435197999911e-05,
+      "loss": 1.0247,
+      "step": 16862
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.94754230099558e-05,
+      "loss": 1.0781,
+      "step": 16863
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.94664940399125e-05,
+      "loss": 0.9801,
+      "step": 16864
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9457565069869196e-05,
+      "loss": 0.9792,
+      "step": 16865
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.944863609982589e-05,
+      "loss": 0.9887,
+      "step": 16866
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9439707129782584e-05,
+      "loss": 1.0176,
+      "step": 16867
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9430778159739275e-05,
+      "loss": 0.9935,
+      "step": 16868
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.942184918969597e-05,
+      "loss": 0.9967,
+      "step": 16869
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.941292021965266e-05,
+      "loss": 1.0213,
+      "step": 16870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.940399124960936e-05,
+      "loss": 1.0919,
+      "step": 16871
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.939506227956606e-05,
+      "loss": 1.0669,
+      "step": 16872
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.938613330952275e-05,
+      "loss": 0.9632,
+      "step": 16873
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9377204339479445e-05,
+      "loss": 0.9444,
+      "step": 16874
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.936827536943614e-05,
+      "loss": 0.939,
+      "step": 16875
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.935934639939283e-05,
+      "loss": 0.9771,
+      "step": 16876
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.935041742934953e-05,
+      "loss": 1.0592,
+      "step": 16877
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.934148845930622e-05,
+      "loss": 0.942,
+      "step": 16878
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.933255948926291e-05,
+      "loss": 1.0925,
+      "step": 16879
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.932363051921961e-05,
+      "loss": 1.0423,
+      "step": 16880
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9314701549176305e-05,
+      "loss": 0.9607,
+      "step": 16881
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9305772579132996e-05,
+      "loss": 1.0494,
+      "step": 16882
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.929684360908969e-05,
+      "loss": 1.0881,
+      "step": 16883
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.928791463904639e-05,
+      "loss": 1.0028,
+      "step": 16884
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.927898566900308e-05,
+      "loss": 1.0639,
+      "step": 16885
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.927005669895978e-05,
+      "loss": 1.0073,
+      "step": 16886
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9261127728916475e-05,
+      "loss": 1.0909,
+      "step": 16887
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9252198758873166e-05,
+      "loss": 0.9879,
+      "step": 16888
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.924326978882986e-05,
+      "loss": 0.9025,
+      "step": 16889
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.923434081878655e-05,
+      "loss": 0.9468,
+      "step": 16890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.922541184874325e-05,
+      "loss": 0.9303,
+      "step": 16891
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.921648287869994e-05,
+      "loss": 1.1251,
+      "step": 16892
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.920755390865664e-05,
+      "loss": 0.9742,
+      "step": 16893
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9198624938613336e-05,
+      "loss": 1.0189,
+      "step": 16894
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9189695968570026e-05,
+      "loss": 0.9632,
+      "step": 16895
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.918076699852672e-05,
+      "loss": 0.9826,
+      "step": 16896
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.917183802848342e-05,
+      "loss": 1.0046,
+      "step": 16897
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.916290905844011e-05,
+      "loss": 1.0583,
+      "step": 16898
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.915398008839681e-05,
+      "loss": 0.9037,
+      "step": 16899
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9145051118353506e-05,
+      "loss": 0.9151,
+      "step": 16900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9136122148310196e-05,
+      "loss": 0.967,
+      "step": 16901
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9127193178266886e-05,
+      "loss": 0.9807,
+      "step": 16902
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9118264208223584e-05,
+      "loss": 1.0261,
+      "step": 16903
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9109335238180274e-05,
+      "loss": 1.113,
+      "step": 16904
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.910040626813697e-05,
+      "loss": 1.0172,
+      "step": 16905
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.909147729809367e-05,
+      "loss": 1.171,
+      "step": 16906
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.908254832805036e-05,
+      "loss": 0.9796,
+      "step": 16907
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9073619358007056e-05,
+      "loss": 0.8785,
+      "step": 16908
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9064690387963754e-05,
+      "loss": 0.9285,
+      "step": 16909
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9055761417920444e-05,
+      "loss": 1.0323,
+      "step": 16910
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.904683244787714e-05,
+      "loss": 0.9809,
+      "step": 16911
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.903790347783383e-05,
+      "loss": 1.0572,
+      "step": 16912
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.902897450779053e-05,
+      "loss": 1.0128,
+      "step": 16913
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.902004553774722e-05,
+      "loss": 0.9607,
+      "step": 16914
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.901111656770392e-05,
+      "loss": 1.0852,
+      "step": 16915
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9002187597660614e-05,
+      "loss": 0.9768,
+      "step": 16916
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8993258627617305e-05,
+      "loss": 0.9761,
+      "step": 16917
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8984329657574e-05,
+      "loss": 0.9215,
+      "step": 16918
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.89754006875307e-05,
+      "loss": 1.0731,
+      "step": 16919
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.896647171748739e-05,
+      "loss": 0.9727,
+      "step": 16920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.895754274744409e-05,
+      "loss": 0.9554,
+      "step": 16921
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8948613777400784e-05,
+      "loss": 1.135,
+      "step": 16922
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8939684807357475e-05,
+      "loss": 1.0431,
+      "step": 16923
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8930755837314165e-05,
+      "loss": 1.0062,
+      "step": 16924
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.892182686727086e-05,
+      "loss": 1.1441,
+      "step": 16925
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.891289789722755e-05,
+      "loss": 1.0091,
+      "step": 16926
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.890396892718425e-05,
+      "loss": 1.192,
+      "step": 16927
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.889503995714095e-05,
+      "loss": 0.9182,
+      "step": 16928
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.888611098709764e-05,
+      "loss": 0.8702,
+      "step": 16929
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8877182017054335e-05,
+      "loss": 1.1296,
+      "step": 16930
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.886825304701103e-05,
+      "loss": 0.9111,
+      "step": 16931
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.885932407696772e-05,
+      "loss": 1.0235,
+      "step": 16932
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.885039510692442e-05,
+      "loss": 0.9194,
+      "step": 16933
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.884146613688112e-05,
+      "loss": 1.1208,
+      "step": 16934
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.883253716683781e-05,
+      "loss": 1.0594,
+      "step": 16935
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.88236081967945e-05,
+      "loss": 0.9559,
+      "step": 16936
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8814679226751196e-05,
+      "loss": 0.9513,
+      "step": 16937
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8805750256707886e-05,
+      "loss": 1.0447,
+      "step": 16938
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.879682128666458e-05,
+      "loss": 0.9286,
+      "step": 16939
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.878789231662128e-05,
+      "loss": 0.9484,
+      "step": 16940
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.877896334657798e-05,
+      "loss": 1.1858,
+      "step": 16941
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.877003437653467e-05,
+      "loss": 0.9762,
+      "step": 16942
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8761105406491366e-05,
+      "loss": 0.9896,
+      "step": 16943
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.875217643644806e-05,
+      "loss": 1.1067,
+      "step": 16944
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.874324746640475e-05,
+      "loss": 1.0012,
+      "step": 16945
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.873431849636145e-05,
+      "loss": 0.9666,
+      "step": 16946
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.872538952631814e-05,
+      "loss": 0.9277,
+      "step": 16947
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.871646055627483e-05,
+      "loss": 1.1131,
+      "step": 16948
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.870753158623153e-05,
+      "loss": 1.0051,
+      "step": 16949
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8698602616188226e-05,
+      "loss": 0.9956,
+      "step": 16950
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8689673646144917e-05,
+      "loss": 0.9667,
+      "step": 16951
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8680744676101614e-05,
+      "loss": 0.9874,
+      "step": 16952
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.867181570605831e-05,
+      "loss": 0.9911,
+      "step": 16953
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8662886736015e-05,
+      "loss": 1.1246,
+      "step": 16954
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.86539577659717e-05,
+      "loss": 0.9967,
+      "step": 16955
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8645028795928396e-05,
+      "loss": 1.1837,
+      "step": 16956
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8636099825885087e-05,
+      "loss": 0.8785,
+      "step": 16957
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8627170855841784e-05,
+      "loss": 1.0246,
+      "step": 16958
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8618241885798474e-05,
+      "loss": 0.9952,
+      "step": 16959
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8609312915755165e-05,
+      "loss": 0.9437,
+      "step": 16960
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.860038394571186e-05,
+      "loss": 1.0913,
+      "step": 16961
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.859145497566856e-05,
+      "loss": 1.0294,
+      "step": 16962
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.858252600562525e-05,
+      "loss": 0.9952,
+      "step": 16963
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.857359703558195e-05,
+      "loss": 1.1961,
+      "step": 16964
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8564668065538644e-05,
+      "loss": 1.022,
+      "step": 16965
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.855573909549534e-05,
+      "loss": 0.9799,
+      "step": 16966
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.854681012545203e-05,
+      "loss": 0.9569,
+      "step": 16967
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.853788115540873e-05,
+      "loss": 0.8886,
+      "step": 16968
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.852895218536542e-05,
+      "loss": 1.0344,
+      "step": 16969
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.852002321532211e-05,
+      "loss": 1.0159,
+      "step": 16970
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.851109424527881e-05,
+      "loss": 1.1076,
+      "step": 16971
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8502165275235505e-05,
+      "loss": 1.0345,
+      "step": 16972
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8493236305192195e-05,
+      "loss": 0.8759,
+      "step": 16973
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.848430733514889e-05,
+      "loss": 1.1145,
+      "step": 16974
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.847537836510559e-05,
+      "loss": 0.9395,
+      "step": 16975
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.846644939506228e-05,
+      "loss": 0.9903,
+      "step": 16976
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.845752042501898e-05,
+      "loss": 0.9671,
+      "step": 16977
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8448591454975675e-05,
+      "loss": 1.0794,
+      "step": 16978
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8439662484932365e-05,
+      "loss": 1.0127,
+      "step": 16979
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.843073351488906e-05,
+      "loss": 1.1011,
+      "step": 16980
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.842180454484575e-05,
+      "loss": 0.9101,
+      "step": 16981
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8412875574802443e-05,
+      "loss": 1.1087,
+      "step": 16982
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.840394660475914e-05,
+      "loss": 0.9753,
+      "step": 16983
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.839501763471584e-05,
+      "loss": 1.1387,
+      "step": 16984
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.838608866467253e-05,
+      "loss": 1.0041,
+      "step": 16985
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8377159694629226e-05,
+      "loss": 1.0285,
+      "step": 16986
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.836823072458592e-05,
+      "loss": 1.0104,
+      "step": 16987
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.835930175454262e-05,
+      "loss": 0.9871,
+      "step": 16988
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.835037278449931e-05,
+      "loss": 1.0416,
+      "step": 16989
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.834144381445601e-05,
+      "loss": 1.0243,
+      "step": 16990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8332514844412705e-05,
+      "loss": 1.041,
+      "step": 16991
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8323585874369396e-05,
+      "loss": 0.9217,
+      "step": 16992
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8314656904326086e-05,
+      "loss": 0.9883,
+      "step": 16993
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.830572793428278e-05,
+      "loss": 1.0868,
+      "step": 16994
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8296798964239474e-05,
+      "loss": 1.0463,
+      "step": 16995
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.828786999419617e-05,
+      "loss": 1.0175,
+      "step": 16996
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.827894102415287e-05,
+      "loss": 0.9886,
+      "step": 16997
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.827001205410956e-05,
+      "loss": 1.1016,
+      "step": 16998
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8261083084066256e-05,
+      "loss": 1.004,
+      "step": 16999
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.825215411402295e-05,
+      "loss": 0.9384,
+      "step": 17000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8243225143979644e-05,
+      "loss": 1.095,
+      "step": 17001
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.823429617393634e-05,
+      "loss": 1.0048,
+      "step": 17002
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.822536720389303e-05,
+      "loss": 1.1217,
+      "step": 17003
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.821643823384973e-05,
+      "loss": 1.2151,
+      "step": 17004
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.820750926380642e-05,
+      "loss": 1.101,
+      "step": 17005
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8198580293763117e-05,
+      "loss": 0.9744,
+      "step": 17006
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.818965132371981e-05,
+      "loss": 1.0849,
+      "step": 17007
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8180722353676504e-05,
+      "loss": 1.0119,
+      "step": 17008
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.81717933836332e-05,
+      "loss": 1.0063,
+      "step": 17009
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.816286441358989e-05,
+      "loss": 0.9821,
+      "step": 17010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.815393544354659e-05,
+      "loss": 1.0066,
+      "step": 17011
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8145006473503287e-05,
+      "loss": 0.953,
+      "step": 17012
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8136077503459984e-05,
+      "loss": 1.1806,
+      "step": 17013
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8127148533416674e-05,
+      "loss": 1.1243,
+      "step": 17014
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8118219563373365e-05,
+      "loss": 1.095,
+      "step": 17015
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.810929059333006e-05,
+      "loss": 1.1482,
+      "step": 17016
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.810036162328675e-05,
+      "loss": 0.9563,
+      "step": 17017
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.809143265324345e-05,
+      "loss": 1.0119,
+      "step": 17018
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.808250368320015e-05,
+      "loss": 1.1022,
+      "step": 17019
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.807357471315684e-05,
+      "loss": 1.1367,
+      "step": 17020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8064645743113535e-05,
+      "loss": 1.0187,
+      "step": 17021
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.805571677307023e-05,
+      "loss": 0.9987,
+      "step": 17022
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.804678780302692e-05,
+      "loss": 0.9931,
+      "step": 17023
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.803785883298362e-05,
+      "loss": 1.0375,
+      "step": 17024
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.802892986294032e-05,
+      "loss": 1.0532,
+      "step": 17025
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.802000089289701e-05,
+      "loss": 0.9962,
+      "step": 17026
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.80110719228537e-05,
+      "loss": 1.0264,
+      "step": 17027
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8002142952810395e-05,
+      "loss": 1.0023,
+      "step": 17028
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7993213982767086e-05,
+      "loss": 0.9996,
+      "step": 17029
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.798428501272378e-05,
+      "loss": 0.9383,
+      "step": 17030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.797535604268048e-05,
+      "loss": 1.0989,
+      "step": 17031
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.796642707263717e-05,
+      "loss": 1.0055,
+      "step": 17032
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.795749810259387e-05,
+      "loss": 1.0017,
+      "step": 17033
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7948569132550565e-05,
+      "loss": 0.9612,
+      "step": 17034
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7939640162507256e-05,
+      "loss": 1.0562,
+      "step": 17035
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.793071119246395e-05,
+      "loss": 0.9781,
+      "step": 17036
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.792178222242065e-05,
+      "loss": 1.071,
+      "step": 17037
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.791285325237734e-05,
+      "loss": 1.0719,
+      "step": 17038
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.790392428233403e-05,
+      "loss": 1.0405,
+      "step": 17039
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.789499531229073e-05,
+      "loss": 1.0273,
+      "step": 17040
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7886066342247426e-05,
+      "loss": 0.9767,
+      "step": 17041
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7877137372204116e-05,
+      "loss": 0.9453,
+      "step": 17042
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7868208402160813e-05,
+      "loss": 1.0384,
+      "step": 17043
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.785927943211751e-05,
+      "loss": 1.0212,
+      "step": 17044
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.78503504620742e-05,
+      "loss": 0.9885,
+      "step": 17045
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.78414214920309e-05,
+      "loss": 0.9732,
+      "step": 17046
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7832492521987596e-05,
+      "loss": 0.9988,
+      "step": 17047
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7823563551944286e-05,
+      "loss": 1.0967,
+      "step": 17048
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.781463458190098e-05,
+      "loss": 1.0881,
+      "step": 17049
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7805705611857674e-05,
+      "loss": 1.1202,
+      "step": 17050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7796776641814364e-05,
+      "loss": 0.9961,
+      "step": 17051
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.778784767177106e-05,
+      "loss": 0.8784,
+      "step": 17052
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.777891870172776e-05,
+      "loss": 1.1526,
+      "step": 17053
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.776998973168445e-05,
+      "loss": 1.0022,
+      "step": 17054
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.776106076164115e-05,
+      "loss": 1.1057,
+      "step": 17055
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7752131791597844e-05,
+      "loss": 0.9244,
+      "step": 17056
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7743202821554534e-05,
+      "loss": 1.017,
+      "step": 17057
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.773427385151123e-05,
+      "loss": 1.1759,
+      "step": 17058
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.772534488146793e-05,
+      "loss": 1.0221,
+      "step": 17059
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.771641591142462e-05,
+      "loss": 1.0325,
+      "step": 17060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.770748694138131e-05,
+      "loss": 1.0141,
+      "step": 17061
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.769855797133801e-05,
+      "loss": 1.0565,
+      "step": 17062
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7689629001294704e-05,
+      "loss": 1.1652,
+      "step": 17063
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7680700031251395e-05,
+      "loss": 0.885,
+      "step": 17064
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.767177106120809e-05,
+      "loss": 1.0904,
+      "step": 17065
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.766284209116479e-05,
+      "loss": 1.0507,
+      "step": 17066
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.765391312112148e-05,
+      "loss": 1.0491,
+      "step": 17067
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.764498415107818e-05,
+      "loss": 0.9884,
+      "step": 17068
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7636055181034874e-05,
+      "loss": 1.0159,
+      "step": 17069
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7627126210991565e-05,
+      "loss": 0.9502,
+      "step": 17070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.761819724094826e-05,
+      "loss": 0.9875,
+      "step": 17071
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.760926827090495e-05,
+      "loss": 1.1639,
+      "step": 17072
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.760033930086164e-05,
+      "loss": 0.9464,
+      "step": 17073
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.759141033081834e-05,
+      "loss": 1.0241,
+      "step": 17074
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.758248136077504e-05,
+      "loss": 1.0559,
+      "step": 17075
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.757355239073173e-05,
+      "loss": 0.9844,
+      "step": 17076
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7564623420688425e-05,
+      "loss": 1.0726,
+      "step": 17077
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.755569445064512e-05,
+      "loss": 1.0189,
+      "step": 17078
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.754676548060181e-05,
+      "loss": 1.1486,
+      "step": 17079
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.753783651055851e-05,
+      "loss": 1.0692,
+      "step": 17080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.752890754051521e-05,
+      "loss": 0.9935,
+      "step": 17081
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.75199785704719e-05,
+      "loss": 1.1245,
+      "step": 17082
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7511049600428595e-05,
+      "loss": 1.1963,
+      "step": 17083
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7502120630385286e-05,
+      "loss": 1.0266,
+      "step": 17084
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7493191660341976e-05,
+      "loss": 0.962,
+      "step": 17085
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7484262690298673e-05,
+      "loss": 1.1223,
+      "step": 17086
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.747533372025537e-05,
+      "loss": 1.0391,
+      "step": 17087
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.746640475021207e-05,
+      "loss": 1.1836,
+      "step": 17088
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.745747578016876e-05,
+      "loss": 1.1376,
+      "step": 17089
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7448546810125456e-05,
+      "loss": 1.06,
+      "step": 17090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.743961784008215e-05,
+      "loss": 1.0001,
+      "step": 17091
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7430688870038843e-05,
+      "loss": 1.0264,
+      "step": 17092
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.742175989999554e-05,
+      "loss": 1.0531,
+      "step": 17093
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.741283092995223e-05,
+      "loss": 0.9922,
+      "step": 17094
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.740390195990893e-05,
+      "loss": 1.051,
+      "step": 17095
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.739497298986562e-05,
+      "loss": 1.0745,
+      "step": 17096
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7386044019822316e-05,
+      "loss": 1.0317,
+      "step": 17097
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.737711504977901e-05,
+      "loss": 0.9302,
+      "step": 17098
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7368186079735704e-05,
+      "loss": 0.95,
+      "step": 17099
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.73592571096924e-05,
+      "loss": 1.1331,
+      "step": 17100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.735032813964909e-05,
+      "loss": 1.1447,
+      "step": 17101
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.734139916960579e-05,
+      "loss": 1.1845,
+      "step": 17102
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7332470199562486e-05,
+      "loss": 1.2388,
+      "step": 17103
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.732354122951918e-05,
+      "loss": 1.0791,
+      "step": 17104
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7314612259475874e-05,
+      "loss": 1.045,
+      "step": 17105
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7305683289432564e-05,
+      "loss": 1.0638,
+      "step": 17106
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7296754319389255e-05,
+      "loss": 1.2045,
+      "step": 17107
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.728782534934595e-05,
+      "loss": 0.9811,
+      "step": 17108
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.727889637930265e-05,
+      "loss": 1.0901,
+      "step": 17109
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.726996740925935e-05,
+      "loss": 0.9124,
+      "step": 17110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.726103843921604e-05,
+      "loss": 0.9968,
+      "step": 17111
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7252109469172734e-05,
+      "loss": 0.8872,
+      "step": 17112
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.724318049912943e-05,
+      "loss": 1.0641,
+      "step": 17113
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.723425152908612e-05,
+      "loss": 0.9434,
+      "step": 17114
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.722532255904282e-05,
+      "loss": 1.0249,
+      "step": 17115
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.721639358899952e-05,
+      "loss": 1.1321,
+      "step": 17116
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.720746461895621e-05,
+      "loss": 0.9746,
+      "step": 17117
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.71985356489129e-05,
+      "loss": 0.9896,
+      "step": 17118
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7189606678869595e-05,
+      "loss": 1.0353,
+      "step": 17119
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7180677708826285e-05,
+      "loss": 0.8982,
+      "step": 17120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.717174873878298e-05,
+      "loss": 0.949,
+      "step": 17121
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.716281976873968e-05,
+      "loss": 0.8878,
+      "step": 17122
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.715389079869637e-05,
+      "loss": 0.9899,
+      "step": 17123
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.714496182865307e-05,
+      "loss": 1.0006,
+      "step": 17124
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7136032858609765e-05,
+      "loss": 0.8699,
+      "step": 17125
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7127103888566455e-05,
+      "loss": 1.1246,
+      "step": 17126
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.711817491852315e-05,
+      "loss": 1.1755,
+      "step": 17127
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.710924594847985e-05,
+      "loss": 1.0449,
+      "step": 17128
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.710031697843654e-05,
+      "loss": 1.0646,
+      "step": 17129
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.709138800839323e-05,
+      "loss": 1.1607,
+      "step": 17130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.708245903834993e-05,
+      "loss": 1.0874,
+      "step": 17131
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.707353006830662e-05,
+      "loss": 1.0505,
+      "step": 17132
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7064601098263316e-05,
+      "loss": 0.9644,
+      "step": 17133
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.705567212822001e-05,
+      "loss": 0.9989,
+      "step": 17134
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.704674315817671e-05,
+      "loss": 1.0008,
+      "step": 17135
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.70378141881334e-05,
+      "loss": 1.2359,
+      "step": 17136
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.70288852180901e-05,
+      "loss": 1.1492,
+      "step": 17137
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7019956248046795e-05,
+      "loss": 1.1376,
+      "step": 17138
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7011027278003486e-05,
+      "loss": 1.1067,
+      "step": 17139
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7002098307960176e-05,
+      "loss": 0.9369,
+      "step": 17140
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6993169337916874e-05,
+      "loss": 0.907,
+      "step": 17141
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6984240367873564e-05,
+      "loss": 1.0821,
+      "step": 17142
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.697531139783026e-05,
+      "loss": 1.0272,
+      "step": 17143
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.696638242778696e-05,
+      "loss": 1.0481,
+      "step": 17144
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.695745345774365e-05,
+      "loss": 0.8993,
+      "step": 17145
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6948524487700346e-05,
+      "loss": 1.0055,
+      "step": 17146
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6939595517657044e-05,
+      "loss": 1.0428,
+      "step": 17147
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6930666547613734e-05,
+      "loss": 0.9893,
+      "step": 17148
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.692173757757043e-05,
+      "loss": 1.0398,
+      "step": 17149
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.691280860752713e-05,
+      "loss": 1.0822,
+      "step": 17150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.690387963748382e-05,
+      "loss": 1.0636,
+      "step": 17151
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.689495066744051e-05,
+      "loss": 1.0121,
+      "step": 17152
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.688602169739721e-05,
+      "loss": 1.0195,
+      "step": 17153
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.68770927273539e-05,
+      "loss": 1.1339,
+      "step": 17154
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6868163757310594e-05,
+      "loss": 1.1918,
+      "step": 17155
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.685923478726729e-05,
+      "loss": 1.159,
+      "step": 17156
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.685030581722398e-05,
+      "loss": 1.0067,
+      "step": 17157
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.684137684718068e-05,
+      "loss": 0.9876,
+      "step": 17158
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.683244787713738e-05,
+      "loss": 0.9611,
+      "step": 17159
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6823518907094074e-05,
+      "loss": 0.9518,
+      "step": 17160
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6814589937050764e-05,
+      "loss": 0.9715,
+      "step": 17161
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.680566096700746e-05,
+      "loss": 0.9892,
+      "step": 17162
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.679673199696415e-05,
+      "loss": 1.027,
+      "step": 17163
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.678780302692084e-05,
+      "loss": 1.0178,
+      "step": 17164
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.677887405687754e-05,
+      "loss": 0.8857,
+      "step": 17165
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.676994508683424e-05,
+      "loss": 1.1016,
+      "step": 17166
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.676101611679093e-05,
+      "loss": 0.9945,
+      "step": 17167
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6752087146747625e-05,
+      "loss": 0.9646,
+      "step": 17168
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.674315817670432e-05,
+      "loss": 0.9709,
+      "step": 17169
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.673422920666101e-05,
+      "loss": 1.0853,
+      "step": 17170
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.672530023661771e-05,
+      "loss": 1.0866,
+      "step": 17171
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.671637126657441e-05,
+      "loss": 1.0955,
+      "step": 17172
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.67074422965311e-05,
+      "loss": 1.0355,
+      "step": 17173
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6698513326487795e-05,
+      "loss": 1.1132,
+      "step": 17174
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6689584356444485e-05,
+      "loss": 1.0967,
+      "step": 17175
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6680655386401176e-05,
+      "loss": 1.0447,
+      "step": 17176
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.667172641635787e-05,
+      "loss": 0.9813,
+      "step": 17177
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.666279744631457e-05,
+      "loss": 0.9988,
+      "step": 17178
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.665386847627126e-05,
+      "loss": 1.0635,
+      "step": 17179
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.664493950622796e-05,
+      "loss": 0.9512,
+      "step": 17180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6636010536184655e-05,
+      "loss": 1.0032,
+      "step": 17181
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6627081566141346e-05,
+      "loss": 1.1731,
+      "step": 17182
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.661815259609804e-05,
+      "loss": 1.0145,
+      "step": 17183
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.660922362605474e-05,
+      "loss": 1.0134,
+      "step": 17184
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.660029465601143e-05,
+      "loss": 1.1711,
+      "step": 17185
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.659136568596812e-05,
+      "loss": 0.9324,
+      "step": 17186
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.658243671592482e-05,
+      "loss": 0.9817,
+      "step": 17187
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6573507745881516e-05,
+      "loss": 1.0231,
+      "step": 17188
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6564578775838206e-05,
+      "loss": 0.9329,
+      "step": 17189
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6555649805794904e-05,
+      "loss": 0.9826,
+      "step": 17190
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.65467208357516e-05,
+      "loss": 1.0765,
+      "step": 17191
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.653779186570829e-05,
+      "loss": 1.0079,
+      "step": 17192
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.652886289566499e-05,
+      "loss": 1.0604,
+      "step": 17193
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6519933925621686e-05,
+      "loss": 0.9776,
+      "step": 17194
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6511004955578376e-05,
+      "loss": 1.0308,
+      "step": 17195
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6502075985535074e-05,
+      "loss": 1.0327,
+      "step": 17196
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6493147015491764e-05,
+      "loss": 1.0499,
+      "step": 17197
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6484218045448455e-05,
+      "loss": 0.9261,
+      "step": 17198
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.647528907540515e-05,
+      "loss": 1.0139,
+      "step": 17199
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.646636010536185e-05,
+      "loss": 1.1017,
+      "step": 17200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.645743113531854e-05,
+      "loss": 1.0758,
+      "step": 17201
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.644850216527524e-05,
+      "loss": 1.011,
+      "step": 17202
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6439573195231934e-05,
+      "loss": 1.0376,
+      "step": 17203
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6430644225188625e-05,
+      "loss": 1.001,
+      "step": 17204
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.642171525514532e-05,
+      "loss": 0.9091,
+      "step": 17205
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.641278628510202e-05,
+      "loss": 0.9912,
+      "step": 17206
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.640385731505871e-05,
+      "loss": 0.9395,
+      "step": 17207
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.639492834501541e-05,
+      "loss": 0.9754,
+      "step": 17208
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.63859993749721e-05,
+      "loss": 0.9642,
+      "step": 17209
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6377070404928794e-05,
+      "loss": 0.9917,
+      "step": 17210
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6368141434885485e-05,
+      "loss": 1.0045,
+      "step": 17211
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.635921246484218e-05,
+      "loss": 1.0222,
+      "step": 17212
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.635028349479888e-05,
+      "loss": 0.8957,
+      "step": 17213
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.634135452475557e-05,
+      "loss": 1.0538,
+      "step": 17214
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.633242555471227e-05,
+      "loss": 0.9601,
+      "step": 17215
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6323496584668964e-05,
+      "loss": 0.9145,
+      "step": 17216
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6314567614625655e-05,
+      "loss": 0.9904,
+      "step": 17217
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.630563864458235e-05,
+      "loss": 1.0252,
+      "step": 17218
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.629670967453904e-05,
+      "loss": 1.0874,
+      "step": 17219
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.628778070449574e-05,
+      "loss": 0.9093,
+      "step": 17220
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.627885173445243e-05,
+      "loss": 0.9588,
+      "step": 17221
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.626992276440913e-05,
+      "loss": 0.9976,
+      "step": 17222
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.626099379436582e-05,
+      "loss": 0.9796,
+      "step": 17223
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6252064824322515e-05,
+      "loss": 1.0953,
+      "step": 17224
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.624313585427921e-05,
+      "loss": 0.9834,
+      "step": 17225
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.62342068842359e-05,
+      "loss": 1.0501,
+      "step": 17226
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.62252779141926e-05,
+      "loss": 1.107,
+      "step": 17227
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.62163489441493e-05,
+      "loss": 0.9628,
+      "step": 17228
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.620741997410599e-05,
+      "loss": 1.1184,
+      "step": 17229
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6198491004062685e-05,
+      "loss": 0.907,
+      "step": 17230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6189562034019376e-05,
+      "loss": 1.0475,
+      "step": 17231
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.618063306397607e-05,
+      "loss": 1.0224,
+      "step": 17232
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6171704093932764e-05,
+      "loss": 0.9747,
+      "step": 17233
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.616277512388946e-05,
+      "loss": 0.9224,
+      "step": 17234
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 1.102,
+      "step": 17235
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.614491718380285e-05,
+      "loss": 1.1108,
+      "step": 17236
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6135988213759546e-05,
+      "loss": 1.0642,
+      "step": 17237
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.612705924371624e-05,
+      "loss": 0.9357,
+      "step": 17238
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6118130273672934e-05,
+      "loss": 1.2452,
+      "step": 17239
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.610920130362963e-05,
+      "loss": 0.9583,
+      "step": 17240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.610027233358633e-05,
+      "loss": 1.1549,
+      "step": 17241
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.609134336354302e-05,
+      "loss": 0.9978,
+      "step": 17242
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.608241439349971e-05,
+      "loss": 0.9644,
+      "step": 17243
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6073485423456406e-05,
+      "loss": 1.1475,
+      "step": 17244
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.60645564534131e-05,
+      "loss": 0.9496,
+      "step": 17245
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6055627483369794e-05,
+      "loss": 0.983,
+      "step": 17246
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.604669851332649e-05,
+      "loss": 1.0362,
+      "step": 17247
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.603776954328318e-05,
+      "loss": 1.0356,
+      "step": 17248
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.602884057323988e-05,
+      "loss": 1.0037,
+      "step": 17249
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6019911603196576e-05,
+      "loss": 1.1005,
+      "step": 17250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.601098263315327e-05,
+      "loss": 1.1004,
+      "step": 17251
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6002053663109964e-05,
+      "loss": 1.0202,
+      "step": 17252
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.599312469306666e-05,
+      "loss": 1.0848,
+      "step": 17253
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.598419572302335e-05,
+      "loss": 0.9694,
+      "step": 17254
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.597526675298004e-05,
+      "loss": 1.1301,
+      "step": 17255
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.596633778293674e-05,
+      "loss": 0.9706,
+      "step": 17256
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.595740881289344e-05,
+      "loss": 0.9317,
+      "step": 17257
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.594847984285013e-05,
+      "loss": 0.9901,
+      "step": 17258
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5939550872806825e-05,
+      "loss": 0.9965,
+      "step": 17259
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.593062190276352e-05,
+      "loss": 0.9073,
+      "step": 17260
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.592169293272021e-05,
+      "loss": 0.9613,
+      "step": 17261
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.591276396267691e-05,
+      "loss": 1.0522,
+      "step": 17262
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.590383499263361e-05,
+      "loss": 1.3294,
+      "step": 17263
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.58949060225903e-05,
+      "loss": 1.154,
+      "step": 17264
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5885977052546995e-05,
+      "loss": 0.9695,
+      "step": 17265
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5877048082503685e-05,
+      "loss": 1.035,
+      "step": 17266
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5868119112460375e-05,
+      "loss": 0.9098,
+      "step": 17267
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.585919014241707e-05,
+      "loss": 1.1446,
+      "step": 17268
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.585026117237377e-05,
+      "loss": 0.9609,
+      "step": 17269
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.584133220233046e-05,
+      "loss": 0.9959,
+      "step": 17270
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.583240323228716e-05,
+      "loss": 0.9854,
+      "step": 17271
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5823474262243855e-05,
+      "loss": 1.004,
+      "step": 17272
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5814545292200545e-05,
+      "loss": 1.1144,
+      "step": 17273
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.580561632215724e-05,
+      "loss": 0.9819,
+      "step": 17274
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.579668735211394e-05,
+      "loss": 1.0456,
+      "step": 17275
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.578775838207063e-05,
+      "loss": 0.9313,
+      "step": 17276
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.577882941202732e-05,
+      "loss": 1.0905,
+      "step": 17277
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.576990044198402e-05,
+      "loss": 1.0555,
+      "step": 17278
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.576097147194071e-05,
+      "loss": 0.9899,
+      "step": 17279
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5752042501897406e-05,
+      "loss": 1.0307,
+      "step": 17280
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.57431135318541e-05,
+      "loss": 1.0028,
+      "step": 17281
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.57341845618108e-05,
+      "loss": 1.0001,
+      "step": 17282
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.572525559176749e-05,
+      "loss": 0.9887,
+      "step": 17283
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.571632662172419e-05,
+      "loss": 1.002,
+      "step": 17284
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5707397651680885e-05,
+      "loss": 0.9723,
+      "step": 17285
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5698468681637576e-05,
+      "loss": 1.1086,
+      "step": 17286
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.568953971159427e-05,
+      "loss": 1.0557,
+      "step": 17287
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5680610741550964e-05,
+      "loss": 1.2334,
+      "step": 17288
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.5671681771507654e-05,
+      "loss": 0.9629,
+      "step": 17289
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.566275280146435e-05,
+      "loss": 1.0765,
+      "step": 17290
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.565382383142105e-05,
+      "loss": 0.9865,
+      "step": 17291
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.564489486137774e-05,
+      "loss": 1.1539,
+      "step": 17292
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5635965891334436e-05,
+      "loss": 1.0219,
+      "step": 17293
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5627036921291134e-05,
+      "loss": 1.0186,
+      "step": 17294
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5618107951247824e-05,
+      "loss": 1.0892,
+      "step": 17295
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.560917898120452e-05,
+      "loss": 0.8802,
+      "step": 17296
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.560025001116122e-05,
+      "loss": 1.0515,
+      "step": 17297
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.559132104111791e-05,
+      "loss": 0.9623,
+      "step": 17298
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5582392071074606e-05,
+      "loss": 1.1485,
+      "step": 17299
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.55734631010313e-05,
+      "loss": 1.0158,
+      "step": 17300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.556453413098799e-05,
+      "loss": 1.1344,
+      "step": 17301
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5555605160944685e-05,
+      "loss": 0.9355,
+      "step": 17302
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.554667619090138e-05,
+      "loss": 0.9361,
+      "step": 17303
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.553774722085807e-05,
+      "loss": 0.9929,
+      "step": 17304
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.552881825081477e-05,
+      "loss": 1.0396,
+      "step": 17305
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.551988928077147e-05,
+      "loss": 0.955,
+      "step": 17306
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5510960310728164e-05,
+      "loss": 0.9616,
+      "step": 17307
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5502031340684855e-05,
+      "loss": 0.9702,
+      "step": 17308
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.549310237064155e-05,
+      "loss": 0.9924,
+      "step": 17309
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.548417340059824e-05,
+      "loss": 1.0314,
+      "step": 17310
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.547524443055494e-05,
+      "loss": 1.06,
+      "step": 17311
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.546631546051163e-05,
+      "loss": 1.075,
+      "step": 17312
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.545738649046833e-05,
+      "loss": 0.9735,
+      "step": 17313
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.544845752042502e-05,
+      "loss": 0.9658,
+      "step": 17314
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5439528550381715e-05,
+      "loss": 0.9612,
+      "step": 17315
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.543059958033841e-05,
+      "loss": 1.0174,
+      "step": 17316
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.54216706102951e-05,
+      "loss": 0.9948,
+      "step": 17317
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.54127416402518e-05,
+      "loss": 0.9512,
+      "step": 17318
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.54038126702085e-05,
+      "loss": 1.0234,
+      "step": 17319
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.539488370016519e-05,
+      "loss": 1.0931,
+      "step": 17320
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5385954730121885e-05,
+      "loss": 1.0843,
+      "step": 17321
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5377025760078576e-05,
+      "loss": 1.0959,
+      "step": 17322
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5368096790035266e-05,
+      "loss": 0.9746,
+      "step": 17323
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.535916781999196e-05,
+      "loss": 0.9932,
+      "step": 17324
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.535023884994866e-05,
+      "loss": 1.0209,
+      "step": 17325
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.534130987990535e-05,
+      "loss": 1.226,
+      "step": 17326
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.533238090986205e-05,
+      "loss": 1.0593,
+      "step": 17327
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5323451939818746e-05,
+      "loss": 1.0335,
+      "step": 17328
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5314522969775436e-05,
+      "loss": 1.0538,
+      "step": 17329
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.530559399973213e-05,
+      "loss": 1.0559,
+      "step": 17330
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.529666502968883e-05,
+      "loss": 0.9833,
+      "step": 17331
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.528773605964553e-05,
+      "loss": 0.9656,
+      "step": 17332
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.527880708960222e-05,
+      "loss": 1.0109,
+      "step": 17333
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.526987811955891e-05,
+      "loss": 0.97,
+      "step": 17334
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5260949149515606e-05,
+      "loss": 1.1091,
+      "step": 17335
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5252020179472296e-05,
+      "loss": 0.9625,
+      "step": 17336
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5243091209428994e-05,
+      "loss": 1.017,
+      "step": 17337
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.523416223938569e-05,
+      "loss": 1.1342,
+      "step": 17338
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.522523326934238e-05,
+      "loss": 0.9705,
+      "step": 17339
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.521630429929908e-05,
+      "loss": 0.9418,
+      "step": 17340
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5207375329255776e-05,
+      "loss": 0.988,
+      "step": 17341
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5198446359212466e-05,
+      "loss": 1.0862,
+      "step": 17342
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5189517389169164e-05,
+      "loss": 0.9098,
+      "step": 17343
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.518058841912586e-05,
+      "loss": 0.9933,
+      "step": 17344
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.517165944908255e-05,
+      "loss": 0.9898,
+      "step": 17345
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.516273047903924e-05,
+      "loss": 0.9683,
+      "step": 17346
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.515380150899594e-05,
+      "loss": 1.0058,
+      "step": 17347
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.514487253895263e-05,
+      "loss": 1.0004,
+      "step": 17348
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.513594356890933e-05,
+      "loss": 1.0335,
+      "step": 17349
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5127014598866024e-05,
+      "loss": 0.8785,
+      "step": 17350
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5118085628822715e-05,
+      "loss": 1.0047,
+      "step": 17351
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.510915665877941e-05,
+      "loss": 1.063,
+      "step": 17352
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.510022768873611e-05,
+      "loss": 1.0129,
+      "step": 17353
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.50912987186928e-05,
+      "loss": 1.0383,
+      "step": 17354
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.50823697486495e-05,
+      "loss": 1.017,
+      "step": 17355
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.507344077860619e-05,
+      "loss": 1.0251,
+      "step": 17356
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5064511808562885e-05,
+      "loss": 1.023,
+      "step": 17357
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5055582838519575e-05,
+      "loss": 1.0255,
+      "step": 17358
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.504665386847627e-05,
+      "loss": 1.0308,
+      "step": 17359
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.503772489843297e-05,
+      "loss": 1.0074,
+      "step": 17360
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.502879592838966e-05,
+      "loss": 0.9946,
+      "step": 17361
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.501986695834636e-05,
+      "loss": 1.0139,
+      "step": 17362
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5010937988303055e-05,
+      "loss": 1.0207,
+      "step": 17363
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5002009018259745e-05,
+      "loss": 1.2164,
+      "step": 17364
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.499308004821644e-05,
+      "loss": 0.9969,
+      "step": 17365
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.498415107817314e-05,
+      "loss": 1.0315,
+      "step": 17366
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.497522210812983e-05,
+      "loss": 0.9414,
+      "step": 17367
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.496629313808652e-05,
+      "loss": 0.9879,
+      "step": 17368
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.495736416804322e-05,
+      "loss": 1.4816,
+      "step": 17369
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.494843519799991e-05,
+      "loss": 1.1853,
+      "step": 17370
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4939506227956606e-05,
+      "loss": 1.0269,
+      "step": 17371
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.49305772579133e-05,
+      "loss": 0.9922,
+      "step": 17372
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.492164828786999e-05,
+      "loss": 0.9136,
+      "step": 17373
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.491271931782669e-05,
+      "loss": 0.9958,
+      "step": 17374
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.490379034778339e-05,
+      "loss": 1.1531,
+      "step": 17375
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.489486137774008e-05,
+      "loss": 1.0463,
+      "step": 17376
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4885932407696776e-05,
+      "loss": 1.0789,
+      "step": 17377
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.487700343765347e-05,
+      "loss": 0.9722,
+      "step": 17378
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.486807446761016e-05,
+      "loss": 1.1522,
+      "step": 17379
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4859145497566854e-05,
+      "loss": 1.0336,
+      "step": 17380
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.485021652752355e-05,
+      "loss": 1.1723,
+      "step": 17381
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.484128755748025e-05,
+      "loss": 1.0067,
+      "step": 17382
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.483235858743694e-05,
+      "loss": 0.9675,
+      "step": 17383
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4823429617393636e-05,
+      "loss": 0.9584,
+      "step": 17384
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.481450064735033e-05,
+      "loss": 1.0285,
+      "step": 17385
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4805571677307024e-05,
+      "loss": 1.1999,
+      "step": 17386
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.479664270726372e-05,
+      "loss": 1.0837,
+      "step": 17387
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.478771373722042e-05,
+      "loss": 1.0594,
+      "step": 17388
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.477878476717711e-05,
+      "loss": 1.1158,
+      "step": 17389
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4769855797133806e-05,
+      "loss": 1.0865,
+      "step": 17390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4760926827090496e-05,
+      "loss": 0.8814,
+      "step": 17391
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.475199785704719e-05,
+      "loss": 1.1239,
+      "step": 17392
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4743068887003884e-05,
+      "loss": 1.0269,
+      "step": 17393
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.473413991696058e-05,
+      "loss": 1.0942,
+      "step": 17394
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.472521094691727e-05,
+      "loss": 0.9505,
+      "step": 17395
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.471628197687397e-05,
+      "loss": 1.0163,
+      "step": 17396
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4707353006830666e-05,
+      "loss": 0.9986,
+      "step": 17397
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.469842403678736e-05,
+      "loss": 1.135,
+      "step": 17398
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4689495066744054e-05,
+      "loss": 1.0547,
+      "step": 17399
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.468056609670075e-05,
+      "loss": 1.0446,
+      "step": 17400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.467163712665744e-05,
+      "loss": 0.9626,
+      "step": 17401
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.466270815661414e-05,
+      "loss": 1.0147,
+      "step": 17402
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.465377918657083e-05,
+      "loss": 1.0469,
+      "step": 17403
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.464485021652753e-05,
+      "loss": 1.0177,
+      "step": 17404
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.463592124648422e-05,
+      "loss": 0.9027,
+      "step": 17405
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4626992276440915e-05,
+      "loss": 1.1609,
+      "step": 17406
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.461806330639761e-05,
+      "loss": 1.132,
+      "step": 17407
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.46091343363543e-05,
+      "loss": 1.0102,
+      "step": 17408
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4600205366311e-05,
+      "loss": 1.0021,
+      "step": 17409
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.45912763962677e-05,
+      "loss": 1.086,
+      "step": 17410
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.458234742622439e-05,
+      "loss": 1.0551,
+      "step": 17411
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4573418456181085e-05,
+      "loss": 0.9838,
+      "step": 17412
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4564489486137775e-05,
+      "loss": 0.9296,
+      "step": 17413
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4555560516094466e-05,
+      "loss": 0.9534,
+      "step": 17414
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.454663154605116e-05,
+      "loss": 1.0229,
+      "step": 17415
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.453770257600786e-05,
+      "loss": 1.0495,
+      "step": 17416
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.452877360596455e-05,
+      "loss": 1.076,
+      "step": 17417
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.451984463592125e-05,
+      "loss": 1.0234,
+      "step": 17418
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4510915665877945e-05,
+      "loss": 1.0244,
+      "step": 17419
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4501986695834636e-05,
+      "loss": 0.9857,
+      "step": 17420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.449305772579133e-05,
+      "loss": 1.1271,
+      "step": 17421
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.448412875574803e-05,
+      "loss": 1.0126,
+      "step": 17422
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.447519978570472e-05,
+      "loss": 0.9584,
+      "step": 17423
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.446627081566142e-05,
+      "loss": 1.0798,
+      "step": 17424
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.445734184561811e-05,
+      "loss": 1.0644,
+      "step": 17425
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.44484128755748e-05,
+      "loss": 1.1212,
+      "step": 17426
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4439483905531496e-05,
+      "loss": 1.0815,
+      "step": 17427
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.443055493548819e-05,
+      "loss": 1.0379,
+      "step": 17428
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.442162596544489e-05,
+      "loss": 1.0333,
+      "step": 17429
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.441269699540158e-05,
+      "loss": 1.0204,
+      "step": 17430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.440376802535828e-05,
+      "loss": 0.9637,
+      "step": 17431
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4394839055314976e-05,
+      "loss": 1.0363,
+      "step": 17432
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.4385910085271666e-05,
+      "loss": 1.0271,
+      "step": 17433
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.437698111522836e-05,
+      "loss": 1.0327,
+      "step": 17434
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.436805214518506e-05,
+      "loss": 0.9838,
+      "step": 17435
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.435912317514175e-05,
+      "loss": 0.9118,
+      "step": 17436
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.435019420509844e-05,
+      "loss": 1.138,
+      "step": 17437
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.434126523505514e-05,
+      "loss": 0.983,
+      "step": 17438
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.433233626501183e-05,
+      "loss": 1.0149,
+      "step": 17439
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4323407294968527e-05,
+      "loss": 1.0369,
+      "step": 17440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4314478324925224e-05,
+      "loss": 0.9999,
+      "step": 17441
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4305549354881914e-05,
+      "loss": 1.1142,
+      "step": 17442
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.429662038483861e-05,
+      "loss": 0.9134,
+      "step": 17443
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.428769141479531e-05,
+      "loss": 0.9927,
+      "step": 17444
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4278762444752e-05,
+      "loss": 1.0747,
+      "step": 17445
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4269833474708697e-05,
+      "loss": 1.029,
+      "step": 17446
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.426090450466539e-05,
+      "loss": 0.9695,
+      "step": 17447
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4251975534622084e-05,
+      "loss": 1.179,
+      "step": 17448
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4243046564578775e-05,
+      "loss": 1.0565,
+      "step": 17449
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.423411759453547e-05,
+      "loss": 1.0367,
+      "step": 17450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.422518862449216e-05,
+      "loss": 1.0193,
+      "step": 17451
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.421625965444886e-05,
+      "loss": 0.95,
+      "step": 17452
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.420733068440556e-05,
+      "loss": 0.9695,
+      "step": 17453
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4198401714362254e-05,
+      "loss": 0.9176,
+      "step": 17454
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4189472744318945e-05,
+      "loss": 1.0471,
+      "step": 17455
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.418054377427564e-05,
+      "loss": 0.9904,
+      "step": 17456
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.417161480423234e-05,
+      "loss": 0.9935,
+      "step": 17457
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.416268583418903e-05,
+      "loss": 1.0945,
+      "step": 17458
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.415375686414572e-05,
+      "loss": 0.9555,
+      "step": 17459
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.414482789410242e-05,
+      "loss": 1.1503,
+      "step": 17460
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.413589892405911e-05,
+      "loss": 1.1047,
+      "step": 17461
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4126969954015805e-05,
+      "loss": 1.0785,
+      "step": 17462
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.41180409839725e-05,
+      "loss": 1.0169,
+      "step": 17463
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.410911201392919e-05,
+      "loss": 1.0523,
+      "step": 17464
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.410018304388589e-05,
+      "loss": 1.1297,
+      "step": 17465
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.409125407384259e-05,
+      "loss": 0.9619,
+      "step": 17466
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.408232510379928e-05,
+      "loss": 0.9557,
+      "step": 17467
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4073396133755975e-05,
+      "loss": 1.1225,
+      "step": 17468
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.406446716371267e-05,
+      "loss": 0.8702,
+      "step": 17469
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.405553819366936e-05,
+      "loss": 0.9722,
+      "step": 17470
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4046609223626053e-05,
+      "loss": 1.0888,
+      "step": 17471
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.403768025358275e-05,
+      "loss": 1.0246,
+      "step": 17472
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.402875128353944e-05,
+      "loss": 1.0036,
+      "step": 17473
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.401982231349614e-05,
+      "loss": 1.0081,
+      "step": 17474
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4010893343452836e-05,
+      "loss": 1.0855,
+      "step": 17475
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4001964373409526e-05,
+      "loss": 1.0017,
+      "step": 17476
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3993035403366223e-05,
+      "loss": 1.0946,
+      "step": 17477
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.398410643332292e-05,
+      "loss": 1.0008,
+      "step": 17478
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.397517746327962e-05,
+      "loss": 1.1994,
+      "step": 17479
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.396624849323631e-05,
+      "loss": 1.052,
+      "step": 17480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3957319523193006e-05,
+      "loss": 1.0256,
+      "step": 17481
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3948390553149696e-05,
+      "loss": 1.1178,
+      "step": 17482
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3939461583106387e-05,
+      "loss": 1.142,
+      "step": 17483
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3930532613063084e-05,
+      "loss": 1.0235,
+      "step": 17484
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.392160364301978e-05,
+      "loss": 1.0001,
+      "step": 17485
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.391267467297647e-05,
+      "loss": 1.0124,
+      "step": 17486
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.390374570293317e-05,
+      "loss": 1.0179,
+      "step": 17487
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3894816732889866e-05,
+      "loss": 1.0569,
+      "step": 17488
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3885887762846557e-05,
+      "loss": 1.1663,
+      "step": 17489
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3876958792803254e-05,
+      "loss": 1.1146,
+      "step": 17490
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.386802982275995e-05,
+      "loss": 1.0109,
+      "step": 17491
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.385910085271664e-05,
+      "loss": 0.9159,
+      "step": 17492
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.385017188267333e-05,
+      "loss": 1.0282,
+      "step": 17493
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.384124291263003e-05,
+      "loss": 0.8713,
+      "step": 17494
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.383231394258672e-05,
+      "loss": 0.9802,
+      "step": 17495
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.382338497254342e-05,
+      "loss": 1.0161,
+      "step": 17496
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3814456002500114e-05,
+      "loss": 0.9446,
+      "step": 17497
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3805527032456805e-05,
+      "loss": 1.096,
+      "step": 17498
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.37965980624135e-05,
+      "loss": 1.4278,
+      "step": 17499
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.37876690923702e-05,
+      "loss": 1.067,
+      "step": 17500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.377874012232689e-05,
+      "loss": 0.9688,
+      "step": 17501
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.376981115228359e-05,
+      "loss": 0.9831,
+      "step": 17502
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3760882182240284e-05,
+      "loss": 1.1266,
+      "step": 17503
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3751953212196975e-05,
+      "loss": 1.237,
+      "step": 17504
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3743024242153665e-05,
+      "loss": 1.1172,
+      "step": 17505
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.373409527211036e-05,
+      "loss": 0.9248,
+      "step": 17506
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.372516630206706e-05,
+      "loss": 1.0362,
+      "step": 17507
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.371623733202375e-05,
+      "loss": 1.1514,
+      "step": 17508
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.370730836198045e-05,
+      "loss": 1.0322,
+      "step": 17509
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3698379391937145e-05,
+      "loss": 1.0985,
+      "step": 17510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3689450421893835e-05,
+      "loss": 0.9571,
+      "step": 17511
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.368052145185053e-05,
+      "loss": 1.0368,
+      "step": 17512
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.367159248180723e-05,
+      "loss": 1.0047,
+      "step": 17513
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.366266351176392e-05,
+      "loss": 1.0701,
+      "step": 17514
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.365373454172062e-05,
+      "loss": 0.9048,
+      "step": 17515
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.364480557167731e-05,
+      "loss": 0.9823,
+      "step": 17516
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3635876601634e-05,
+      "loss": 1.1314,
+      "step": 17517
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3626947631590696e-05,
+      "loss": 1.0369,
+      "step": 17518
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.361801866154739e-05,
+      "loss": 0.9843,
+      "step": 17519
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3609089691504083e-05,
+      "loss": 0.8958,
+      "step": 17520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.360016072146078e-05,
+      "loss": 1.0977,
+      "step": 17521
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.359123175141748e-05,
+      "loss": 1.0008,
+      "step": 17522
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.358230278137417e-05,
+      "loss": 1.0336,
+      "step": 17523
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3573373811330866e-05,
+      "loss": 0.9558,
+      "step": 17524
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.356444484128756e-05,
+      "loss": 0.9336,
+      "step": 17525
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3555515871244253e-05,
+      "loss": 1.1241,
+      "step": 17526
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.354658690120095e-05,
+      "loss": 1.119,
+      "step": 17527
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.353765793115764e-05,
+      "loss": 0.9276,
+      "step": 17528
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.352872896111434e-05,
+      "loss": 1.0292,
+      "step": 17529
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.351979999107103e-05,
+      "loss": 0.9486,
+      "step": 17530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3510871021027726e-05,
+      "loss": 1.1113,
+      "step": 17531
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3501942050984423e-05,
+      "loss": 0.923,
+      "step": 17532
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3493013080941114e-05,
+      "loss": 0.9571,
+      "step": 17533
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.348408411089781e-05,
+      "loss": 1.1195,
+      "step": 17534
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.347515514085451e-05,
+      "loss": 1.052,
+      "step": 17535
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.34662261708112e-05,
+      "loss": 1.0566,
+      "step": 17536
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3457297200767896e-05,
+      "loss": 1.0417,
+      "step": 17537
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.344836823072459e-05,
+      "loss": 1.1027,
+      "step": 17538
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3439439260681284e-05,
+      "loss": 0.9475,
+      "step": 17539
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3430510290637974e-05,
+      "loss": 0.9768,
+      "step": 17540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.342158132059467e-05,
+      "loss": 0.9328,
+      "step": 17541
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.341265235055136e-05,
+      "loss": 1.0511,
+      "step": 17542
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.340372338050806e-05,
+      "loss": 1.1558,
+      "step": 17543
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.339479441046476e-05,
+      "loss": 1.0625,
+      "step": 17544
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.338586544042145e-05,
+      "loss": 0.9713,
+      "step": 17545
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3376936470378144e-05,
+      "loss": 0.9217,
+      "step": 17546
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.336800750033484e-05,
+      "loss": 0.9497,
+      "step": 17547
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.335907853029153e-05,
+      "loss": 1.0629,
+      "step": 17548
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.335014956024823e-05,
+      "loss": 1.1683,
+      "step": 17549
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.334122059020492e-05,
+      "loss": 1.052,
+      "step": 17550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.333229162016162e-05,
+      "loss": 0.9934,
+      "step": 17551
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.332336265011831e-05,
+      "loss": 0.9576,
+      "step": 17552
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3314433680075005e-05,
+      "loss": 0.9643,
+      "step": 17553
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.33055047100317e-05,
+      "loss": 0.9302,
+      "step": 17554
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.329657573998839e-05,
+      "loss": 0.8962,
+      "step": 17555
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.328764676994509e-05,
+      "loss": 0.9139,
+      "step": 17556
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.327871779990179e-05,
+      "loss": 0.9917,
+      "step": 17557
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.326978882985848e-05,
+      "loss": 1.055,
+      "step": 17558
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3260859859815175e-05,
+      "loss": 1.1117,
+      "step": 17559
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.325193088977187e-05,
+      "loss": 1.0155,
+      "step": 17560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.324300191972856e-05,
+      "loss": 0.9678,
+      "step": 17561
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.323407294968525e-05,
+      "loss": 1.0297,
+      "step": 17562
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.322514397964195e-05,
+      "loss": 1.1098,
+      "step": 17563
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.321621500959864e-05,
+      "loss": 1.0053,
+      "step": 17564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.320728603955534e-05,
+      "loss": 0.9654,
+      "step": 17565
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3198357069512035e-05,
+      "loss": 0.9629,
+      "step": 17566
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3189428099468726e-05,
+      "loss": 1.0615,
+      "step": 17567
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.318049912942542e-05,
+      "loss": 1.0032,
+      "step": 17568
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.317157015938212e-05,
+      "loss": 0.9493,
+      "step": 17569
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.316264118933881e-05,
+      "loss": 1.0588,
+      "step": 17570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.315371221929551e-05,
+      "loss": 0.9615,
+      "step": 17571
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3144783249252205e-05,
+      "loss": 1.1916,
+      "step": 17572
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3135854279208896e-05,
+      "loss": 1.0688,
+      "step": 17573
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3126925309165586e-05,
+      "loss": 1.0778,
+      "step": 17574
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3117996339122283e-05,
+      "loss": 0.9906,
+      "step": 17575
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.310906736907898e-05,
+      "loss": 0.9221,
+      "step": 17576
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.310013839903567e-05,
+      "loss": 0.9706,
+      "step": 17577
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.309120942899237e-05,
+      "loss": 0.9529,
+      "step": 17578
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3082280458949066e-05,
+      "loss": 0.8693,
+      "step": 17579
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3073351488905756e-05,
+      "loss": 1.0498,
+      "step": 17580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3064422518862453e-05,
+      "loss": 1.0297,
+      "step": 17581
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.305549354881915e-05,
+      "loss": 0.9918,
+      "step": 17582
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.304656457877584e-05,
+      "loss": 1.0047,
+      "step": 17583
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.303763560873253e-05,
+      "loss": 1.0681,
+      "step": 17584
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.302870663868923e-05,
+      "loss": 0.9249,
+      "step": 17585
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.301977766864592e-05,
+      "loss": 0.9544,
+      "step": 17586
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.301084869860262e-05,
+      "loss": 1.1105,
+      "step": 17587
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3001919728559314e-05,
+      "loss": 1.1031,
+      "step": 17588
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2992990758516004e-05,
+      "loss": 1.0135,
+      "step": 17589
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.29840617884727e-05,
+      "loss": 0.9655,
+      "step": 17590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.29751328184294e-05,
+      "loss": 1.119,
+      "step": 17591
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.296620384838609e-05,
+      "loss": 1.08,
+      "step": 17592
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.295727487834279e-05,
+      "loss": 1.0987,
+      "step": 17593
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2948345908299484e-05,
+      "loss": 1.0117,
+      "step": 17594
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2939416938256174e-05,
+      "loss": 1.0319,
+      "step": 17595
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2930487968212865e-05,
+      "loss": 1.1387,
+      "step": 17596
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.292155899816956e-05,
+      "loss": 1.0271,
+      "step": 17597
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.291263002812625e-05,
+      "loss": 1.0122,
+      "step": 17598
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.290370105808295e-05,
+      "loss": 0.9616,
+      "step": 17599
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.289477208803965e-05,
+      "loss": 1.0326,
+      "step": 17600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2885843117996344e-05,
+      "loss": 1.2008,
+      "step": 17601
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2876914147953035e-05,
+      "loss": 1.0454,
+      "step": 17602
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.286798517790973e-05,
+      "loss": 0.884,
+      "step": 17603
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.285905620786643e-05,
+      "loss": 1.0904,
+      "step": 17604
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.285012723782312e-05,
+      "loss": 1.0745,
+      "step": 17605
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.284119826777982e-05,
+      "loss": 1.0502,
+      "step": 17606
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.283226929773651e-05,
+      "loss": 1.1702,
+      "step": 17607
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.28233403276932e-05,
+      "loss": 0.9814,
+      "step": 17608
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2814411357649895e-05,
+      "loss": 0.9747,
+      "step": 17609
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.280548238760659e-05,
+      "loss": 1.1439,
+      "step": 17610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.279655341756328e-05,
+      "loss": 0.9947,
+      "step": 17611
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.278762444751998e-05,
+      "loss": 0.994,
+      "step": 17612
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.277869547747668e-05,
+      "loss": 1.0728,
+      "step": 17613
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.276976650743337e-05,
+      "loss": 0.9627,
+      "step": 17614
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2760837537390065e-05,
+      "loss": 0.9809,
+      "step": 17615
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.275190856734676e-05,
+      "loss": 1.0617,
+      "step": 17616
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.274297959730345e-05,
+      "loss": 1.0534,
+      "step": 17617
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.273405062726015e-05,
+      "loss": 0.9664,
+      "step": 17618
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.272512165721684e-05,
+      "loss": 1.0507,
+      "step": 17619
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.271619268717353e-05,
+      "loss": 0.9315,
+      "step": 17620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.270726371713023e-05,
+      "loss": 1.1013,
+      "step": 17621
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2698334747086926e-05,
+      "loss": 1.0706,
+      "step": 17622
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2689405777043616e-05,
+      "loss": 1.0333,
+      "step": 17623
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2680476807000314e-05,
+      "loss": 1.0306,
+      "step": 17624
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.267154783695701e-05,
+      "loss": 1.0928,
+      "step": 17625
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.266261886691371e-05,
+      "loss": 1.164,
+      "step": 17626
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.26536898968704e-05,
+      "loss": 1.0245,
+      "step": 17627
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2644760926827096e-05,
+      "loss": 1.0159,
+      "step": 17628
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2635831956783786e-05,
+      "loss": 1.0433,
+      "step": 17629
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.262690298674048e-05,
+      "loss": 1.0064,
+      "step": 17630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2617974016697174e-05,
+      "loss": 1.1646,
+      "step": 17631
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.260904504665387e-05,
+      "loss": 0.947,
+      "step": 17632
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.260011607661056e-05,
+      "loss": 1.0672,
+      "step": 17633
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.259118710656726e-05,
+      "loss": 1.0205,
+      "step": 17634
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2582258136523956e-05,
+      "loss": 0.946,
+      "step": 17635
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.257332916648065e-05,
+      "loss": 0.9589,
+      "step": 17636
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2564400196437344e-05,
+      "loss": 1.0183,
+      "step": 17637
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.255547122639404e-05,
+      "loss": 1.155,
+      "step": 17638
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.254654225635073e-05,
+      "loss": 1.2606,
+      "step": 17639
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.253761328630743e-05,
+      "loss": 1.0304,
+      "step": 17640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.252868431626412e-05,
+      "loss": 1.0047,
+      "step": 17641
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.251975534622081e-05,
+      "loss": 0.9897,
+      "step": 17642
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.251082637617751e-05,
+      "loss": 1.0018,
+      "step": 17643
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2501897406134204e-05,
+      "loss": 0.9651,
+      "step": 17644
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2492968436090895e-05,
+      "loss": 1.0316,
+      "step": 17645
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.248403946604759e-05,
+      "loss": 1.072,
+      "step": 17646
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.247511049600429e-05,
+      "loss": 1.0695,
+      "step": 17647
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.246618152596099e-05,
+      "loss": 0.9363,
+      "step": 17648
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.245725255591768e-05,
+      "loss": 1.063,
+      "step": 17649
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2448323585874374e-05,
+      "loss": 1.0548,
+      "step": 17650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.243939461583107e-05,
+      "loss": 1.05,
+      "step": 17651
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.243046564578776e-05,
+      "loss": 0.9065,
+      "step": 17652
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.242153667574445e-05,
+      "loss": 1.1542,
+      "step": 17653
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.241260770570115e-05,
+      "loss": 0.9703,
+      "step": 17654
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.240367873565784e-05,
+      "loss": 0.9116,
+      "step": 17655
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.239474976561454e-05,
+      "loss": 0.904,
+      "step": 17656
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2385820795571235e-05,
+      "loss": 1.0818,
+      "step": 17657
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2376891825527925e-05,
+      "loss": 0.8885,
+      "step": 17658
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.236796285548462e-05,
+      "loss": 1.0245,
+      "step": 17659
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.235903388544132e-05,
+      "loss": 1.0272,
+      "step": 17660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.235010491539801e-05,
+      "loss": 1.1526,
+      "step": 17661
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.234117594535471e-05,
+      "loss": 1.0381,
+      "step": 17662
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.23322469753114e-05,
+      "loss": 1.0325,
+      "step": 17663
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2323318005268095e-05,
+      "loss": 0.995,
+      "step": 17664
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2314389035224786e-05,
+      "loss": 1.0525,
+      "step": 17665
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.230546006518148e-05,
+      "loss": 1.0215,
+      "step": 17666
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2296531095138174e-05,
+      "loss": 1.1198,
+      "step": 17667
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.228760212509487e-05,
+      "loss": 0.9396,
+      "step": 17668
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.227867315505157e-05,
+      "loss": 1.0609,
+      "step": 17669
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.226974418500826e-05,
+      "loss": 1.0183,
+      "step": 17670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2260815214964956e-05,
+      "loss": 1.3051,
+      "step": 17671
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.225188624492165e-05,
+      "loss": 0.8517,
+      "step": 17672
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.224295727487835e-05,
+      "loss": 1.0846,
+      "step": 17673
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.223402830483504e-05,
+      "loss": 0.9703,
+      "step": 17674
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.222509933479173e-05,
+      "loss": 1.0713,
+      "step": 17675
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.221617036474843e-05,
+      "loss": 0.9947,
+      "step": 17676
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.220724139470512e-05,
+      "loss": 0.9322,
+      "step": 17677
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2198312424661816e-05,
+      "loss": 1.0971,
+      "step": 17678
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2189383454618514e-05,
+      "loss": 1.0211,
+      "step": 17679
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2180454484575204e-05,
+      "loss": 1.0075,
+      "step": 17680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.21715255145319e-05,
+      "loss": 1.1486,
+      "step": 17681
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.21625965444886e-05,
+      "loss": 0.9109,
+      "step": 17682
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.215366757444529e-05,
+      "loss": 1.1193,
+      "step": 17683
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2144738604401986e-05,
+      "loss": 1.1671,
+      "step": 17684
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2135809634358684e-05,
+      "loss": 1.101,
+      "step": 17685
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2126880664315374e-05,
+      "loss": 0.9125,
+      "step": 17686
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2117951694272065e-05,
+      "loss": 1.0597,
+      "step": 17687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.210902272422876e-05,
+      "loss": 0.9625,
+      "step": 17688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.210009375418545e-05,
+      "loss": 1.0031,
+      "step": 17689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.209116478414215e-05,
+      "loss": 0.9307,
+      "step": 17690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.208223581409885e-05,
+      "loss": 1.1089,
+      "step": 17691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.207330684405554e-05,
+      "loss": 1.0271,
+      "step": 17692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2064377874012235e-05,
+      "loss": 1.0767,
+      "step": 17693
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.205544890396893e-05,
+      "loss": 1.0578,
+      "step": 17694
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.204651993392562e-05,
+      "loss": 0.9897,
+      "step": 17695
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.203759096388232e-05,
+      "loss": 1.0173,
+      "step": 17696
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.202866199383902e-05,
+      "loss": 0.9823,
+      "step": 17697
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.201973302379571e-05,
+      "loss": 1.1692,
+      "step": 17698
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.20108040537524e-05,
+      "loss": 1.046,
+      "step": 17699
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2001875083709095e-05,
+      "loss": 1.1146,
+      "step": 17700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.199294611366579e-05,
+      "loss": 1.0022,
+      "step": 17701
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.198401714362248e-05,
+      "loss": 1.1582,
+      "step": 17702
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.197508817357918e-05,
+      "loss": 1.1169,
+      "step": 17703
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.196615920353588e-05,
+      "loss": 0.903,
+      "step": 17704
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.195723023349257e-05,
+      "loss": 1.1132,
+      "step": 17705
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1948301263449265e-05,
+      "loss": 1.0191,
+      "step": 17706
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.193937229340596e-05,
+      "loss": 1.0485,
+      "step": 17707
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.193044332336265e-05,
+      "loss": 1.1991,
+      "step": 17708
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.192151435331935e-05,
+      "loss": 1.0672,
+      "step": 17709
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.191258538327604e-05,
+      "loss": 1.0497,
+      "step": 17710
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.190365641323273e-05,
+      "loss": 1.0078,
+      "step": 17711
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.189472744318943e-05,
+      "loss": 0.9465,
+      "step": 17712
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1885798473146125e-05,
+      "loss": 1.0195,
+      "step": 17713
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1876869503102816e-05,
+      "loss": 1.0162,
+      "step": 17714
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.186794053305951e-05,
+      "loss": 1.1693,
+      "step": 17715
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.185901156301621e-05,
+      "loss": 1.0201,
+      "step": 17716
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.18500825929729e-05,
+      "loss": 1.0994,
+      "step": 17717
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.18411536229296e-05,
+      "loss": 1.1451,
+      "step": 17718
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1832224652886295e-05,
+      "loss": 1.06,
+      "step": 17719
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1823295682842986e-05,
+      "loss": 1.1152,
+      "step": 17720
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1814366712799676e-05,
+      "loss": 1.2268,
+      "step": 17721
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1805437742756374e-05,
+      "loss": 1.0661,
+      "step": 17722
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.179650877271307e-05,
+      "loss": 0.9692,
+      "step": 17723
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.178757980266976e-05,
+      "loss": 0.9793,
+      "step": 17724
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.177865083262646e-05,
+      "loss": 1.1476,
+      "step": 17725
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1769721862583156e-05,
+      "loss": 1.102,
+      "step": 17726
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1760792892539846e-05,
+      "loss": 0.9673,
+      "step": 17727
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1751863922496544e-05,
+      "loss": 0.9916,
+      "step": 17728
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.174293495245324e-05,
+      "loss": 1.0988,
+      "step": 17729
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.173400598240993e-05,
+      "loss": 1.0178,
+      "step": 17730
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.172507701236663e-05,
+      "loss": 1.0616,
+      "step": 17731
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.171614804232332e-05,
+      "loss": 1.0101,
+      "step": 17732
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.170721907228001e-05,
+      "loss": 1.0508,
+      "step": 17733
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.169829010223671e-05,
+      "loss": 1.0869,
+      "step": 17734
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1689361132193404e-05,
+      "loss": 1.0505,
+      "step": 17735
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1680432162150095e-05,
+      "loss": 0.9865,
+      "step": 17736
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.167150319210679e-05,
+      "loss": 0.9669,
+      "step": 17737
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.166257422206349e-05,
+      "loss": 1.1026,
+      "step": 17738
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.165364525202018e-05,
+      "loss": 0.9803,
+      "step": 17739
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.164471628197688e-05,
+      "loss": 1.0308,
+      "step": 17740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1635787311933574e-05,
+      "loss": 1.0528,
+      "step": 17741
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1626858341890265e-05,
+      "loss": 1.1828,
+      "step": 17742
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.161792937184696e-05,
+      "loss": 0.9472,
+      "step": 17743
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.160900040180365e-05,
+      "loss": 1.0684,
+      "step": 17744
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.160007143176034e-05,
+      "loss": 0.9865,
+      "step": 17745
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.159114246171704e-05,
+      "loss": 1.0589,
+      "step": 17746
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.158221349167374e-05,
+      "loss": 1.0298,
+      "step": 17747
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1573284521630435e-05,
+      "loss": 0.9628,
+      "step": 17748
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1564355551587125e-05,
+      "loss": 0.906,
+      "step": 17749
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.155542658154382e-05,
+      "loss": 1.1278,
+      "step": 17750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.154649761150052e-05,
+      "loss": 0.905,
+      "step": 17751
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.153756864145721e-05,
+      "loss": 0.8113,
+      "step": 17752
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.152863967141391e-05,
+      "loss": 1.0441,
+      "step": 17753
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.15197107013706e-05,
+      "loss": 0.9497,
+      "step": 17754
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1510781731327295e-05,
+      "loss": 1.0619,
+      "step": 17755
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1501852761283985e-05,
+      "loss": 0.9517,
+      "step": 17756
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.149292379124068e-05,
+      "loss": 1.003,
+      "step": 17757
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.148399482119737e-05,
+      "loss": 1.0357,
+      "step": 17758
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.147506585115407e-05,
+      "loss": 1.1069,
+      "step": 17759
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.146613688111077e-05,
+      "loss": 1.0363,
+      "step": 17760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.145720791106746e-05,
+      "loss": 1.0825,
+      "step": 17761
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1448278941024155e-05,
+      "loss": 1.1398,
+      "step": 17762
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.143934997098085e-05,
+      "loss": 1.0623,
+      "step": 17763
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.143042100093754e-05,
+      "loss": 1.0551,
+      "step": 17764
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.142149203089424e-05,
+      "loss": 1.1442,
+      "step": 17765
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.141256306085093e-05,
+      "loss": 0.9139,
+      "step": 17766
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.140363409080762e-05,
+      "loss": 1.107,
+      "step": 17767
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.139470512076432e-05,
+      "loss": 0.9263,
+      "step": 17768
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1385776150721016e-05,
+      "loss": 0.9639,
+      "step": 17769
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1376847180677706e-05,
+      "loss": 0.9391,
+      "step": 17770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1367918210634404e-05,
+      "loss": 1.1109,
+      "step": 17771
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.13589892405911e-05,
+      "loss": 1.0094,
+      "step": 17772
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.13500602705478e-05,
+      "loss": 0.9394,
+      "step": 17773
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.134113130050449e-05,
+      "loss": 1.0646,
+      "step": 17774
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1332202330461186e-05,
+      "loss": 1.0474,
+      "step": 17775
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.132327336041788e-05,
+      "loss": 0.8811,
+      "step": 17776
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1314344390374574e-05,
+      "loss": 0.9759,
+      "step": 17777
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1305415420331264e-05,
+      "loss": 1.1054,
+      "step": 17778
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.129648645028796e-05,
+      "loss": 1.013,
+      "step": 17779
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.128755748024465e-05,
+      "loss": 0.9486,
+      "step": 17780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.127862851020135e-05,
+      "loss": 1.011,
+      "step": 17781
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1269699540158046e-05,
+      "loss": 1.1577,
+      "step": 17782
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.126077057011474e-05,
+      "loss": 0.9823,
+      "step": 17783
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1251841600071434e-05,
+      "loss": 0.9299,
+      "step": 17784
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.124291263002813e-05,
+      "loss": 1.1019,
+      "step": 17785
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.123398365998482e-05,
+      "loss": 1.0547,
+      "step": 17786
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.122505468994152e-05,
+      "loss": 1.0353,
+      "step": 17787
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1216125719898216e-05,
+      "loss": 1.0567,
+      "step": 17788
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.120719674985491e-05,
+      "loss": 0.9364,
+      "step": 17789
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.11982677798116e-05,
+      "loss": 0.9952,
+      "step": 17790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1189338809768295e-05,
+      "loss": 1.1119,
+      "step": 17791
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1180409839724985e-05,
+      "loss": 1.1277,
+      "step": 17792
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.117148086968168e-05,
+      "loss": 0.9745,
+      "step": 17793
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.116255189963838e-05,
+      "loss": 1.0449,
+      "step": 17794
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.115362292959508e-05,
+      "loss": 1.0462,
+      "step": 17795
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.114469395955177e-05,
+      "loss": 1.0074,
+      "step": 17796
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1135764989508465e-05,
+      "loss": 1.0194,
+      "step": 17797
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.112683601946516e-05,
+      "loss": 0.9928,
+      "step": 17798
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.111790704942185e-05,
+      "loss": 0.923,
+      "step": 17799
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.110897807937854e-05,
+      "loss": 0.9515,
+      "step": 17800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.110004910933524e-05,
+      "loss": 1.0395,
+      "step": 17801
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.109112013929193e-05,
+      "loss": 1.1158,
+      "step": 17802
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.108219116924863e-05,
+      "loss": 0.9891,
+      "step": 17803
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1073262199205325e-05,
+      "loss": 1.2511,
+      "step": 17804
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1064333229162016e-05,
+      "loss": 1.035,
+      "step": 17805
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.105540425911871e-05,
+      "loss": 1.006,
+      "step": 17806
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.104647528907541e-05,
+      "loss": 1.008,
+      "step": 17807
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.10375463190321e-05,
+      "loss": 1.0325,
+      "step": 17808
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.10286173489888e-05,
+      "loss": 1.003,
+      "step": 17809
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1019688378945495e-05,
+      "loss": 0.9306,
+      "step": 17810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1010759408902186e-05,
+      "loss": 1.0423,
+      "step": 17811
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1001830438858876e-05,
+      "loss": 0.9871,
+      "step": 17812
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.099290146881557e-05,
+      "loss": 1.0912,
+      "step": 17813
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0983972498772264e-05,
+      "loss": 1.0106,
+      "step": 17814
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.097504352872896e-05,
+      "loss": 0.9392,
+      "step": 17815
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.096611455868566e-05,
+      "loss": 0.9857,
+      "step": 17816
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.095718558864235e-05,
+      "loss": 1.1026,
+      "step": 17817
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0948256618599046e-05,
+      "loss": 0.9518,
+      "step": 17818
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.093932764855574e-05,
+      "loss": 1.008,
+      "step": 17819
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.093039867851244e-05,
+      "loss": 1.0215,
+      "step": 17820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.092146970846913e-05,
+      "loss": 0.9602,
+      "step": 17821
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.091254073842583e-05,
+      "loss": 0.9217,
+      "step": 17822
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.090361176838252e-05,
+      "loss": 1.1057,
+      "step": 17823
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.089468279833921e-05,
+      "loss": 1.0158,
+      "step": 17824
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0885753828295906e-05,
+      "loss": 1.0127,
+      "step": 17825
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0876824858252604e-05,
+      "loss": 0.9464,
+      "step": 17826
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0867895888209294e-05,
+      "loss": 1.0362,
+      "step": 17827
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.085896691816599e-05,
+      "loss": 1.0031,
+      "step": 17828
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.085003794812269e-05,
+      "loss": 1.1754,
+      "step": 17829
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.084110897807938e-05,
+      "loss": 0.9912,
+      "step": 17830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0832180008036076e-05,
+      "loss": 0.9567,
+      "step": 17831
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0823251037992774e-05,
+      "loss": 1.0198,
+      "step": 17832
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0814322067949464e-05,
+      "loss": 0.8772,
+      "step": 17833
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.080539309790616e-05,
+      "loss": 0.9993,
+      "step": 17834
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.079646412786285e-05,
+      "loss": 1.0762,
+      "step": 17835
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.078753515781954e-05,
+      "loss": 1.016,
+      "step": 17836
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.077860618777624e-05,
+      "loss": 1.0563,
+      "step": 17837
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.076967721773294e-05,
+      "loss": 1.1277,
+      "step": 17838
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.076074824768963e-05,
+      "loss": 1.0641,
+      "step": 17839
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0751819277646325e-05,
+      "loss": 1.1817,
+      "step": 17840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.074289030760302e-05,
+      "loss": 1.1075,
+      "step": 17841
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.073396133755971e-05,
+      "loss": 1.0061,
+      "step": 17842
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.072503236751641e-05,
+      "loss": 1.0106,
+      "step": 17843
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.071610339747311e-05,
+      "loss": 1.0128,
+      "step": 17844
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.07071744274298e-05,
+      "loss": 1.0675,
+      "step": 17845
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0698245457386495e-05,
+      "loss": 1.1085,
+      "step": 17846
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0689316487343185e-05,
+      "loss": 1.1135,
+      "step": 17847
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.068038751729988e-05,
+      "loss": 1.0283,
+      "step": 17848
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.067145854725657e-05,
+      "loss": 0.9915,
+      "step": 17849
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.066252957721327e-05,
+      "loss": 1.0669,
+      "step": 17850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.065360060716997e-05,
+      "loss": 1.143,
+      "step": 17851
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.064467163712666e-05,
+      "loss": 1.065,
+      "step": 17852
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0635742667083355e-05,
+      "loss": 1.1088,
+      "step": 17853
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.062681369704005e-05,
+      "loss": 1.118,
+      "step": 17854
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.061788472699674e-05,
+      "loss": 1.0242,
+      "step": 17855
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.060895575695344e-05,
+      "loss": 1.1207,
+      "step": 17856
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.060002678691013e-05,
+      "loss": 0.947,
+      "step": 17857
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.059109781686682e-05,
+      "loss": 0.977,
+      "step": 17858
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.058216884682352e-05,
+      "loss": 1.0185,
+      "step": 17859
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0573239876780216e-05,
+      "loss": 1.0963,
+      "step": 17860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0564310906736906e-05,
+      "loss": 1.0116,
+      "step": 17861
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.05553819366936e-05,
+      "loss": 1.1816,
+      "step": 17862
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.05464529666503e-05,
+      "loss": 1.2158,
+      "step": 17863
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.053752399660699e-05,
+      "loss": 0.9859,
+      "step": 17864
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.052859502656369e-05,
+      "loss": 1.0196,
+      "step": 17865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0519666056520386e-05,
+      "loss": 1.0153,
+      "step": 17866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0510737086477076e-05,
+      "loss": 0.9884,
+      "step": 17867
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.050180811643377e-05,
+      "loss": 0.9501,
+      "step": 17868
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0492879146390464e-05,
+      "loss": 1.0532,
+      "step": 17869
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.048395017634716e-05,
+      "loss": 1.0261,
+      "step": 17870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.047502120630385e-05,
+      "loss": 1.0592,
+      "step": 17871
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.046609223626055e-05,
+      "loss": 1.0081,
+      "step": 17872
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0457163266217246e-05,
+      "loss": 1.02,
+      "step": 17873
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0448234296173937e-05,
+      "loss": 1.049,
+      "step": 17874
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0439305326130634e-05,
+      "loss": 1.0093,
+      "step": 17875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.043037635608733e-05,
+      "loss": 1.084,
+      "step": 17876
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.042144738604402e-05,
+      "loss": 1.1217,
+      "step": 17877
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.041251841600072e-05,
+      "loss": 1.0246,
+      "step": 17878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0403589445957416e-05,
+      "loss": 0.9479,
+      "step": 17879
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0394660475914107e-05,
+      "loss": 0.9649,
+      "step": 17880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.03857315058708e-05,
+      "loss": 0.9764,
+      "step": 17881
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0376802535827494e-05,
+      "loss": 1.1088,
+      "step": 17882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.0367873565784185e-05,
+      "loss": 0.9697,
+      "step": 17883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.035894459574088e-05,
+      "loss": 0.9717,
+      "step": 17884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.035001562569758e-05,
+      "loss": 1.0067,
+      "step": 17885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.034108665565427e-05,
+      "loss": 1.0422,
+      "step": 17886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.033215768561097e-05,
+      "loss": 1.1101,
+      "step": 17887
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0323228715567664e-05,
+      "loss": 1.0495,
+      "step": 17888
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0314299745524355e-05,
+      "loss": 1.1019,
+      "step": 17889
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.030537077548105e-05,
+      "loss": 0.977,
+      "step": 17890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.029644180543774e-05,
+      "loss": 1.0474,
+      "step": 17891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.028751283539444e-05,
+      "loss": 1.0947,
+      "step": 17892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.027858386535113e-05,
+      "loss": 1.1332,
+      "step": 17893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.026965489530783e-05,
+      "loss": 0.9653,
+      "step": 17894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0260725925264525e-05,
+      "loss": 1.0061,
+      "step": 17895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0251796955221215e-05,
+      "loss": 1.0012,
+      "step": 17896
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.024286798517791e-05,
+      "loss": 0.9508,
+      "step": 17897
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.023393901513461e-05,
+      "loss": 1.0435,
+      "step": 17898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.02250100450913e-05,
+      "loss": 1.167,
+      "step": 17899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0216081075048e-05,
+      "loss": 0.9601,
+      "step": 17900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0207152105004695e-05,
+      "loss": 1.0034,
+      "step": 17901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0198223134961385e-05,
+      "loss": 1.0712,
+      "step": 17902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0189294164918076e-05,
+      "loss": 1.1834,
+      "step": 17903
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.018036519487477e-05,
+      "loss": 1.065,
+      "step": 17904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.017143622483146e-05,
+      "loss": 1.2435,
+      "step": 17905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.016250725478816e-05,
+      "loss": 0.9895,
+      "step": 17906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.015357828474486e-05,
+      "loss": 1.0254,
+      "step": 17907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.014464931470155e-05,
+      "loss": 1.0249,
+      "step": 17908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0135720344658246e-05,
+      "loss": 1.1378,
+      "step": 17909
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.012679137461494e-05,
+      "loss": 1.0113,
+      "step": 17910
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.011786240457163e-05,
+      "loss": 0.9914,
+      "step": 17911
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.010893343452833e-05,
+      "loss": 1.1653,
+      "step": 17912
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.010000446448503e-05,
+      "loss": 1.0915,
+      "step": 17913
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.009107549444172e-05,
+      "loss": 1.0855,
+      "step": 17914
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.008214652439841e-05,
+      "loss": 1.1307,
+      "step": 17915
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0073217554355106e-05,
+      "loss": 0.9924,
+      "step": 17916
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.00642885843118e-05,
+      "loss": 1.1213,
+      "step": 17917
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0055359614268494e-05,
+      "loss": 1.1157,
+      "step": 17918
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.004643064422519e-05,
+      "loss": 1.0222,
+      "step": 17919
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.003750167418189e-05,
+      "loss": 1.0327,
+      "step": 17920
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.002857270413858e-05,
+      "loss": 1.1185,
+      "step": 17921
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0019643734095276e-05,
+      "loss": 1.1082,
+      "step": 17922
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.001071476405197e-05,
+      "loss": 0.9988,
+      "step": 17923
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0001785794008664e-05,
+      "loss": 1.0324,
+      "step": 17924
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.999285682396536e-05,
+      "loss": 1.0187,
+      "step": 17925
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.998392785392205e-05,
+      "loss": 1.11,
+      "step": 17926
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.997499888387874e-05,
+      "loss": 0.9612,
+      "step": 17927
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.996606991383544e-05,
+      "loss": 1.0636,
+      "step": 17928
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9957140943792137e-05,
+      "loss": 1.1647,
+      "step": 17929
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.994821197374883e-05,
+      "loss": 1.0211,
+      "step": 17930
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9939283003705524e-05,
+      "loss": 0.9125,
+      "step": 17931
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.993035403366222e-05,
+      "loss": 1.0236,
+      "step": 17932
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.992142506361891e-05,
+      "loss": 1.1214,
+      "step": 17933
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.991249609357561e-05,
+      "loss": 1.0661,
+      "step": 17934
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9903567123532307e-05,
+      "loss": 0.9702,
+      "step": 17935
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9894638153489e-05,
+      "loss": 0.919,
+      "step": 17936
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.988570918344569e-05,
+      "loss": 1.0314,
+      "step": 17937
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9876780213402385e-05,
+      "loss": 1.0362,
+      "step": 17938
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9867851243359075e-05,
+      "loss": 1.0619,
+      "step": 17939
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.985892227331577e-05,
+      "loss": 1.0207,
+      "step": 17940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.984999330327247e-05,
+      "loss": 1.036,
+      "step": 17941
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.984106433322917e-05,
+      "loss": 1.0588,
+      "step": 17942
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.983213536318586e-05,
+      "loss": 0.932,
+      "step": 17943
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9823206393142555e-05,
+      "loss": 1.0416,
+      "step": 17944
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.981427742309925e-05,
+      "loss": 0.9894,
+      "step": 17945
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.980534845305594e-05,
+      "loss": 1.0493,
+      "step": 17946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.979641948301264e-05,
+      "loss": 1.0284,
+      "step": 17947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.978749051296933e-05,
+      "loss": 0.8797,
+      "step": 17948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.977856154292602e-05,
+      "loss": 0.9366,
+      "step": 17949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.976963257288272e-05,
+      "loss": 1.0343,
+      "step": 17950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9760703602839415e-05,
+      "loss": 0.9873,
+      "step": 17951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9751774632796106e-05,
+      "loss": 1.1075,
+      "step": 17952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.97428456627528e-05,
+      "loss": 1.1616,
+      "step": 17953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.97339166927095e-05,
+      "loss": 1.002,
+      "step": 17954
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.972498772266619e-05,
+      "loss": 1.0181,
+      "step": 17955
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.971605875262289e-05,
+      "loss": 1.0617,
+      "step": 17956
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9707129782579585e-05,
+      "loss": 1.0239,
+      "step": 17957
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9698200812536276e-05,
+      "loss": 0.9139,
+      "step": 17958
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.968927184249297e-05,
+      "loss": 0.9755,
+      "step": 17959
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9680342872449663e-05,
+      "loss": 0.9298,
+      "step": 17960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.9671413902406354e-05,
+      "loss": 1.0978,
+      "step": 17961
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.966248493236305e-05,
+      "loss": 1.0981,
+      "step": 17962
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.965355596231975e-05,
+      "loss": 0.9983,
+      "step": 17963
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.964462699227644e-05,
+      "loss": 1.0502,
+      "step": 17964
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9635698022233136e-05,
+      "loss": 1.0813,
+      "step": 17965
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9626769052189833e-05,
+      "loss": 1.0151,
+      "step": 17966
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.961784008214653e-05,
+      "loss": 1.1097,
+      "step": 17967
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.960891111210322e-05,
+      "loss": 0.957,
+      "step": 17968
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.959998214205992e-05,
+      "loss": 0.8835,
+      "step": 17969
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.959105317201661e-05,
+      "loss": 1.1543,
+      "step": 17970
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9582124201973306e-05,
+      "loss": 1.0341,
+      "step": 17971
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.957319523193e-05,
+      "loss": 0.9449,
+      "step": 17972
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9564266261886694e-05,
+      "loss": 0.9768,
+      "step": 17973
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9555337291843384e-05,
+      "loss": 1.0377,
+      "step": 17974
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.954640832180008e-05,
+      "loss": 1.0278,
+      "step": 17975
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.953747935175678e-05,
+      "loss": 1.1045,
+      "step": 17976
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.952855038171347e-05,
+      "loss": 0.9942,
+      "step": 17977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9519621411670167e-05,
+      "loss": 1.0581,
+      "step": 17978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9510692441626864e-05,
+      "loss": 1.0297,
+      "step": 17979
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9501763471583554e-05,
+      "loss": 1.0832,
+      "step": 17980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.949283450154025e-05,
+      "loss": 0.968,
+      "step": 17981
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.948390553149694e-05,
+      "loss": 0.9864,
+      "step": 17982
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.947497656145364e-05,
+      "loss": 1.0884,
+      "step": 17983
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.946604759141033e-05,
+      "loss": 1.0137,
+      "step": 17984
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.945711862136703e-05,
+      "loss": 0.9887,
+      "step": 17985
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.944818965132372e-05,
+      "loss": 1.0432,
+      "step": 17986
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9439260681280415e-05,
+      "loss": 1.0248,
+      "step": 17987
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.943033171123711e-05,
+      "loss": 1.0094,
+      "step": 17988
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.94214027411938e-05,
+      "loss": 1.0271,
+      "step": 17989
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.94124737711505e-05,
+      "loss": 1.1963,
+      "step": 17990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.94035448011072e-05,
+      "loss": 1.0797,
+      "step": 17991
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9394615831063894e-05,
+      "loss": 1.0023,
+      "step": 17992
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9385686861020585e-05,
+      "loss": 0.9886,
+      "step": 17993
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9376757890977275e-05,
+      "loss": 1.0709,
+      "step": 17994
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.936782892093397e-05,
+      "loss": 1.1032,
+      "step": 17995
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.935889995089066e-05,
+      "loss": 0.9682,
+      "step": 17996
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.934997098084736e-05,
+      "loss": 1.1744,
+      "step": 17997
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.934104201080406e-05,
+      "loss": 1.0978,
+      "step": 17998
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.933211304076075e-05,
+      "loss": 1.1406,
+      "step": 17999
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9323184070717445e-05,
+      "loss": 0.9022,
+      "step": 18000
+    },
+    {
+      "epoch": 2.41,
+      "eval_loss": 1.0255261659622192,
+      "eval_runtime": 3047.0293,
+      "eval_samples_per_second": 6.172,
+      "eval_steps_per_second": 0.772,
+      "step": 18000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.931425510067414e-05,
+      "loss": 0.8997,
+      "step": 18001
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.930532613063083e-05,
+      "loss": 1.1423,
+      "step": 18002
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.929639716058753e-05,
+      "loss": 1.0655,
+      "step": 18003
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.928746819054423e-05,
+      "loss": 1.236,
+      "step": 18004
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.927853922050092e-05,
+      "loss": 1.1433,
+      "step": 18005
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.926961025045761e-05,
+      "loss": 0.9061,
+      "step": 18006
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9260681280414306e-05,
+      "loss": 1.0591,
+      "step": 18007
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9251752310370996e-05,
+      "loss": 1.0319,
+      "step": 18008
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9242823340327693e-05,
+      "loss": 0.9855,
+      "step": 18009
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.923389437028439e-05,
+      "loss": 1.119,
+      "step": 18010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.922496540024108e-05,
+      "loss": 1.0093,
+      "step": 18011
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.921603643019778e-05,
+      "loss": 1.1776,
+      "step": 18012
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9207107460154476e-05,
+      "loss": 1.0624,
+      "step": 18013
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9198178490111166e-05,
+      "loss": 1.1107,
+      "step": 18014
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9189249520067863e-05,
+      "loss": 1.0238,
+      "step": 18015
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.918032055002456e-05,
+      "loss": 1.1428,
+      "step": 18016
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.917139157998125e-05,
+      "loss": 0.9779,
+      "step": 18017
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.916246260993794e-05,
+      "loss": 1.0226,
+      "step": 18018
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.915353363989464e-05,
+      "loss": 1.0227,
+      "step": 18019
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9144604669851336e-05,
+      "loss": 1.0935,
+      "step": 18020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.913567569980803e-05,
+      "loss": 1.0262,
+      "step": 18021
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9126746729764724e-05,
+      "loss": 0.951,
+      "step": 18022
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.911781775972142e-05,
+      "loss": 1.0342,
+      "step": 18023
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.910888878967811e-05,
+      "loss": 1.0635,
+      "step": 18024
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.909995981963481e-05,
+      "loss": 1.13,
+      "step": 18025
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9091030849591506e-05,
+      "loss": 0.9419,
+      "step": 18026
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.90821018795482e-05,
+      "loss": 0.9771,
+      "step": 18027
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.907317290950489e-05,
+      "loss": 1.0274,
+      "step": 18028
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9064243939461584e-05,
+      "loss": 1.0472,
+      "step": 18029
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9055314969418275e-05,
+      "loss": 0.9431,
+      "step": 18030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.904638599937497e-05,
+      "loss": 0.9382,
+      "step": 18031
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.903745702933167e-05,
+      "loss": 1.0726,
+      "step": 18032
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.902852805928836e-05,
+      "loss": 1.0274,
+      "step": 18033
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.901959908924506e-05,
+      "loss": 0.968,
+      "step": 18034
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9010670119201754e-05,
+      "loss": 0.9142,
+      "step": 18035
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9001741149158445e-05,
+      "loss": 1.0502,
+      "step": 18036
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.899281217911514e-05,
+      "loss": 1.01,
+      "step": 18037
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.898388320907184e-05,
+      "loss": 0.9777,
+      "step": 18038
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.897495423902853e-05,
+      "loss": 1.0086,
+      "step": 18039
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.896602526898522e-05,
+      "loss": 0.865,
+      "step": 18040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.895709629894192e-05,
+      "loss": 1.0989,
+      "step": 18041
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8948167328898615e-05,
+      "loss": 1.0186,
+      "step": 18042
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8939238358855305e-05,
+      "loss": 0.9128,
+      "step": 18043
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8930309388812e-05,
+      "loss": 1.0658,
+      "step": 18044
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.89213804187687e-05,
+      "loss": 1.0876,
+      "step": 18045
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.891245144872539e-05,
+      "loss": 1.1136,
+      "step": 18046
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.890352247868209e-05,
+      "loss": 0.9654,
+      "step": 18047
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8894593508638785e-05,
+      "loss": 1.0839,
+      "step": 18048
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8885664538595475e-05,
+      "loss": 1.0814,
+      "step": 18049
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.887673556855217e-05,
+      "loss": 1.0059,
+      "step": 18050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.886780659850886e-05,
+      "loss": 1.0416,
+      "step": 18051
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8858877628465554e-05,
+      "loss": 1.0004,
+      "step": 18052
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.884994865842225e-05,
+      "loss": 0.9726,
+      "step": 18053
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.884101968837895e-05,
+      "loss": 0.9817,
+      "step": 18054
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.883209071833564e-05,
+      "loss": 0.994,
+      "step": 18055
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8823161748292336e-05,
+      "loss": 1.0887,
+      "step": 18056
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.881423277824903e-05,
+      "loss": 1.2074,
+      "step": 18057
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8805303808205724e-05,
+      "loss": 0.9583,
+      "step": 18058
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.879637483816242e-05,
+      "loss": 1.0171,
+      "step": 18059
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.878744586811912e-05,
+      "loss": 1.0383,
+      "step": 18060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.877851689807581e-05,
+      "loss": 0.9716,
+      "step": 18061
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8769587928032506e-05,
+      "loss": 1.0789,
+      "step": 18062
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8760658957989196e-05,
+      "loss": 1.0603,
+      "step": 18063
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8751729987945893e-05,
+      "loss": 1.0307,
+      "step": 18064
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8742801017902584e-05,
+      "loss": 0.9188,
+      "step": 18065
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.873387204785928e-05,
+      "loss": 1.14,
+      "step": 18066
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.872494307781598e-05,
+      "loss": 1.0231,
+      "step": 18067
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.871601410777267e-05,
+      "loss": 1.0515,
+      "step": 18068
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8707085137729366e-05,
+      "loss": 1.0294,
+      "step": 18069
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8698156167686063e-05,
+      "loss": 0.9999,
+      "step": 18070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8689227197642754e-05,
+      "loss": 1.0873,
+      "step": 18071
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.868029822759945e-05,
+      "loss": 0.9884,
+      "step": 18072
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.867136925755614e-05,
+      "loss": 0.9933,
+      "step": 18073
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.866244028751283e-05,
+      "loss": 1.0858,
+      "step": 18074
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.865351131746953e-05,
+      "loss": 0.974,
+      "step": 18075
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.864458234742623e-05,
+      "loss": 1.0492,
+      "step": 18076
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.863565337738292e-05,
+      "loss": 1.0424,
+      "step": 18077
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8626724407339614e-05,
+      "loss": 1.0212,
+      "step": 18078
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.861779543729631e-05,
+      "loss": 1.0839,
+      "step": 18079
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8608866467253e-05,
+      "loss": 0.9274,
+      "step": 18080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.85999374972097e-05,
+      "loss": 0.98,
+      "step": 18081
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.85910085271664e-05,
+      "loss": 1.018,
+      "step": 18082
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.858207955712309e-05,
+      "loss": 1.0856,
+      "step": 18083
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8573150587079784e-05,
+      "loss": 0.9718,
+      "step": 18084
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8564221617036475e-05,
+      "loss": 1.0395,
+      "step": 18085
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8555292646993165e-05,
+      "loss": 1.0251,
+      "step": 18086
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.854636367694986e-05,
+      "loss": 1.0739,
+      "step": 18087
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.853743470690656e-05,
+      "loss": 1.0674,
+      "step": 18088
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.852850573686326e-05,
+      "loss": 0.9381,
+      "step": 18089
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.851957676681995e-05,
+      "loss": 1.0317,
+      "step": 18090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8510647796776645e-05,
+      "loss": 1.0236,
+      "step": 18091
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.850171882673334e-05,
+      "loss": 1.0037,
+      "step": 18092
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.849278985669003e-05,
+      "loss": 1.1169,
+      "step": 18093
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.848386088664673e-05,
+      "loss": 0.9559,
+      "step": 18094
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.847493191660343e-05,
+      "loss": 1.0357,
+      "step": 18095
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.846600294656012e-05,
+      "loss": 0.9714,
+      "step": 18096
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.845707397651681e-05,
+      "loss": 0.8986,
+      "step": 18097
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8448145006473505e-05,
+      "loss": 1.0124,
+      "step": 18098
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8439216036430196e-05,
+      "loss": 1.1751,
+      "step": 18099
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.843028706638689e-05,
+      "loss": 0.9209,
+      "step": 18100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.842135809634359e-05,
+      "loss": 1.0339,
+      "step": 18101
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.841242912630028e-05,
+      "loss": 1.0246,
+      "step": 18102
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.840350015625698e-05,
+      "loss": 1.097,
+      "step": 18103
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8394571186213675e-05,
+      "loss": 1.0041,
+      "step": 18104
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8385642216170366e-05,
+      "loss": 1.077,
+      "step": 18105
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.837671324612706e-05,
+      "loss": 1.0066,
+      "step": 18106
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8367784276083754e-05,
+      "loss": 0.9664,
+      "step": 18107
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.835885530604045e-05,
+      "loss": 1.0071,
+      "step": 18108
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.834992633599714e-05,
+      "loss": 1.0043,
+      "step": 18109
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.834099736595384e-05,
+      "loss": 1.0413,
+      "step": 18110
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.833206839591053e-05,
+      "loss": 0.9866,
+      "step": 18111
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8323139425867226e-05,
+      "loss": 1.1017,
+      "step": 18112
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8314210455823924e-05,
+      "loss": 1.0916,
+      "step": 18113
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.830528148578062e-05,
+      "loss": 0.9069,
+      "step": 18114
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.829635251573731e-05,
+      "loss": 0.9861,
+      "step": 18115
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.828742354569401e-05,
+      "loss": 1.0128,
+      "step": 18116
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8278494575650706e-05,
+      "loss": 1.1219,
+      "step": 18117
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8269565605607396e-05,
+      "loss": 0.9894,
+      "step": 18118
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.826063663556409e-05,
+      "loss": 1.1039,
+      "step": 18119
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8251707665520784e-05,
+      "loss": 1.1487,
+      "step": 18120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8242778695477474e-05,
+      "loss": 1.0502,
+      "step": 18121
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.823384972543417e-05,
+      "loss": 1.0192,
+      "step": 18122
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.822492075539087e-05,
+      "loss": 1.1277,
+      "step": 18123
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.821599178534756e-05,
+      "loss": 0.9775,
+      "step": 18124
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.820706281530426e-05,
+      "loss": 1.0308,
+      "step": 18125
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8198133845260954e-05,
+      "loss": 1.0738,
+      "step": 18126
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8189204875217644e-05,
+      "loss": 1.0551,
+      "step": 18127
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.818027590517434e-05,
+      "loss": 1.0564,
+      "step": 18128
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.817134693513104e-05,
+      "loss": 1.0156,
+      "step": 18129
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.816241796508773e-05,
+      "loss": 1.0372,
+      "step": 18130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.815348899504442e-05,
+      "loss": 1.0421,
+      "step": 18131
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.814456002500112e-05,
+      "loss": 1.0936,
+      "step": 18132
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.813563105495781e-05,
+      "loss": 1.096,
+      "step": 18133
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8126702084914505e-05,
+      "loss": 1.0437,
+      "step": 18134
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.81177731148712e-05,
+      "loss": 0.9655,
+      "step": 18135
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.810884414482789e-05,
+      "loss": 1.0412,
+      "step": 18136
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.809991517478459e-05,
+      "loss": 0.9324,
+      "step": 18137
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.809098620474129e-05,
+      "loss": 0.9953,
+      "step": 18138
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8082057234697984e-05,
+      "loss": 1.0788,
+      "step": 18139
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8073128264654675e-05,
+      "loss": 1.0438,
+      "step": 18140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.806419929461137e-05,
+      "loss": 1.1268,
+      "step": 18141
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.805527032456806e-05,
+      "loss": 0.9627,
+      "step": 18142
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.804634135452475e-05,
+      "loss": 1.1529,
+      "step": 18143
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.803741238448145e-05,
+      "loss": 0.9806,
+      "step": 18144
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.802848341443815e-05,
+      "loss": 0.9668,
+      "step": 18145
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.801955444439484e-05,
+      "loss": 1.1872,
+      "step": 18146
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8010625474351535e-05,
+      "loss": 0.9689,
+      "step": 18147
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.800169650430823e-05,
+      "loss": 1.0592,
+      "step": 18148
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.799276753426492e-05,
+      "loss": 0.8095,
+      "step": 18149
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.798383856422162e-05,
+      "loss": 1.1247,
+      "step": 18150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.797490959417832e-05,
+      "loss": 1.1038,
+      "step": 18151
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.796598062413501e-05,
+      "loss": 0.9703,
+      "step": 18152
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7957051654091705e-05,
+      "loss": 1.0449,
+      "step": 18153
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7948122684048396e-05,
+      "loss": 1.0834,
+      "step": 18154
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7939193714005086e-05,
+      "loss": 1.0842,
+      "step": 18155
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7930264743961784e-05,
+      "loss": 1.0978,
+      "step": 18156
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.792133577391848e-05,
+      "loss": 0.9886,
+      "step": 18157
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.791240680387517e-05,
+      "loss": 1.01,
+      "step": 18158
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.790347783383187e-05,
+      "loss": 1.0546,
+      "step": 18159
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7894548863788566e-05,
+      "loss": 0.9597,
+      "step": 18160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7885619893745256e-05,
+      "loss": 0.9628,
+      "step": 18161
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7876690923701954e-05,
+      "loss": 1.1279,
+      "step": 18162
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.786776195365865e-05,
+      "loss": 1.0093,
+      "step": 18163
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.785883298361534e-05,
+      "loss": 0.9243,
+      "step": 18164
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.784990401357203e-05,
+      "loss": 0.986,
+      "step": 18165
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.784097504352873e-05,
+      "loss": 1.0695,
+      "step": 18166
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7832046073485426e-05,
+      "loss": 1.1464,
+      "step": 18167
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.782311710344212e-05,
+      "loss": 1.0412,
+      "step": 18168
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7814188133398814e-05,
+      "loss": 1.1145,
+      "step": 18169
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.780525916335551e-05,
+      "loss": 0.93,
+      "step": 18170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.77963301933122e-05,
+      "loss": 1.0376,
+      "step": 18171
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.77874012232689e-05,
+      "loss": 1.0516,
+      "step": 18172
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7778472253225596e-05,
+      "loss": 0.972,
+      "step": 18173
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.776954328318229e-05,
+      "loss": 0.9808,
+      "step": 18174
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7760614313138984e-05,
+      "loss": 1.0413,
+      "step": 18175
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7751685343095675e-05,
+      "loss": 0.9687,
+      "step": 18176
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7742756373052365e-05,
+      "loss": 1.0183,
+      "step": 18177
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.773382740300906e-05,
+      "loss": 1.0443,
+      "step": 18178
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.772489843296576e-05,
+      "loss": 0.9474,
+      "step": 18179
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.771596946292245e-05,
+      "loss": 1.0355,
+      "step": 18180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.770704049287915e-05,
+      "loss": 1.0626,
+      "step": 18181
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7698111522835845e-05,
+      "loss": 1.0974,
+      "step": 18182
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7689182552792535e-05,
+      "loss": 1.0186,
+      "step": 18183
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.768025358274923e-05,
+      "loss": 1.043,
+      "step": 18184
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.767132461270593e-05,
+      "loss": 1.0268,
+      "step": 18185
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.766239564266263e-05,
+      "loss": 0.9799,
+      "step": 18186
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.765346667261932e-05,
+      "loss": 0.8492,
+      "step": 18187
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.764453770257601e-05,
+      "loss": 1.2364,
+      "step": 18188
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7635608732532705e-05,
+      "loss": 0.955,
+      "step": 18189
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7626679762489395e-05,
+      "loss": 0.9921,
+      "step": 18190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.761775079244609e-05,
+      "loss": 1.0277,
+      "step": 18191
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.760882182240279e-05,
+      "loss": 1.0046,
+      "step": 18192
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.759989285235948e-05,
+      "loss": 1.0699,
+      "step": 18193
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.759096388231618e-05,
+      "loss": 1.0078,
+      "step": 18194
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7582034912272875e-05,
+      "loss": 0.9489,
+      "step": 18195
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7573105942229565e-05,
+      "loss": 1.0568,
+      "step": 18196
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.756417697218626e-05,
+      "loss": 1.0363,
+      "step": 18197
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.755524800214295e-05,
+      "loss": 1.0816,
+      "step": 18198
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.754631903209965e-05,
+      "loss": 0.9158,
+      "step": 18199
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.753739006205634e-05,
+      "loss": 1.0328,
+      "step": 18200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.752846109201304e-05,
+      "loss": 1.092,
+      "step": 18201
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.751953212196973e-05,
+      "loss": 0.9751,
+      "step": 18202
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7510603151926426e-05,
+      "loss": 1.1753,
+      "step": 18203
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.750167418188312e-05,
+      "loss": 1.0532,
+      "step": 18204
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7492745211839814e-05,
+      "loss": 1.0128,
+      "step": 18205
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.748381624179651e-05,
+      "loss": 0.9572,
+      "step": 18206
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.747488727175321e-05,
+      "loss": 1.0494,
+      "step": 18207
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.74659583017099e-05,
+      "loss": 1.147,
+      "step": 18208
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7457029331666596e-05,
+      "loss": 0.884,
+      "step": 18209
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7448100361623286e-05,
+      "loss": 1.1246,
+      "step": 18210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7439171391579984e-05,
+      "loss": 1.087,
+      "step": 18211
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7430242421536674e-05,
+      "loss": 0.9442,
+      "step": 18212
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.742131345149337e-05,
+      "loss": 0.9125,
+      "step": 18213
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.741238448145007e-05,
+      "loss": 1.0582,
+      "step": 18214
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.740345551140676e-05,
+      "loss": 0.8899,
+      "step": 18215
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7394526541363456e-05,
+      "loss": 0.9958,
+      "step": 18216
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7385597571320154e-05,
+      "loss": 0.8965,
+      "step": 18217
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7376668601276844e-05,
+      "loss": 1.0495,
+      "step": 18218
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.736773963123354e-05,
+      "loss": 0.984,
+      "step": 18219
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.735881066119024e-05,
+      "loss": 1.0398,
+      "step": 18220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.734988169114693e-05,
+      "loss": 1.005,
+      "step": 18221
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.734095272110362e-05,
+      "loss": 0.9819,
+      "step": 18222
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.733202375106032e-05,
+      "loss": 1.114,
+      "step": 18223
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.732309478101701e-05,
+      "loss": 0.981,
+      "step": 18224
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7314165810973705e-05,
+      "loss": 0.96,
+      "step": 18225
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.73052368409304e-05,
+      "loss": 1.0256,
+      "step": 18226
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.729630787088709e-05,
+      "loss": 1.0184,
+      "step": 18227
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.728737890084379e-05,
+      "loss": 1.1248,
+      "step": 18228
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.727844993080049e-05,
+      "loss": 1.0007,
+      "step": 18229
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.726952096075718e-05,
+      "loss": 1.0196,
+      "step": 18230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7260591990713875e-05,
+      "loss": 1.0544,
+      "step": 18231
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.725166302067057e-05,
+      "loss": 0.935,
+      "step": 18232
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.724273405062726e-05,
+      "loss": 1.0383,
+      "step": 18233
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.723380508058395e-05,
+      "loss": 1.0468,
+      "step": 18234
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.722487611054065e-05,
+      "loss": 0.9914,
+      "step": 18235
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.721594714049735e-05,
+      "loss": 0.9719,
+      "step": 18236
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.720701817045404e-05,
+      "loss": 1.0193,
+      "step": 18237
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7198089200410735e-05,
+      "loss": 0.985,
+      "step": 18238
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.718916023036743e-05,
+      "loss": 1.1861,
+      "step": 18239
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.718023126032412e-05,
+      "loss": 0.9998,
+      "step": 18240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.717130229028082e-05,
+      "loss": 1.0162,
+      "step": 18241
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.716237332023752e-05,
+      "loss": 0.9349,
+      "step": 18242
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.715344435019421e-05,
+      "loss": 0.941,
+      "step": 18243
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.71445153801509e-05,
+      "loss": 1.0956,
+      "step": 18244
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7135586410107596e-05,
+      "loss": 1.0625,
+      "step": 18245
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7126657440064286e-05,
+      "loss": 1.1228,
+      "step": 18246
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.711772847002098e-05,
+      "loss": 1.0725,
+      "step": 18247
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.710879949997768e-05,
+      "loss": 0.9888,
+      "step": 18248
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.709987052993437e-05,
+      "loss": 1.0933,
+      "step": 18249
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.709094155989107e-05,
+      "loss": 1.0221,
+      "step": 18250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7082012589847765e-05,
+      "loss": 0.9865,
+      "step": 18251
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7073083619804456e-05,
+      "loss": 1.1586,
+      "step": 18252
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.706415464976115e-05,
+      "loss": 1.0392,
+      "step": 18253
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.705522567971785e-05,
+      "loss": 1.007,
+      "step": 18254
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.704629670967454e-05,
+      "loss": 1.1887,
+      "step": 18255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.703736773963123e-05,
+      "loss": 1.1454,
+      "step": 18256
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.702843876958793e-05,
+      "loss": 0.9966,
+      "step": 18257
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.701950979954462e-05,
+      "loss": 0.9956,
+      "step": 18258
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7010580829501316e-05,
+      "loss": 1.0892,
+      "step": 18259
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7001651859458014e-05,
+      "loss": 1.123,
+      "step": 18260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.699272288941471e-05,
+      "loss": 1.0068,
+      "step": 18261
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.69837939193714e-05,
+      "loss": 0.9815,
+      "step": 18262
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.69748649493281e-05,
+      "loss": 1.0165,
+      "step": 18263
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6965935979284796e-05,
+      "loss": 1.0364,
+      "step": 18264
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6957007009241486e-05,
+      "loss": 0.8886,
+      "step": 18265
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6948078039198184e-05,
+      "loss": 0.9923,
+      "step": 18266
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6939149069154874e-05,
+      "loss": 0.9781,
+      "step": 18267
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6930220099111565e-05,
+      "loss": 0.9584,
+      "step": 18268
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.692129112906826e-05,
+      "loss": 0.992,
+      "step": 18269
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.691236215902496e-05,
+      "loss": 1.0652,
+      "step": 18270
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.690343318898165e-05,
+      "loss": 0.9266,
+      "step": 18271
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.689450421893835e-05,
+      "loss": 1.1726,
+      "step": 18272
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6885575248895044e-05,
+      "loss": 1.118,
+      "step": 18273
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6876646278851735e-05,
+      "loss": 0.9958,
+      "step": 18274
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.686771730880843e-05,
+      "loss": 1.1743,
+      "step": 18275
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.685878833876513e-05,
+      "loss": 0.9466,
+      "step": 18276
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.684985936872182e-05,
+      "loss": 1.0098,
+      "step": 18277
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.684093039867852e-05,
+      "loss": 1.0093,
+      "step": 18278
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.683200142863521e-05,
+      "loss": 0.9757,
+      "step": 18279
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.68230724585919e-05,
+      "loss": 1.1122,
+      "step": 18280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6814143488548595e-05,
+      "loss": 1.0352,
+      "step": 18281
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.680521451850529e-05,
+      "loss": 1.0427,
+      "step": 18282
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.679628554846198e-05,
+      "loss": 1.0048,
+      "step": 18283
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.678735657841868e-05,
+      "loss": 0.8854,
+      "step": 18284
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.677842760837538e-05,
+      "loss": 1.0003,
+      "step": 18285
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6769498638332075e-05,
+      "loss": 0.9875,
+      "step": 18286
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6760569668288765e-05,
+      "loss": 1.0435,
+      "step": 18287
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.675164069824546e-05,
+      "loss": 0.9659,
+      "step": 18288
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.674271172820215e-05,
+      "loss": 0.9553,
+      "step": 18289
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.673378275815885e-05,
+      "loss": 0.9506,
+      "step": 18290
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.672485378811554e-05,
+      "loss": 0.9768,
+      "step": 18291
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.671592481807224e-05,
+      "loss": 1.1019,
+      "step": 18292
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.670699584802893e-05,
+      "loss": 0.9883,
+      "step": 18293
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6698066877985626e-05,
+      "loss": 1.0192,
+      "step": 18294
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.668913790794232e-05,
+      "loss": 0.9538,
+      "step": 18295
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.668020893789901e-05,
+      "loss": 0.8984,
+      "step": 18296
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.667127996785571e-05,
+      "loss": 1.0255,
+      "step": 18297
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.666235099781241e-05,
+      "loss": 0.8879,
+      "step": 18298
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.66534220277691e-05,
+      "loss": 1.0767,
+      "step": 18299
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6644493057725796e-05,
+      "loss": 1.0517,
+      "step": 18300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6635564087682486e-05,
+      "loss": 1.0086,
+      "step": 18301
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6626635117639177e-05,
+      "loss": 1.1387,
+      "step": 18302
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6617706147595874e-05,
+      "loss": 1.055,
+      "step": 18303
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.660877717755257e-05,
+      "loss": 0.9085,
+      "step": 18304
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.659984820750926e-05,
+      "loss": 0.9471,
+      "step": 18305
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.659091923746596e-05,
+      "loss": 0.9493,
+      "step": 18306
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6581990267422656e-05,
+      "loss": 1.1978,
+      "step": 18307
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6573061297379346e-05,
+      "loss": 0.9726,
+      "step": 18308
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6564132327336044e-05,
+      "loss": 1.0881,
+      "step": 18309
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.655520335729274e-05,
+      "loss": 0.932,
+      "step": 18310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.654627438724944e-05,
+      "loss": 0.9511,
+      "step": 18311
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.653734541720613e-05,
+      "loss": 1.013,
+      "step": 18312
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.652841644716282e-05,
+      "loss": 1.0936,
+      "step": 18313
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6519487477119516e-05,
+      "loss": 1.0189,
+      "step": 18314
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.651055850707621e-05,
+      "loss": 1.0711,
+      "step": 18315
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6501629537032904e-05,
+      "loss": 1.0531,
+      "step": 18316
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.64927005669896e-05,
+      "loss": 1.0134,
+      "step": 18317
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.648377159694629e-05,
+      "loss": 0.9473,
+      "step": 18318
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.647484262690299e-05,
+      "loss": 1.0285,
+      "step": 18319
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6465913656859686e-05,
+      "loss": 0.9055,
+      "step": 18320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.645698468681638e-05,
+      "loss": 0.9508,
+      "step": 18321
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6448055716773074e-05,
+      "loss": 1.0955,
+      "step": 18322
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.643912674672977e-05,
+      "loss": 1.0599,
+      "step": 18323
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.643019777668646e-05,
+      "loss": 0.9536,
+      "step": 18324
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.642126880664315e-05,
+      "loss": 1.0008,
+      "step": 18325
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.641233983659985e-05,
+      "loss": 1.1934,
+      "step": 18326
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.640341086655654e-05,
+      "loss": 1.0173,
+      "step": 18327
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.639448189651324e-05,
+      "loss": 0.9306,
+      "step": 18328
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6385552926469935e-05,
+      "loss": 0.9634,
+      "step": 18329
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6376623956426625e-05,
+      "loss": 1.0016,
+      "step": 18330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.636769498638332e-05,
+      "loss": 1.1392,
+      "step": 18331
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.635876601634002e-05,
+      "loss": 0.9789,
+      "step": 18332
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.634983704629672e-05,
+      "loss": 1.0083,
+      "step": 18333
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.634090807625341e-05,
+      "loss": 1.0768,
+      "step": 18334
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.63319791062101e-05,
+      "loss": 0.966,
+      "step": 18335
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6323050136166795e-05,
+      "loss": 1.06,
+      "step": 18336
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6314121166123486e-05,
+      "loss": 0.9847,
+      "step": 18337
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.630519219608018e-05,
+      "loss": 1.0941,
+      "step": 18338
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.629626322603688e-05,
+      "loss": 1.0773,
+      "step": 18339
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.628733425599357e-05,
+      "loss": 1.0265,
+      "step": 18340
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.627840528595027e-05,
+      "loss": 1.0826,
+      "step": 18341
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6269476315906965e-05,
+      "loss": 1.1042,
+      "step": 18342
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6260547345863656e-05,
+      "loss": 1.0788,
+      "step": 18343
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.625161837582035e-05,
+      "loss": 1.0367,
+      "step": 18344
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.624268940577705e-05,
+      "loss": 1.1327,
+      "step": 18345
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.623376043573374e-05,
+      "loss": 0.9595,
+      "step": 18346
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.622483146569043e-05,
+      "loss": 1.2164,
+      "step": 18347
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.621590249564713e-05,
+      "loss": 0.9634,
+      "step": 18348
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.620697352560382e-05,
+      "loss": 0.9442,
+      "step": 18349
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6198044555560516e-05,
+      "loss": 1.1271,
+      "step": 18350
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.618911558551721e-05,
+      "loss": 1.0173,
+      "step": 18351
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6180186615473904e-05,
+      "loss": 1.0209,
+      "step": 18352
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.61712576454306e-05,
+      "loss": 1.1001,
+      "step": 18353
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.61623286753873e-05,
+      "loss": 0.9787,
+      "step": 18354
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.615339970534399e-05,
+      "loss": 1.0098,
+      "step": 18355
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6144470735300686e-05,
+      "loss": 0.9773,
+      "step": 18356
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.613554176525738e-05,
+      "loss": 0.982,
+      "step": 18357
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6126612795214074e-05,
+      "loss": 1.1664,
+      "step": 18358
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6117683825170764e-05,
+      "loss": 1.1233,
+      "step": 18359
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.610875485512746e-05,
+      "loss": 1.006,
+      "step": 18360
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.609982588508416e-05,
+      "loss": 1.0179,
+      "step": 18361
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.609089691504085e-05,
+      "loss": 1.1283,
+      "step": 18362
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6081967944997547e-05,
+      "loss": 1.1264,
+      "step": 18363
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6073038974954244e-05,
+      "loss": 1.1338,
+      "step": 18364
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6064110004910934e-05,
+      "loss": 0.9898,
+      "step": 18365
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.605518103486763e-05,
+      "loss": 0.8805,
+      "step": 18366
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.604625206482433e-05,
+      "loss": 1.1302,
+      "step": 18367
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.603732309478102e-05,
+      "loss": 1.1222,
+      "step": 18368
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6028394124737717e-05,
+      "loss": 1.0412,
+      "step": 18369
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.601946515469441e-05,
+      "loss": 0.9084,
+      "step": 18370
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.60105361846511e-05,
+      "loss": 0.941,
+      "step": 18371
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6001607214607795e-05,
+      "loss": 1.0745,
+      "step": 18372
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.599267824456449e-05,
+      "loss": 1.01,
+      "step": 18373
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.598374927452118e-05,
+      "loss": 0.9599,
+      "step": 18374
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.597482030447788e-05,
+      "loss": 0.9511,
+      "step": 18375
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.596589133443458e-05,
+      "loss": 1.0705,
+      "step": 18376
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.595696236439127e-05,
+      "loss": 1.0939,
+      "step": 18377
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5948033394347965e-05,
+      "loss": 1.0302,
+      "step": 18378
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.593910442430466e-05,
+      "loss": 0.9843,
+      "step": 18379
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.593017545426135e-05,
+      "loss": 1.0138,
+      "step": 18380
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.592124648421804e-05,
+      "loss": 0.9895,
+      "step": 18381
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.591231751417474e-05,
+      "loss": 0.8663,
+      "step": 18382
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.590338854413144e-05,
+      "loss": 1.119,
+      "step": 18383
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.589445957408813e-05,
+      "loss": 1.0275,
+      "step": 18384
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5885530604044825e-05,
+      "loss": 1.0306,
+      "step": 18385
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.587660163400152e-05,
+      "loss": 0.9471,
+      "step": 18386
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.586767266395821e-05,
+      "loss": 0.936,
+      "step": 18387
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.585874369391491e-05,
+      "loss": 0.9734,
+      "step": 18388
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.584981472387161e-05,
+      "loss": 1.0473,
+      "step": 18389
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.58408857538283e-05,
+      "loss": 1.1358,
+      "step": 18390
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5831956783784995e-05,
+      "loss": 1.0521,
+      "step": 18391
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5823027813741686e-05,
+      "loss": 1.1806,
+      "step": 18392
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5814098843698376e-05,
+      "loss": 0.9531,
+      "step": 18393
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.580516987365507e-05,
+      "loss": 1.004,
+      "step": 18394
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.579624090361177e-05,
+      "loss": 1.005,
+      "step": 18395
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.578731193356846e-05,
+      "loss": 1.051,
+      "step": 18396
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.577838296352516e-05,
+      "loss": 1.1734,
+      "step": 18397
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5769453993481856e-05,
+      "loss": 0.9601,
+      "step": 18398
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5760525023438546e-05,
+      "loss": 1.1872,
+      "step": 18399
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.575159605339524e-05,
+      "loss": 1.0399,
+      "step": 18400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.574266708335194e-05,
+      "loss": 1.0828,
+      "step": 18401
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.573373811330863e-05,
+      "loss": 1.1524,
+      "step": 18402
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.572480914326533e-05,
+      "loss": 1.0442,
+      "step": 18403
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.571588017322202e-05,
+      "loss": 1.1408,
+      "step": 18404
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.570695120317871e-05,
+      "loss": 1.0321,
+      "step": 18405
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5698022233135407e-05,
+      "loss": 0.9933,
+      "step": 18406
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5689093263092104e-05,
+      "loss": 0.9418,
+      "step": 18407
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.56801642930488e-05,
+      "loss": 1.0231,
+      "step": 18408
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.567123532300549e-05,
+      "loss": 0.9943,
+      "step": 18409
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.566230635296219e-05,
+      "loss": 1.0684,
+      "step": 18410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5653377382918886e-05,
+      "loss": 1.0359,
+      "step": 18411
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5644448412875577e-05,
+      "loss": 1.0423,
+      "step": 18412
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5635519442832274e-05,
+      "loss": 1.0072,
+      "step": 18413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5626590472788964e-05,
+      "loss": 1.0053,
+      "step": 18414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.561766150274566e-05,
+      "loss": 0.9644,
+      "step": 18415
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.560873253270235e-05,
+      "loss": 0.9244,
+      "step": 18416
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.559980356265905e-05,
+      "loss": 1.0801,
+      "step": 18417
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.559087459261574e-05,
+      "loss": 1.0322,
+      "step": 18418
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.558194562257244e-05,
+      "loss": 0.8594,
+      "step": 18419
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5573016652529134e-05,
+      "loss": 1.3644,
+      "step": 18420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5564087682485825e-05,
+      "loss": 0.9749,
+      "step": 18421
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.555515871244252e-05,
+      "loss": 1.0608,
+      "step": 18422
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.554622974239922e-05,
+      "loss": 1.0534,
+      "step": 18423
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.553730077235591e-05,
+      "loss": 1.1232,
+      "step": 18424
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.552837180231261e-05,
+      "loss": 1.0726,
+      "step": 18425
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.55194428322693e-05,
+      "loss": 0.9009,
+      "step": 18426
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5510513862225995e-05,
+      "loss": 1.0104,
+      "step": 18427
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5501584892182685e-05,
+      "loss": 1.0726,
+      "step": 18428
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.549265592213938e-05,
+      "loss": 1.0336,
+      "step": 18429
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.548372695209607e-05,
+      "loss": 1.0308,
+      "step": 18430
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.547479798205277e-05,
+      "loss": 0.9944,
+      "step": 18431
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.546586901200947e-05,
+      "loss": 1.2306,
+      "step": 18432
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5456940041966165e-05,
+      "loss": 0.9966,
+      "step": 18433
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5448011071922855e-05,
+      "loss": 1.1154,
+      "step": 18434
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.543908210187955e-05,
+      "loss": 0.9659,
+      "step": 18435
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.543015313183625e-05,
+      "loss": 1.0262,
+      "step": 18436
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.542122416179294e-05,
+      "loss": 1.0636,
+      "step": 18437
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.541229519174963e-05,
+      "loss": 1.199,
+      "step": 18438
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.540336622170633e-05,
+      "loss": 0.9992,
+      "step": 18439
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.539443725166302e-05,
+      "loss": 1.0428,
+      "step": 18440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5385508281619716e-05,
+      "loss": 1.198,
+      "step": 18441
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.537657931157641e-05,
+      "loss": 1.0414,
+      "step": 18442
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5367650341533103e-05,
+      "loss": 1.1478,
+      "step": 18443
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.53587213714898e-05,
+      "loss": 1.0515,
+      "step": 18444
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.53497924014465e-05,
+      "loss": 1.0254,
+      "step": 18445
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.534086343140319e-05,
+      "loss": 0.9447,
+      "step": 18446
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5331934461359886e-05,
+      "loss": 0.9778,
+      "step": 18447
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.532300549131658e-05,
+      "loss": 1.0026,
+      "step": 18448
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5314076521273273e-05,
+      "loss": 1.0834,
+      "step": 18449
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5305147551229964e-05,
+      "loss": 0.9914,
+      "step": 18450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.529621858118666e-05,
+      "loss": 1.0581,
+      "step": 18451
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.528728961114335e-05,
+      "loss": 0.9791,
+      "step": 18452
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.527836064110005e-05,
+      "loss": 1.0528,
+      "step": 18453
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5269431671056746e-05,
+      "loss": 1.0102,
+      "step": 18454
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5260502701013443e-05,
+      "loss": 1.0458,
+      "step": 18455
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5251573730970134e-05,
+      "loss": 0.9506,
+      "step": 18456
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.524264476092683e-05,
+      "loss": 1.1226,
+      "step": 18457
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.523371579088353e-05,
+      "loss": 1.0545,
+      "step": 18458
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.522478682084022e-05,
+      "loss": 1.013,
+      "step": 18459
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5215857850796916e-05,
+      "loss": 1.2398,
+      "step": 18460
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.520692888075361e-05,
+      "loss": 0.9658,
+      "step": 18461
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.51979999107103e-05,
+      "loss": 1.0632,
+      "step": 18462
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5189070940666994e-05,
+      "loss": 1.0343,
+      "step": 18463
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.518014197062369e-05,
+      "loss": 1.0812,
+      "step": 18464
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.517121300058038e-05,
+      "loss": 1.044,
+      "step": 18465
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.516228403053708e-05,
+      "loss": 1.138,
+      "step": 18466
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5153355060493777e-05,
+      "loss": 1.1165,
+      "step": 18467
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.514442609045047e-05,
+      "loss": 1.0298,
+      "step": 18468
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5135497120407164e-05,
+      "loss": 0.9972,
+      "step": 18469
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.512656815036386e-05,
+      "loss": 1.0735,
+      "step": 18470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.511763918032055e-05,
+      "loss": 1.0121,
+      "step": 18471
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.510871021027724e-05,
+      "loss": 1.0873,
+      "step": 18472
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.509978124023394e-05,
+      "loss": 0.9217,
+      "step": 18473
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.509085227019063e-05,
+      "loss": 1.0151,
+      "step": 18474
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.508192330014733e-05,
+      "loss": 1.0596,
+      "step": 18475
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5072994330104025e-05,
+      "loss": 0.9862,
+      "step": 18476
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5064065360060715e-05,
+      "loss": 0.9534,
+      "step": 18477
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.505513639001741e-05,
+      "loss": 0.9458,
+      "step": 18478
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.504620741997411e-05,
+      "loss": 1.1366,
+      "step": 18479
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.503727844993081e-05,
+      "loss": 1.077,
+      "step": 18480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.50283494798875e-05,
+      "loss": 0.9558,
+      "step": 18481
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5019420509844195e-05,
+      "loss": 1.0193,
+      "step": 18482
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5010491539800885e-05,
+      "loss": 1.0716,
+      "step": 18483
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5001562569757576e-05,
+      "loss": 1.0081,
+      "step": 18484
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.499263359971427e-05,
+      "loss": 0.8499,
+      "step": 18485
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.498370462967097e-05,
+      "loss": 0.9303,
+      "step": 18486
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.497477565962766e-05,
+      "loss": 0.981,
+      "step": 18487
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.496584668958436e-05,
+      "loss": 1.0745,
+      "step": 18488
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4956917719541055e-05,
+      "loss": 0.988,
+      "step": 18489
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4947988749497746e-05,
+      "loss": 1.0336,
+      "step": 18490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.493905977945444e-05,
+      "loss": 1.1706,
+      "step": 18491
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.493013080941114e-05,
+      "loss": 1.0591,
+      "step": 18492
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.492120183936783e-05,
+      "loss": 1.0016,
+      "step": 18493
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.491227286932453e-05,
+      "loss": 0.9719,
+      "step": 18494
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.490334389928122e-05,
+      "loss": 1.0105,
+      "step": 18495
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.489441492923791e-05,
+      "loss": 0.9905,
+      "step": 18496
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4885485959194606e-05,
+      "loss": 1.1212,
+      "step": 18497
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4876556989151303e-05,
+      "loss": 0.9961,
+      "step": 18498
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4867628019107994e-05,
+      "loss": 1.0107,
+      "step": 18499
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.485869904906469e-05,
+      "loss": 1.0331,
+      "step": 18500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.484977007902139e-05,
+      "loss": 1.079,
+      "step": 18501
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.484084110897808e-05,
+      "loss": 1.016,
+      "step": 18502
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4831912138934776e-05,
+      "loss": 1.0813,
+      "step": 18503
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4822983168891473e-05,
+      "loss": 1.04,
+      "step": 18504
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4814054198848164e-05,
+      "loss": 0.9881,
+      "step": 18505
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.480512522880486e-05,
+      "loss": 0.9271,
+      "step": 18506
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.479619625876155e-05,
+      "loss": 1.5138,
+      "step": 18507
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.478726728871825e-05,
+      "loss": 1.1025,
+      "step": 18508
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.477833831867494e-05,
+      "loss": 1.014,
+      "step": 18509
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.476940934863164e-05,
+      "loss": 1.0471,
+      "step": 18510
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4760480378588334e-05,
+      "loss": 0.9785,
+      "step": 18511
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4751551408545024e-05,
+      "loss": 1.028,
+      "step": 18512
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.474262243850172e-05,
+      "loss": 1.1149,
+      "step": 18513
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.473369346845842e-05,
+      "loss": 1.0076,
+      "step": 18514
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.472476449841511e-05,
+      "loss": 0.9278,
+      "step": 18515
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.471583552837181e-05,
+      "loss": 1.0547,
+      "step": 18516
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.47069065583285e-05,
+      "loss": 1.0406,
+      "step": 18517
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.469797758828519e-05,
+      "loss": 1.1828,
+      "step": 18518
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4689048618241885e-05,
+      "loss": 0.9496,
+      "step": 18519
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.468011964819858e-05,
+      "loss": 1.0093,
+      "step": 18520
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.467119067815527e-05,
+      "loss": 1.0462,
+      "step": 18521
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.466226170811197e-05,
+      "loss": 1.0252,
+      "step": 18522
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.465333273806867e-05,
+      "loss": 1.1155,
+      "step": 18523
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.464440376802536e-05,
+      "loss": 1.0081,
+      "step": 18524
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4635474797982055e-05,
+      "loss": 1.0494,
+      "step": 18525
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.462654582793875e-05,
+      "loss": 1.0695,
+      "step": 18526
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.461761685789544e-05,
+      "loss": 0.9844,
+      "step": 18527
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.460868788785214e-05,
+      "loss": 1.0113,
+      "step": 18528
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.459975891780883e-05,
+      "loss": 1.0128,
+      "step": 18529
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.459082994776553e-05,
+      "loss": 1.1297,
+      "step": 18530
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.458190097772222e-05,
+      "loss": 0.9658,
+      "step": 18531
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4572972007678915e-05,
+      "loss": 1.0814,
+      "step": 18532
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.456404303763561e-05,
+      "loss": 1.0546,
+      "step": 18533
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.45551140675923e-05,
+      "loss": 0.8544,
+      "step": 18534
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4546185097549e-05,
+      "loss": 1.0112,
+      "step": 18535
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.45372561275057e-05,
+      "loss": 0.9688,
+      "step": 18536
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.452832715746239e-05,
+      "loss": 1.0054,
+      "step": 18537
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4519398187419085e-05,
+      "loss": 0.9172,
+      "step": 18538
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.451046921737578e-05,
+      "loss": 0.9856,
+      "step": 18539
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.450154024733247e-05,
+      "loss": 1.0039,
+      "step": 18540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4492611277289164e-05,
+      "loss": 1.1182,
+      "step": 18541
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.448368230724586e-05,
+      "loss": 1.4019,
+      "step": 18542
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.447475333720255e-05,
+      "loss": 0.9376,
+      "step": 18543
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.446582436715925e-05,
+      "loss": 0.9915,
+      "step": 18544
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4456895397115946e-05,
+      "loss": 1.0293,
+      "step": 18545
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4447966427072636e-05,
+      "loss": 0.9826,
+      "step": 18546
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4439037457029334e-05,
+      "loss": 1.0886,
+      "step": 18547
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.443010848698603e-05,
+      "loss": 1.003,
+      "step": 18548
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.442117951694272e-05,
+      "loss": 1.0222,
+      "step": 18549
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.441225054689942e-05,
+      "loss": 0.9981,
+      "step": 18550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.440332157685611e-05,
+      "loss": 0.9587,
+      "step": 18551
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4394392606812806e-05,
+      "loss": 0.9868,
+      "step": 18552
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.43854636367695e-05,
+      "loss": 1.0004,
+      "step": 18553
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4376534666726194e-05,
+      "loss": 1.005,
+      "step": 18554
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.436760569668289e-05,
+      "loss": 0.9196,
+      "step": 18555
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.435867672663958e-05,
+      "loss": 1.014,
+      "step": 18556
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.434974775659628e-05,
+      "loss": 1.037,
+      "step": 18557
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4340818786552976e-05,
+      "loss": 0.954,
+      "step": 18558
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.433188981650967e-05,
+      "loss": 1.0564,
+      "step": 18559
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4322960846466364e-05,
+      "loss": 1.1617,
+      "step": 18560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.431403187642306e-05,
+      "loss": 0.9569,
+      "step": 18561
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.430510290637975e-05,
+      "loss": 1.0991,
+      "step": 18562
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.429617393633644e-05,
+      "loss": 0.9471,
+      "step": 18563
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.428724496629314e-05,
+      "loss": 0.9671,
+      "step": 18564
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.427831599624983e-05,
+      "loss": 0.983,
+      "step": 18565
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.426938702620653e-05,
+      "loss": 1.0815,
+      "step": 18566
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4260458056163224e-05,
+      "loss": 1.0822,
+      "step": 18567
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4251529086119915e-05,
+      "loss": 1.0215,
+      "step": 18568
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.424260011607661e-05,
+      "loss": 1.0292,
+      "step": 18569
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.423367114603331e-05,
+      "loss": 0.9297,
+      "step": 18570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.422474217599e-05,
+      "loss": 1.1482,
+      "step": 18571
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.42158132059467e-05,
+      "loss": 1.1245,
+      "step": 18572
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4206884235903394e-05,
+      "loss": 1.0056,
+      "step": 18573
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4197955265860085e-05,
+      "loss": 0.8406,
+      "step": 18574
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4189026295816775e-05,
+      "loss": 1.0619,
+      "step": 18575
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.418009732577347e-05,
+      "loss": 1.0794,
+      "step": 18576
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.417116835573016e-05,
+      "loss": 0.9895,
+      "step": 18577
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.416223938568686e-05,
+      "loss": 1.1048,
+      "step": 18578
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.415331041564356e-05,
+      "loss": 1.0261,
+      "step": 18579
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4144381445600255e-05,
+      "loss": 1.1226,
+      "step": 18580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4135452475556945e-05,
+      "loss": 1.078,
+      "step": 18581
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.412652350551364e-05,
+      "loss": 0.9244,
+      "step": 18582
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.411759453547034e-05,
+      "loss": 1.1035,
+      "step": 18583
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.410866556542703e-05,
+      "loss": 1.0123,
+      "step": 18584
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.409973659538373e-05,
+      "loss": 0.9465,
+      "step": 18585
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.409080762534042e-05,
+      "loss": 0.9483,
+      "step": 18586
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.408187865529711e-05,
+      "loss": 0.9582,
+      "step": 18587
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4072949685253806e-05,
+      "loss": 0.996,
+      "step": 18588
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.40640207152105e-05,
+      "loss": 1.0187,
+      "step": 18589
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4055091745167194e-05,
+      "loss": 1.0108,
+      "step": 18590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.404616277512389e-05,
+      "loss": 0.9165,
+      "step": 18591
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.403723380508059e-05,
+      "loss": 1.2049,
+      "step": 18592
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.402830483503728e-05,
+      "loss": 0.9873,
+      "step": 18593
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4019375864993976e-05,
+      "loss": 1.0249,
+      "step": 18594
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.401044689495067e-05,
+      "loss": 1.0146,
+      "step": 18595
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4001517924907364e-05,
+      "loss": 1.0921,
+      "step": 18596
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.399258895486406e-05,
+      "loss": 0.9516,
+      "step": 18597
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.398365998482075e-05,
+      "loss": 1.1329,
+      "step": 18598
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.397473101477744e-05,
+      "loss": 1.0323,
+      "step": 18599
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.396580204473414e-05,
+      "loss": 1.0489,
+      "step": 18600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3956873074690836e-05,
+      "loss": 1.0374,
+      "step": 18601
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3947944104647534e-05,
+      "loss": 1.0194,
+      "step": 18602
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3939015134604224e-05,
+      "loss": 0.9754,
+      "step": 18603
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.393008616456092e-05,
+      "loss": 1.0219,
+      "step": 18604
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.392115719451762e-05,
+      "loss": 0.9987,
+      "step": 18605
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.391222822447431e-05,
+      "loss": 1.134,
+      "step": 18606
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3903299254431006e-05,
+      "loss": 0.9347,
+      "step": 18607
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.38943702843877e-05,
+      "loss": 1.1853,
+      "step": 18608
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.388544131434439e-05,
+      "loss": 1.0819,
+      "step": 18609
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3876512344301085e-05,
+      "loss": 1.1729,
+      "step": 18610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.386758337425778e-05,
+      "loss": 0.9701,
+      "step": 18611
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.385865440421447e-05,
+      "loss": 1.0942,
+      "step": 18612
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.384972543417117e-05,
+      "loss": 1.0369,
+      "step": 18613
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.384079646412787e-05,
+      "loss": 1.0577,
+      "step": 18614
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.383186749408456e-05,
+      "loss": 0.8804,
+      "step": 18615
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3822938524041254e-05,
+      "loss": 0.728,
+      "step": 18616
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.381400955399795e-05,
+      "loss": 0.9947,
+      "step": 18617
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.380508058395464e-05,
+      "loss": 1.1221,
+      "step": 18618
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.379615161391134e-05,
+      "loss": 0.9941,
+      "step": 18619
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.378722264386803e-05,
+      "loss": 0.9899,
+      "step": 18620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.377829367382472e-05,
+      "loss": 1.13,
+      "step": 18621
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.376936470378142e-05,
+      "loss": 1.0767,
+      "step": 18622
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3760435733738115e-05,
+      "loss": 0.9887,
+      "step": 18623
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3751506763694805e-05,
+      "loss": 1.0639,
+      "step": 18624
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.37425777936515e-05,
+      "loss": 1.0038,
+      "step": 18625
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.37336488236082e-05,
+      "loss": 0.9517,
+      "step": 18626
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.37247198535649e-05,
+      "loss": 1.0479,
+      "step": 18627
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.371579088352159e-05,
+      "loss": 1.1086,
+      "step": 18628
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3706861913478285e-05,
+      "loss": 1.0531,
+      "step": 18629
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.369793294343498e-05,
+      "loss": 1.0615,
+      "step": 18630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.368900397339167e-05,
+      "loss": 0.953,
+      "step": 18631
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.368007500334836e-05,
+      "loss": 0.9836,
+      "step": 18632
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.367114603330506e-05,
+      "loss": 1.1525,
+      "step": 18633
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.366221706326175e-05,
+      "loss": 0.916,
+      "step": 18634
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.365328809321845e-05,
+      "loss": 1.0509,
+      "step": 18635
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3644359123175145e-05,
+      "loss": 1.1153,
+      "step": 18636
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3635430153131836e-05,
+      "loss": 1.1208,
+      "step": 18637
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.362650118308853e-05,
+      "loss": 1.0676,
+      "step": 18638
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.361757221304523e-05,
+      "loss": 1.0685,
+      "step": 18639
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.360864324300192e-05,
+      "loss": 1.0372,
+      "step": 18640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.359971427295862e-05,
+      "loss": 0.9811,
+      "step": 18641
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.359078530291531e-05,
+      "loss": 1.1321,
+      "step": 18642
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3581856332872006e-05,
+      "loss": 0.9959,
+      "step": 18643
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3572927362828696e-05,
+      "loss": 1.0194,
+      "step": 18644
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3563998392785394e-05,
+      "loss": 0.9662,
+      "step": 18645
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3555069422742084e-05,
+      "loss": 1.1819,
+      "step": 18646
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.354614045269878e-05,
+      "loss": 1.0238,
+      "step": 18647
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.353721148265548e-05,
+      "loss": 0.8245,
+      "step": 18648
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.352828251261217e-05,
+      "loss": 0.9618,
+      "step": 18649
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3519353542568866e-05,
+      "loss": 0.9491,
+      "step": 18650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3510424572525564e-05,
+      "loss": 1.0681,
+      "step": 18651
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.350149560248226e-05,
+      "loss": 1.021,
+      "step": 18652
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.349256663243895e-05,
+      "loss": 1.0733,
+      "step": 18653
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.348363766239564e-05,
+      "loss": 1.0573,
+      "step": 18654
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.347470869235234e-05,
+      "loss": 0.9915,
+      "step": 18655
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.346577972230903e-05,
+      "loss": 0.9983,
+      "step": 18656
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.345685075226573e-05,
+      "loss": 1.0708,
+      "step": 18657
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3447921782222424e-05,
+      "loss": 1.1087,
+      "step": 18658
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3438992812179115e-05,
+      "loss": 1.0745,
+      "step": 18659
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.343006384213581e-05,
+      "loss": 1.0376,
+      "step": 18660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.342113487209251e-05,
+      "loss": 1.064,
+      "step": 18661
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.34122059020492e-05,
+      "loss": 1.0235,
+      "step": 18662
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.34032769320059e-05,
+      "loss": 1.1797,
+      "step": 18663
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3394347961962594e-05,
+      "loss": 1.0028,
+      "step": 18664
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3385418991919285e-05,
+      "loss": 1.0206,
+      "step": 18665
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3376490021875975e-05,
+      "loss": 1.011,
+      "step": 18666
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.336756105183267e-05,
+      "loss": 0.9788,
+      "step": 18667
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.335863208178936e-05,
+      "loss": 0.985,
+      "step": 18668
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.334970311174606e-05,
+      "loss": 1.0441,
+      "step": 18669
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.334077414170276e-05,
+      "loss": 0.9544,
+      "step": 18670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.333184517165945e-05,
+      "loss": 1.0884,
+      "step": 18671
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3322916201616145e-05,
+      "loss": 0.9792,
+      "step": 18672
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.331398723157284e-05,
+      "loss": 0.9757,
+      "step": 18673
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.330505826152953e-05,
+      "loss": 1.0324,
+      "step": 18674
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.329612929148623e-05,
+      "loss": 1.0599,
+      "step": 18675
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.328720032144293e-05,
+      "loss": 1.0705,
+      "step": 18676
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.327827135139962e-05,
+      "loss": 1.2278,
+      "step": 18677
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.326934238135631e-05,
+      "loss": 1.1007,
+      "step": 18678
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3260413411313005e-05,
+      "loss": 0.8686,
+      "step": 18679
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.32514844412697e-05,
+      "loss": 0.9433,
+      "step": 18680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.324255547122639e-05,
+      "loss": 1.0179,
+      "step": 18681
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.323362650118309e-05,
+      "loss": 1.0385,
+      "step": 18682
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.322469753113979e-05,
+      "loss": 0.9557,
+      "step": 18683
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.321576856109648e-05,
+      "loss": 1.0527,
+      "step": 18684
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3206839591053175e-05,
+      "loss": 1.0666,
+      "step": 18685
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.319791062100987e-05,
+      "loss": 1.0705,
+      "step": 18686
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.318898165096656e-05,
+      "loss": 0.9486,
+      "step": 18687
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3180052680923254e-05,
+      "loss": 1.1733,
+      "step": 18688
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.317112371087995e-05,
+      "loss": 0.9968,
+      "step": 18689
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.316219474083664e-05,
+      "loss": 1.0591,
+      "step": 18690
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.315326577079334e-05,
+      "loss": 1.0424,
+      "step": 18691
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3144336800750036e-05,
+      "loss": 0.9506,
+      "step": 18692
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3135407830706726e-05,
+      "loss": 0.9704,
+      "step": 18693
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3126478860663424e-05,
+      "loss": 1.1029,
+      "step": 18694
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.311754989062012e-05,
+      "loss": 1.0912,
+      "step": 18695
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.310862092057681e-05,
+      "loss": 1.0682,
+      "step": 18696
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.309969195053351e-05,
+      "loss": 1.0891,
+      "step": 18697
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3090762980490206e-05,
+      "loss": 0.8396,
+      "step": 18698
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3081834010446896e-05,
+      "loss": 0.9836,
+      "step": 18699
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.307290504040359e-05,
+      "loss": 1.3699,
+      "step": 18700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3063976070360284e-05,
+      "loss": 1.0242,
+      "step": 18701
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.305504710031698e-05,
+      "loss": 0.986,
+      "step": 18702
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.304611813027367e-05,
+      "loss": 1.0976,
+      "step": 18703
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.303718916023037e-05,
+      "loss": 0.8906,
+      "step": 18704
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3028260190187066e-05,
+      "loss": 1.0457,
+      "step": 18705
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.301933122014376e-05,
+      "loss": 1.2196,
+      "step": 18706
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3010402250100454e-05,
+      "loss": 0.9963,
+      "step": 18707
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.300147328005715e-05,
+      "loss": 1.044,
+      "step": 18708
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.299254431001384e-05,
+      "loss": 1.0162,
+      "step": 18709
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.298361533997054e-05,
+      "loss": 1.0826,
+      "step": 18710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.297468636992723e-05,
+      "loss": 1.1596,
+      "step": 18711
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.296575739988392e-05,
+      "loss": 1.0051,
+      "step": 18712
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.295682842984062e-05,
+      "loss": 0.9596,
+      "step": 18713
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2947899459797315e-05,
+      "loss": 1.0516,
+      "step": 18714
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2938970489754005e-05,
+      "loss": 0.983,
+      "step": 18715
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.29300415197107e-05,
+      "loss": 0.9649,
+      "step": 18716
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.29211125496674e-05,
+      "loss": 1.0288,
+      "step": 18717
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.291218357962409e-05,
+      "loss": 0.9765,
+      "step": 18718
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.290325460958079e-05,
+      "loss": 1.1764,
+      "step": 18719
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2894325639537485e-05,
+      "loss": 1.0448,
+      "step": 18720
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2885396669494175e-05,
+      "loss": 1.0129,
+      "step": 18721
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.287646769945087e-05,
+      "loss": 0.9083,
+      "step": 18722
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.286753872940756e-05,
+      "loss": 1.1238,
+      "step": 18723
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.285860975936426e-05,
+      "loss": 1.0698,
+      "step": 18724
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.284968078932095e-05,
+      "loss": 1.1515,
+      "step": 18725
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.284075181927765e-05,
+      "loss": 1.0443,
+      "step": 18726
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2831822849234345e-05,
+      "loss": 1.0237,
+      "step": 18727
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2822893879191036e-05,
+      "loss": 1.0192,
+      "step": 18728
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.281396490914773e-05,
+      "loss": 1.1201,
+      "step": 18729
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.280503593910443e-05,
+      "loss": 1.0721,
+      "step": 18730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.279610696906112e-05,
+      "loss": 0.8956,
+      "step": 18731
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.278717799901782e-05,
+      "loss": 1.0391,
+      "step": 18732
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.277824902897451e-05,
+      "loss": 1.1512,
+      "step": 18733
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2769320058931206e-05,
+      "loss": 0.9679,
+      "step": 18734
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2760391088887896e-05,
+      "loss": 0.9559,
+      "step": 18735
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.275146211884459e-05,
+      "loss": 0.9576,
+      "step": 18736
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2742533148801284e-05,
+      "loss": 0.9287,
+      "step": 18737
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.273360417875798e-05,
+      "loss": 1.0837,
+      "step": 18738
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.272467520871468e-05,
+      "loss": 0.9203,
+      "step": 18739
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.271574623867137e-05,
+      "loss": 1.1167,
+      "step": 18740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2706817268628066e-05,
+      "loss": 0.9658,
+      "step": 18741
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.269788829858476e-05,
+      "loss": 1.1083,
+      "step": 18742
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2688959328541454e-05,
+      "loss": 0.9881,
+      "step": 18743
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.268003035849815e-05,
+      "loss": 1.0455,
+      "step": 18744
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.267110138845484e-05,
+      "loss": 0.9474,
+      "step": 18745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.266217241841153e-05,
+      "loss": 1.0444,
+      "step": 18746
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.265324344836823e-05,
+      "loss": 1.0137,
+      "step": 18747
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2644314478324926e-05,
+      "loss": 0.9868,
+      "step": 18748
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2635385508281624e-05,
+      "loss": 0.859,
+      "step": 18749
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2626456538238314e-05,
+      "loss": 1.1098,
+      "step": 18750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.261752756819501e-05,
+      "loss": 1.0654,
+      "step": 18751
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.260859859815171e-05,
+      "loss": 1.0113,
+      "step": 18752
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.25996696281084e-05,
+      "loss": 1.0919,
+      "step": 18753
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2590740658065096e-05,
+      "loss": 1.0309,
+      "step": 18754
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2581811688021794e-05,
+      "loss": 1.1011,
+      "step": 18755
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2572882717978484e-05,
+      "loss": 1.1723,
+      "step": 18756
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2563953747935175e-05,
+      "loss": 1.0332,
+      "step": 18757
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.255502477789187e-05,
+      "loss": 0.9517,
+      "step": 18758
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.254609580784856e-05,
+      "loss": 1.1234,
+      "step": 18759
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.253716683780526e-05,
+      "loss": 0.9469,
+      "step": 18760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.252823786776196e-05,
+      "loss": 0.9303,
+      "step": 18761
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.251930889771865e-05,
+      "loss": 1.1322,
+      "step": 18762
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2510379927675345e-05,
+      "loss": 0.9023,
+      "step": 18763
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.250145095763204e-05,
+      "loss": 1.2454,
+      "step": 18764
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.249252198758873e-05,
+      "loss": 1.064,
+      "step": 18765
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.248359301754543e-05,
+      "loss": 0.9603,
+      "step": 18766
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.247466404750213e-05,
+      "loss": 0.9815,
+      "step": 18767
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.246573507745882e-05,
+      "loss": 1.0433,
+      "step": 18768
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.245680610741551e-05,
+      "loss": 1.1355,
+      "step": 18769
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2447877137372205e-05,
+      "loss": 1.0295,
+      "step": 18770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2438948167328896e-05,
+      "loss": 1.0159,
+      "step": 18771
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.243001919728559e-05,
+      "loss": 1.1469,
+      "step": 18772
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.242109022724229e-05,
+      "loss": 1.0279,
+      "step": 18773
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.241216125719899e-05,
+      "loss": 1.0695,
+      "step": 18774
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.240323228715568e-05,
+      "loss": 1.2829,
+      "step": 18775
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2394303317112375e-05,
+      "loss": 1.0621,
+      "step": 18776
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.238537434706907e-05,
+      "loss": 0.8548,
+      "step": 18777
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.237644537702576e-05,
+      "loss": 1.1597,
+      "step": 18778
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.236751640698245e-05,
+      "loss": 1.1907,
+      "step": 18779
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.235858743693915e-05,
+      "loss": 0.9151,
+      "step": 18780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.234965846689584e-05,
+      "loss": 1.1105,
+      "step": 18781
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.234072949685254e-05,
+      "loss": 1.0884,
+      "step": 18782
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2331800526809236e-05,
+      "loss": 1.0271,
+      "step": 18783
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2322871556765926e-05,
+      "loss": 1.0374,
+      "step": 18784
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.231394258672262e-05,
+      "loss": 1.0105,
+      "step": 18785
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.230501361667932e-05,
+      "loss": 1.0764,
+      "step": 18786
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.229608464663601e-05,
+      "loss": 0.9946,
+      "step": 18787
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.228715567659271e-05,
+      "loss": 0.9618,
+      "step": 18788
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2278226706549406e-05,
+      "loss": 1.0331,
+      "step": 18789
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2269297736506096e-05,
+      "loss": 1.0108,
+      "step": 18790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2260368766462787e-05,
+      "loss": 1.0783,
+      "step": 18791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2251439796419484e-05,
+      "loss": 0.9453,
+      "step": 18792
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2242510826376174e-05,
+      "loss": 1.0147,
+      "step": 18793
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.223358185633287e-05,
+      "loss": 1.1359,
+      "step": 18794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.222465288628957e-05,
+      "loss": 1.0517,
+      "step": 18795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.221572391624626e-05,
+      "loss": 1.0524,
+      "step": 18796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2206794946202956e-05,
+      "loss": 1.0634,
+      "step": 18797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2197865976159654e-05,
+      "loss": 0.9916,
+      "step": 18798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.218893700611635e-05,
+      "loss": 1.0648,
+      "step": 18799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.218000803607304e-05,
+      "loss": 0.9764,
+      "step": 18800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.217107906602974e-05,
+      "loss": 1.0119,
+      "step": 18801
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.216215009598643e-05,
+      "loss": 0.9908,
+      "step": 18802
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.215322112594312e-05,
+      "loss": 0.9265,
+      "step": 18803
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.214429215589982e-05,
+      "loss": 1.0459,
+      "step": 18804
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2135363185856514e-05,
+      "loss": 0.8987,
+      "step": 18805
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2126434215813205e-05,
+      "loss": 1.0486,
+      "step": 18806
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.21175052457699e-05,
+      "loss": 1.0929,
+      "step": 18807
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.21085762757266e-05,
+      "loss": 1.0566,
+      "step": 18808
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.209964730568329e-05,
+      "loss": 1.0516,
+      "step": 18809
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.209071833563999e-05,
+      "loss": 1.014,
+      "step": 18810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2081789365596684e-05,
+      "loss": 1.1229,
+      "step": 18811
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2072860395553375e-05,
+      "loss": 0.9478,
+      "step": 18812
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.206393142551007e-05,
+      "loss": 1.0798,
+      "step": 18813
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.205500245546676e-05,
+      "loss": 1.1615,
+      "step": 18814
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.204607348542345e-05,
+      "loss": 1.0161,
+      "step": 18815
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.203714451538015e-05,
+      "loss": 0.9694,
+      "step": 18816
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.202821554533685e-05,
+      "loss": 0.959,
+      "step": 18817
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.201928657529354e-05,
+      "loss": 1.0442,
+      "step": 18818
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2010357605250235e-05,
+      "loss": 1.0164,
+      "step": 18819
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.200142863520693e-05,
+      "loss": 0.9683,
+      "step": 18820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.199249966516362e-05,
+      "loss": 1.0159,
+      "step": 18821
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.198357069512032e-05,
+      "loss": 1.0799,
+      "step": 18822
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.197464172507702e-05,
+      "loss": 1.036,
+      "step": 18823
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.196571275503371e-05,
+      "loss": 0.9556,
+      "step": 18824
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.19567837849904e-05,
+      "loss": 1.0601,
+      "step": 18825
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1947854814947096e-05,
+      "loss": 1.0227,
+      "step": 18826
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.193892584490379e-05,
+      "loss": 1.1888,
+      "step": 18827
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.192999687486048e-05,
+      "loss": 1.1403,
+      "step": 18828
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.192106790481718e-05,
+      "loss": 1.054,
+      "step": 18829
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.191213893477388e-05,
+      "loss": 1.0247,
+      "step": 18830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.190320996473057e-05,
+      "loss": 0.9706,
+      "step": 18831
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1894280994687266e-05,
+      "loss": 1.1031,
+      "step": 18832
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.188535202464396e-05,
+      "loss": 1.0718,
+      "step": 18833
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.187642305460065e-05,
+      "loss": 0.9646,
+      "step": 18834
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.186749408455735e-05,
+      "loss": 0.9453,
+      "step": 18835
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.185856511451404e-05,
+      "loss": 1.0302,
+      "step": 18836
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.184963614447073e-05,
+      "loss": 1.0144,
+      "step": 18837
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.184070717442743e-05,
+      "loss": 1.115,
+      "step": 18838
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1831778204384126e-05,
+      "loss": 1.0013,
+      "step": 18839
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1822849234340817e-05,
+      "loss": 1.1291,
+      "step": 18840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1813920264297514e-05,
+      "loss": 0.9786,
+      "step": 18841
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.180499129425421e-05,
+      "loss": 1.0446,
+      "step": 18842
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.17960623242109e-05,
+      "loss": 0.941,
+      "step": 18843
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.17871333541676e-05,
+      "loss": 1.0216,
+      "step": 18844
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1778204384124296e-05,
+      "loss": 0.9361,
+      "step": 18845
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1769275414080987e-05,
+      "loss": 0.9753,
+      "step": 18846
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1760346444037684e-05,
+      "loss": 0.9807,
+      "step": 18847
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1751417473994374e-05,
+      "loss": 1.2131,
+      "step": 18848
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.174248850395107e-05,
+      "loss": 1.0618,
+      "step": 18849
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.173355953390776e-05,
+      "loss": 1.1364,
+      "step": 18850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.172463056386446e-05,
+      "loss": 1.109,
+      "step": 18851
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1715701593821157e-05,
+      "loss": 0.9733,
+      "step": 18852
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.170677262377785e-05,
+      "loss": 1.0003,
+      "step": 18853
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1697843653734544e-05,
+      "loss": 0.8845,
+      "step": 18854
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.168891468369124e-05,
+      "loss": 1.0197,
+      "step": 18855
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.167998571364793e-05,
+      "loss": 0.9866,
+      "step": 18856
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.167105674360463e-05,
+      "loss": 1.0191,
+      "step": 18857
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.166212777356132e-05,
+      "loss": 1.0123,
+      "step": 18858
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.165319880351802e-05,
+      "loss": 0.9772,
+      "step": 18859
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.164426983347471e-05,
+      "loss": 1.0683,
+      "step": 18860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1635340863431405e-05,
+      "loss": 0.9215,
+      "step": 18861
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1626411893388095e-05,
+      "loss": 1.2414,
+      "step": 18862
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.161748292334479e-05,
+      "loss": 1.052,
+      "step": 18863
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.160855395330149e-05,
+      "loss": 1.0736,
+      "step": 18864
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.159962498325818e-05,
+      "loss": 0.9934,
+      "step": 18865
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.159069601321488e-05,
+      "loss": 1.0024,
+      "step": 18866
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1581767043171575e-05,
+      "loss": 1.0026,
+      "step": 18867
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1572838073128265e-05,
+      "loss": 1.0373,
+      "step": 18868
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.156390910308496e-05,
+      "loss": 1.0463,
+      "step": 18869
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.155498013304165e-05,
+      "loss": 1.0016,
+      "step": 18870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.154605116299835e-05,
+      "loss": 1.0209,
+      "step": 18871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.153712219295504e-05,
+      "loss": 1.0702,
+      "step": 18872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.152819322291174e-05,
+      "loss": 1.0278,
+      "step": 18873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1519264252868435e-05,
+      "loss": 1.0667,
+      "step": 18874
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1510335282825126e-05,
+      "loss": 1.1826,
+      "step": 18875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.150140631278182e-05,
+      "loss": 1.0138,
+      "step": 18876
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.149247734273852e-05,
+      "loss": 1.201,
+      "step": 18877
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.148354837269521e-05,
+      "loss": 1.0177,
+      "step": 18878
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.147461940265191e-05,
+      "loss": 1.2063,
+      "step": 18879
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1465690432608605e-05,
+      "loss": 1.0199,
+      "step": 18880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1456761462565296e-05,
+      "loss": 0.9897,
+      "step": 18881
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1447832492521986e-05,
+      "loss": 0.9785,
+      "step": 18882
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1438903522478683e-05,
+      "loss": 0.9545,
+      "step": 18883
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1429974552435374e-05,
+      "loss": 1.0319,
+      "step": 18884
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.142104558239207e-05,
+      "loss": 1.0769,
+      "step": 18885
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.141211661234877e-05,
+      "loss": 1.1798,
+      "step": 18886
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.140318764230546e-05,
+      "loss": 0.9727,
+      "step": 18887
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1394258672262156e-05,
+      "loss": 1.0222,
+      "step": 18888
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.138532970221885e-05,
+      "loss": 1.0817,
+      "step": 18889
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1376400732175544e-05,
+      "loss": 0.9815,
+      "step": 18890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.136747176213224e-05,
+      "loss": 0.8982,
+      "step": 18891
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.135854279208894e-05,
+      "loss": 1.1665,
+      "step": 18892
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.134961382204563e-05,
+      "loss": 1.0225,
+      "step": 18893
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.134068485200232e-05,
+      "loss": 1.052,
+      "step": 18894
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1331755881959017e-05,
+      "loss": 1.0246,
+      "step": 18895
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1322826911915714e-05,
+      "loss": 1.0639,
+      "step": 18896
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1313897941872404e-05,
+      "loss": 0.9691,
+      "step": 18897
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.13049689718291e-05,
+      "loss": 1.0729,
+      "step": 18898
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.12960400017858e-05,
+      "loss": 1.0291,
+      "step": 18899
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.128711103174249e-05,
+      "loss": 1.0186,
+      "step": 18900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1278182061699187e-05,
+      "loss": 0.9682,
+      "step": 18901
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1269253091655884e-05,
+      "loss": 0.9842,
+      "step": 18902
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1260324121612574e-05,
+      "loss": 1.0285,
+      "step": 18903
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.125139515156927e-05,
+      "loss": 0.9443,
+      "step": 18904
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.124246618152596e-05,
+      "loss": 0.9647,
+      "step": 18905
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.123353721148265e-05,
+      "loss": 1.1154,
+      "step": 18906
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.122460824143935e-05,
+      "loss": 1.0642,
+      "step": 18907
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.121567927139605e-05,
+      "loss": 1.1077,
+      "step": 18908
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.120675030135274e-05,
+      "loss": 0.9712,
+      "step": 18909
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1197821331309435e-05,
+      "loss": 0.9702,
+      "step": 18910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.118889236126613e-05,
+      "loss": 0.994,
+      "step": 18911
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.117996339122282e-05,
+      "loss": 0.9858,
+      "step": 18912
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.117103442117952e-05,
+      "loss": 1.0,
+      "step": 18913
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.116210545113622e-05,
+      "loss": 1.1281,
+      "step": 18914
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.115317648109291e-05,
+      "loss": 0.9868,
+      "step": 18915
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.11442475110496e-05,
+      "loss": 1.0202,
+      "step": 18916
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1135318541006295e-05,
+      "loss": 1.1542,
+      "step": 18917
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1126389570962986e-05,
+      "loss": 1.0335,
+      "step": 18918
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.111746060091968e-05,
+      "loss": 1.0842,
+      "step": 18919
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.110853163087638e-05,
+      "loss": 0.956,
+      "step": 18920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.109960266083308e-05,
+      "loss": 1.1123,
+      "step": 18921
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.109067369078977e-05,
+      "loss": 0.9295,
+      "step": 18922
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1081744720746465e-05,
+      "loss": 1.0145,
+      "step": 18923
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.107281575070316e-05,
+      "loss": 1.046,
+      "step": 18924
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.106388678065985e-05,
+      "loss": 1.0157,
+      "step": 18925
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.105495781061655e-05,
+      "loss": 1.0093,
+      "step": 18926
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.104602884057324e-05,
+      "loss": 1.0249,
+      "step": 18927
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.103709987052993e-05,
+      "loss": 1.1534,
+      "step": 18928
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.102817090048663e-05,
+      "loss": 1.0481,
+      "step": 18929
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1019241930443326e-05,
+      "loss": 0.9748,
+      "step": 18930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1010312960400016e-05,
+      "loss": 1.0358,
+      "step": 18931
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1001383990356713e-05,
+      "loss": 1.0704,
+      "step": 18932
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.099245502031341e-05,
+      "loss": 0.9775,
+      "step": 18933
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.09835260502701e-05,
+      "loss": 0.9924,
+      "step": 18934
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.09745970802268e-05,
+      "loss": 0.9733,
+      "step": 18935
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0965668110183496e-05,
+      "loss": 1.2114,
+      "step": 18936
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0956739140140186e-05,
+      "loss": 1.0117,
+      "step": 18937
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0947810170096883e-05,
+      "loss": 1.0731,
+      "step": 18938
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0938881200053574e-05,
+      "loss": 1.1772,
+      "step": 18939
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0929952230010264e-05,
+      "loss": 0.9991,
+      "step": 18940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.092102325996696e-05,
+      "loss": 0.8924,
+      "step": 18941
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.091209428992366e-05,
+      "loss": 1.0488,
+      "step": 18942
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.090316531988035e-05,
+      "loss": 0.9424,
+      "step": 18943
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.089423634983705e-05,
+      "loss": 0.9915,
+      "step": 18944
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0885307379793744e-05,
+      "loss": 1.1142,
+      "step": 18945
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.087637840975044e-05,
+      "loss": 0.9777,
+      "step": 18946
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.086744943970713e-05,
+      "loss": 1.042,
+      "step": 18947
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.085852046966383e-05,
+      "loss": 0.9547,
+      "step": 18948
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.084959149962052e-05,
+      "loss": 0.9798,
+      "step": 18949
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.084066252957722e-05,
+      "loss": 0.9911,
+      "step": 18950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.083173355953391e-05,
+      "loss": 1.1456,
+      "step": 18951
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0822804589490604e-05,
+      "loss": 1.0403,
+      "step": 18952
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0813875619447295e-05,
+      "loss": 1.1097,
+      "step": 18953
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.080494664940399e-05,
+      "loss": 1.0036,
+      "step": 18954
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.079601767936069e-05,
+      "loss": 1.0607,
+      "step": 18955
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.078708870931738e-05,
+      "loss": 0.9576,
+      "step": 18956
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.077815973927408e-05,
+      "loss": 1.0634,
+      "step": 18957
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 1.0039,
+      "step": 18958
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0760301799187465e-05,
+      "loss": 1.071,
+      "step": 18959
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.075137282914416e-05,
+      "loss": 0.8608,
+      "step": 18960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.074244385910085e-05,
+      "loss": 0.9612,
+      "step": 18961
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.073351488905754e-05,
+      "loss": 1.1637,
+      "step": 18962
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.072458591901424e-05,
+      "loss": 1.0662,
+      "step": 18963
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.071565694897094e-05,
+      "loss": 1.0148,
+      "step": 18964
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.070672797892763e-05,
+      "loss": 1.1059,
+      "step": 18965
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0697799008884325e-05,
+      "loss": 1.0002,
+      "step": 18966
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.068887003884102e-05,
+      "loss": 0.9407,
+      "step": 18967
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.067994106879771e-05,
+      "loss": 1.0764,
+      "step": 18968
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.067101209875441e-05,
+      "loss": 0.9769,
+      "step": 18969
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.066208312871111e-05,
+      "loss": 0.8531,
+      "step": 18970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0653154158667805e-05,
+      "loss": 0.8857,
+      "step": 18971
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0644225188624495e-05,
+      "loss": 1.0569,
+      "step": 18972
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0635296218581186e-05,
+      "loss": 1.1043,
+      "step": 18973
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.062636724853788e-05,
+      "loss": 0.9162,
+      "step": 18974
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0617438278494574e-05,
+      "loss": 0.9789,
+      "step": 18975
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.060850930845127e-05,
+      "loss": 1.0897,
+      "step": 18976
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.059958033840797e-05,
+      "loss": 1.1263,
+      "step": 18977
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.059065136836466e-05,
+      "loss": 1.0969,
+      "step": 18978
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0581722398321356e-05,
+      "loss": 1.1619,
+      "step": 18979
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.057279342827805e-05,
+      "loss": 1.131,
+      "step": 18980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0563864458234743e-05,
+      "loss": 0.9728,
+      "step": 18981
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.055493548819144e-05,
+      "loss": 1.084,
+      "step": 18982
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.054600651814814e-05,
+      "loss": 1.0864,
+      "step": 18983
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.053707754810483e-05,
+      "loss": 1.1078,
+      "step": 18984
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.052814857806152e-05,
+      "loss": 1.0147,
+      "step": 18985
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0519219608018216e-05,
+      "loss": 1.0009,
+      "step": 18986
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.051029063797491e-05,
+      "loss": 1.0956,
+      "step": 18987
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0501361667931604e-05,
+      "loss": 1.1786,
+      "step": 18988
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.04924326978883e-05,
+      "loss": 1.0353,
+      "step": 18989
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0483503727844992e-05,
+      "loss": 1.0681,
+      "step": 18990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.047457475780169e-05,
+      "loss": 1.0532,
+      "step": 18991
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0465645787758383e-05,
+      "loss": 1.0215,
+      "step": 18992
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.045671681771508e-05,
+      "loss": 0.9347,
+      "step": 18993
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.044778784767177e-05,
+      "loss": 0.9333,
+      "step": 18994
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0438858877628468e-05,
+      "loss": 1.0277,
+      "step": 18995
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0429929907585165e-05,
+      "loss": 1.0135,
+      "step": 18996
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0421000937541856e-05,
+      "loss": 1.0498,
+      "step": 18997
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.041207196749855e-05,
+      "loss": 1.1288,
+      "step": 18998
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0403142997455247e-05,
+      "loss": 1.0147,
+      "step": 18999
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0394214027411937e-05,
+      "loss": 0.9568,
+      "step": 19000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0385285057368634e-05,
+      "loss": 0.9709,
+      "step": 19001
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.037635608732533e-05,
+      "loss": 1.0431,
+      "step": 19002
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0367427117282022e-05,
+      "loss": 1.053,
+      "step": 19003
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0358498147238716e-05,
+      "loss": 0.9818,
+      "step": 19004
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0349569177195413e-05,
+      "loss": 1.1492,
+      "step": 19005
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0340640207152104e-05,
+      "loss": 1.0627,
+      "step": 19006
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.03317112371088e-05,
+      "loss": 1.0877,
+      "step": 19007
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0322782267065498e-05,
+      "loss": 1.0286,
+      "step": 19008
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.031385329702219e-05,
+      "loss": 1.0009,
+      "step": 19009
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0304924326978883e-05,
+      "loss": 0.8961,
+      "step": 19010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.029599535693558e-05,
+      "loss": 0.9776,
+      "step": 19011
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.028706638689227e-05,
+      "loss": 0.8964,
+      "step": 19012
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0278137416848968e-05,
+      "loss": 1.0269,
+      "step": 19013
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0269208446805665e-05,
+      "loss": 0.9809,
+      "step": 19014
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0260279476762355e-05,
+      "loss": 1.0761,
+      "step": 19015
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.025135050671905e-05,
+      "loss": 0.9551,
+      "step": 19016
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0242421536675746e-05,
+      "loss": 1.0955,
+      "step": 19017
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0233492566632444e-05,
+      "loss": 0.9749,
+      "step": 19018
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0224563596589134e-05,
+      "loss": 1.0666,
+      "step": 19019
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.021563462654583e-05,
+      "loss": 0.9981,
+      "step": 19020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0206705656502525e-05,
+      "loss": 1.0003,
+      "step": 19021
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0197776686459216e-05,
+      "loss": 1.039,
+      "step": 19022
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0188847716415913e-05,
+      "loss": 1.1025,
+      "step": 19023
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.017991874637261e-05,
+      "loss": 1.0825,
+      "step": 19024
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.01709897763293e-05,
+      "loss": 0.938,
+      "step": 19025
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0162060806285998e-05,
+      "loss": 1.1269,
+      "step": 19026
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0153131836242692e-05,
+      "loss": 1.1052,
+      "step": 19027
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0144202866199382e-05,
+      "loss": 1.065,
+      "step": 19028
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.013527389615608e-05,
+      "loss": 0.9535,
+      "step": 19029
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0126344926112777e-05,
+      "loss": 1.021,
+      "step": 19030
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0117415956069467e-05,
+      "loss": 0.998,
+      "step": 19031
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0108486986026165e-05,
+      "loss": 0.9993,
+      "step": 19032
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.009955801598286e-05,
+      "loss": 0.8988,
+      "step": 19033
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.009062904593955e-05,
+      "loss": 1.1086,
+      "step": 19034
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0081700075896246e-05,
+      "loss": 1.0561,
+      "step": 19035
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0072771105852944e-05,
+      "loss": 1.0055,
+      "step": 19036
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0063842135809634e-05,
+      "loss": 1.0327,
+      "step": 19037
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.005491316576633e-05,
+      "loss": 1.0424,
+      "step": 19038
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0045984195723025e-05,
+      "loss": 0.9652,
+      "step": 19039
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0037055225679716e-05,
+      "loss": 1.028,
+      "step": 19040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0028126255636413e-05,
+      "loss": 0.9932,
+      "step": 19041
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.001919728559311e-05,
+      "loss": 1.0839,
+      "step": 19042
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0010268315549804e-05,
+      "loss": 0.9698,
+      "step": 19043
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0001339345506494e-05,
+      "loss": 0.9975,
+      "step": 19044
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9992410375463192e-05,
+      "loss": 1.1354,
+      "step": 19045
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.998348140541989e-05,
+      "loss": 1.0314,
+      "step": 19046
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.997455243537658e-05,
+      "loss": 1.126,
+      "step": 19047
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9965623465333277e-05,
+      "loss": 1.1866,
+      "step": 19048
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.995669449528997e-05,
+      "loss": 1.0693,
+      "step": 19049
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.994776552524666e-05,
+      "loss": 1.0433,
+      "step": 19050
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.993883655520336e-05,
+      "loss": 1.0284,
+      "step": 19051
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9929907585160056e-05,
+      "loss": 1.0236,
+      "step": 19052
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9920978615116746e-05,
+      "loss": 0.9648,
+      "step": 19053
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9912049645073443e-05,
+      "loss": 1.0359,
+      "step": 19054
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9903120675030137e-05,
+      "loss": 0.9368,
+      "step": 19055
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9894191704986828e-05,
+      "loss": 1.0715,
+      "step": 19056
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9885262734943525e-05,
+      "loss": 1.254,
+      "step": 19057
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9876333764900222e-05,
+      "loss": 1.0159,
+      "step": 19058
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9867404794856913e-05,
+      "loss": 0.9951,
+      "step": 19059
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.985847582481361e-05,
+      "loss": 0.9281,
+      "step": 19060
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9849546854770304e-05,
+      "loss": 1.0455,
+      "step": 19061
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9840617884726994e-05,
+      "loss": 1.036,
+      "step": 19062
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.983168891468369e-05,
+      "loss": 1.0325,
+      "step": 19063
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.982275994464039e-05,
+      "loss": 1.1797,
+      "step": 19064
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.981383097459708e-05,
+      "loss": 0.8588,
+      "step": 19065
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9804902004553777e-05,
+      "loss": 0.9914,
+      "step": 19066
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.979597303451047e-05,
+      "loss": 0.9958,
+      "step": 19067
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9787044064467168e-05,
+      "loss": 1.0547,
+      "step": 19068
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9778115094423858e-05,
+      "loss": 0.9965,
+      "step": 19069
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9769186124380555e-05,
+      "loss": 0.9554,
+      "step": 19070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9760257154337253e-05,
+      "loss": 1.1189,
+      "step": 19071
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9751328184293943e-05,
+      "loss": 1.0797,
+      "step": 19072
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9742399214250637e-05,
+      "loss": 1.0288,
+      "step": 19073
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9733470244207334e-05,
+      "loss": 1.0249,
+      "step": 19074
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9724541274164025e-05,
+      "loss": 0.9641,
+      "step": 19075
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9715612304120722e-05,
+      "loss": 1.0019,
+      "step": 19076
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.970668333407742e-05,
+      "loss": 1.0335,
+      "step": 19077
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.969775436403411e-05,
+      "loss": 0.9398,
+      "step": 19078
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9688825393990804e-05,
+      "loss": 1.1395,
+      "step": 19079
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.96798964239475e-05,
+      "loss": 0.9405,
+      "step": 19080
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.967096745390419e-05,
+      "loss": 1.1592,
+      "step": 19081
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.966203848386089e-05,
+      "loss": 1.1167,
+      "step": 19082
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9653109513817582e-05,
+      "loss": 1.0492,
+      "step": 19083
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9644180543774276e-05,
+      "loss": 1.0539,
+      "step": 19084
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.963525157373097e-05,
+      "loss": 1.1136,
+      "step": 19085
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9626322603687667e-05,
+      "loss": 1.0504,
+      "step": 19086
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9617393633644358e-05,
+      "loss": 1.05,
+      "step": 19087
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9608464663601055e-05,
+      "loss": 0.9783,
+      "step": 19088
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.959953569355775e-05,
+      "loss": 0.8788,
+      "step": 19089
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9590606723514443e-05,
+      "loss": 1.0628,
+      "step": 19090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9581677753471137e-05,
+      "loss": 0.9712,
+      "step": 19091
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9572748783427834e-05,
+      "loss": 1.0006,
+      "step": 19092
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.956381981338453e-05,
+      "loss": 0.8856,
+      "step": 19093
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9554890843341222e-05,
+      "loss": 0.9554,
+      "step": 19094
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9545961873297916e-05,
+      "loss": 1.1794,
+      "step": 19095
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9537032903254613e-05,
+      "loss": 1.0816,
+      "step": 19096
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9528103933211303e-05,
+      "loss": 0.9713,
+      "step": 19097
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9519174963168e-05,
+      "loss": 1.1113,
+      "step": 19098
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9510245993124698e-05,
+      "loss": 1.2631,
+      "step": 19099
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.950131702308139e-05,
+      "loss": 0.8337,
+      "step": 19100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9492388053038082e-05,
+      "loss": 0.9881,
+      "step": 19101
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.948345908299478e-05,
+      "loss": 1.0842,
+      "step": 19102
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.947453011295147e-05,
+      "loss": 0.9066,
+      "step": 19103
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9465601142908167e-05,
+      "loss": 1.0543,
+      "step": 19104
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9456672172864865e-05,
+      "loss": 1.0295,
+      "step": 19105
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9447743202821555e-05,
+      "loss": 1.0545,
+      "step": 19106
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.943881423277825e-05,
+      "loss": 1.0257,
+      "step": 19107
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9429885262734946e-05,
+      "loss": 0.9451,
+      "step": 19108
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9420956292691637e-05,
+      "loss": 1.0214,
+      "step": 19109
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9412027322648334e-05,
+      "loss": 1.0294,
+      "step": 19110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.940309835260503e-05,
+      "loss": 1.0853,
+      "step": 19111
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.939416938256172e-05,
+      "loss": 0.9113,
+      "step": 19112
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9385240412518415e-05,
+      "loss": 1.1939,
+      "step": 19113
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9376311442475113e-05,
+      "loss": 1.0203,
+      "step": 19114
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9367382472431803e-05,
+      "loss": 1.0971,
+      "step": 19115
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.93584535023885e-05,
+      "loss": 0.933,
+      "step": 19116
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9349524532345198e-05,
+      "loss": 0.9673,
+      "step": 19117
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.934059556230189e-05,
+      "loss": 0.9809,
+      "step": 19118
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9331666592258582e-05,
+      "loss": 1.0132,
+      "step": 19119
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.932273762221528e-05,
+      "loss": 1.0325,
+      "step": 19120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9313808652171977e-05,
+      "loss": 1.0288,
+      "step": 19121
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9304879682128667e-05,
+      "loss": 1.0004,
+      "step": 19122
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9295950712085364e-05,
+      "loss": 1.0538,
+      "step": 19123
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9287021742042058e-05,
+      "loss": 1.0277,
+      "step": 19124
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.927809277199875e-05,
+      "loss": 1.0742,
+      "step": 19125
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9269163801955446e-05,
+      "loss": 1.1025,
+      "step": 19126
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9260234831912143e-05,
+      "loss": 1.0531,
+      "step": 19127
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9251305861868834e-05,
+      "loss": 1.1457,
+      "step": 19128
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9242376891825528e-05,
+      "loss": 1.0469,
+      "step": 19129
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9233447921782225e-05,
+      "loss": 1.1349,
+      "step": 19130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9224518951738915e-05,
+      "loss": 1.0465,
+      "step": 19131
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9215589981695613e-05,
+      "loss": 1.0159,
+      "step": 19132
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.920666101165231e-05,
+      "loss": 1.0098,
+      "step": 19133
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9197732041609e-05,
+      "loss": 1.0693,
+      "step": 19134
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9188803071565694e-05,
+      "loss": 1.0776,
+      "step": 19135
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.917987410152239e-05,
+      "loss": 1.0227,
+      "step": 19136
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9170945131479082e-05,
+      "loss": 1.04,
+      "step": 19137
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.916201616143578e-05,
+      "loss": 1.0802,
+      "step": 19138
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9153087191392476e-05,
+      "loss": 1.1272,
+      "step": 19139
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.914415822134917e-05,
+      "loss": 1.2033,
+      "step": 19140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.913522925130586e-05,
+      "loss": 1.0095,
+      "step": 19141
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9126300281262558e-05,
+      "loss": 1.0383,
+      "step": 19142
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9117371311219255e-05,
+      "loss": 0.9908,
+      "step": 19143
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9108442341175946e-05,
+      "loss": 1.054,
+      "step": 19144
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9099513371132643e-05,
+      "loss": 0.966,
+      "step": 19145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9090584401089337e-05,
+      "loss": 1.1481,
+      "step": 19146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9081655431046027e-05,
+      "loss": 1.0604,
+      "step": 19147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9072726461002725e-05,
+      "loss": 1.0109,
+      "step": 19148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9063797490959422e-05,
+      "loss": 1.0788,
+      "step": 19149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9054868520916112e-05,
+      "loss": 1.0247,
+      "step": 19150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.904593955087281e-05,
+      "loss": 0.9652,
+      "step": 19151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9037010580829503e-05,
+      "loss": 0.9677,
+      "step": 19152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9028081610786194e-05,
+      "loss": 1.1044,
+      "step": 19153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.901915264074289e-05,
+      "loss": 0.9415,
+      "step": 19154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.901022367069959e-05,
+      "loss": 0.9827,
+      "step": 19155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.900129470065628e-05,
+      "loss": 0.8878,
+      "step": 19156
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8992365730612976e-05,
+      "loss": 1.0396,
+      "step": 19157
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.898343676056967e-05,
+      "loss": 1.1064,
+      "step": 19158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.897450779052636e-05,
+      "loss": 0.9752,
+      "step": 19159
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8965578820483058e-05,
+      "loss": 1.0347,
+      "step": 19160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8956649850439755e-05,
+      "loss": 0.9901,
+      "step": 19161
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8947720880396445e-05,
+      "loss": 1.1359,
+      "step": 19162
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8938791910353143e-05,
+      "loss": 0.9642,
+      "step": 19163
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8929862940309837e-05,
+      "loss": 1.0441,
+      "step": 19164
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8920933970266534e-05,
+      "loss": 0.9353,
+      "step": 19165
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8912005000223224e-05,
+      "loss": 0.9355,
+      "step": 19166
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.890307603017992e-05,
+      "loss": 0.9905,
+      "step": 19167
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8894147060136615e-05,
+      "loss": 1.1159,
+      "step": 19168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.888521809009331e-05,
+      "loss": 1.0101,
+      "step": 19169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8876289120050003e-05,
+      "loss": 0.9426,
+      "step": 19170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.88673601500067e-05,
+      "loss": 1.0884,
+      "step": 19171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.885843117996339e-05,
+      "loss": 1.0358,
+      "step": 19172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8849502209920088e-05,
+      "loss": 1.0337,
+      "step": 19173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8840573239876782e-05,
+      "loss": 0.9274,
+      "step": 19174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8831644269833476e-05,
+      "loss": 0.9774,
+      "step": 19175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.882271529979017e-05,
+      "loss": 1.0735,
+      "step": 19176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8813786329746867e-05,
+      "loss": 1.1378,
+      "step": 19177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8804857359703558e-05,
+      "loss": 0.9769,
+      "step": 19178
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8795928389660255e-05,
+      "loss": 1.0127,
+      "step": 19179
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.878699941961695e-05,
+      "loss": 0.8274,
+      "step": 19180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.877807044957364e-05,
+      "loss": 1.1888,
+      "step": 19181
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8769141479530336e-05,
+      "loss": 1.0945,
+      "step": 19182
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8760212509487034e-05,
+      "loss": 1.0675,
+      "step": 19183
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8751283539443724e-05,
+      "loss": 1.0269,
+      "step": 19184
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.874235456940042e-05,
+      "loss": 0.9909,
+      "step": 19185
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8733425599357115e-05,
+      "loss": 0.8591,
+      "step": 19186
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8724496629313806e-05,
+      "loss": 0.9704,
+      "step": 19187
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8715567659270503e-05,
+      "loss": 0.9491,
+      "step": 19188
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.87066386892272e-05,
+      "loss": 0.8849,
+      "step": 19189
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8697709719183898e-05,
+      "loss": 0.9767,
+      "step": 19190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8688780749140588e-05,
+      "loss": 1.1585,
+      "step": 19191
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8679851779097282e-05,
+      "loss": 0.9982,
+      "step": 19192
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.867092280905398e-05,
+      "loss": 1.0202,
+      "step": 19193
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.866199383901067e-05,
+      "loss": 1.0256,
+      "step": 19194
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8653064868967367e-05,
+      "loss": 1.1093,
+      "step": 19195
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8644135898924064e-05,
+      "loss": 1.0204,
+      "step": 19196
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8635206928880755e-05,
+      "loss": 1.0201,
+      "step": 19197
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.862627795883745e-05,
+      "loss": 0.9686,
+      "step": 19198
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8617348988794146e-05,
+      "loss": 0.9513,
+      "step": 19199
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8608420018750836e-05,
+      "loss": 1.0649,
+      "step": 19200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8599491048707533e-05,
+      "loss": 1.0249,
+      "step": 19201
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.859056207866423e-05,
+      "loss": 1.0124,
+      "step": 19202
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.858163310862092e-05,
+      "loss": 0.9401,
+      "step": 19203
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8572704138577615e-05,
+      "loss": 0.9506,
+      "step": 19204
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8563775168534312e-05,
+      "loss": 1.0989,
+      "step": 19205
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8554846198491003e-05,
+      "loss": 1.0508,
+      "step": 19206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.85459172284477e-05,
+      "loss": 1.0057,
+      "step": 19207
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8536988258404397e-05,
+      "loss": 1.0018,
+      "step": 19208
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8528059288361088e-05,
+      "loss": 0.9361,
+      "step": 19209
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.851913031831778e-05,
+      "loss": 0.9122,
+      "step": 19210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.851020134827448e-05,
+      "loss": 0.985,
+      "step": 19211
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.850127237823117e-05,
+      "loss": 1.1974,
+      "step": 19212
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8492343408187867e-05,
+      "loss": 1.0108,
+      "step": 19213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.848341443814456e-05,
+      "loss": 1.0407,
+      "step": 19214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8474485468101258e-05,
+      "loss": 0.9253,
+      "step": 19215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8465556498057948e-05,
+      "loss": 1.249,
+      "step": 19216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8456627528014646e-05,
+      "loss": 0.9941,
+      "step": 19217
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8447698557971343e-05,
+      "loss": 1.1647,
+      "step": 19218
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8438769587928033e-05,
+      "loss": 1.1102,
+      "step": 19219
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8429840617884727e-05,
+      "loss": 0.9787,
+      "step": 19220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8420911647841424e-05,
+      "loss": 0.9568,
+      "step": 19221
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8411982677798115e-05,
+      "loss": 0.9652,
+      "step": 19222
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8403053707754812e-05,
+      "loss": 1.0916,
+      "step": 19223
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.839412473771151e-05,
+      "loss": 1.0533,
+      "step": 19224
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.83851957676682e-05,
+      "loss": 1.1585,
+      "step": 19225
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8376266797624894e-05,
+      "loss": 0.9922,
+      "step": 19226
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.836733782758159e-05,
+      "loss": 1.1802,
+      "step": 19227
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.835840885753828e-05,
+      "loss": 1.0393,
+      "step": 19228
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.834947988749498e-05,
+      "loss": 0.9878,
+      "step": 19229
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8340550917451676e-05,
+      "loss": 1.0381,
+      "step": 19230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8331621947408366e-05,
+      "loss": 0.8959,
+      "step": 19231
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.832269297736506e-05,
+      "loss": 1.0852,
+      "step": 19232
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8313764007321758e-05,
+      "loss": 1.0516,
+      "step": 19233
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8304835037278448e-05,
+      "loss": 1.0302,
+      "step": 19234
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8295906067235145e-05,
+      "loss": 1.0447,
+      "step": 19235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8286977097191843e-05,
+      "loss": 0.9867,
+      "step": 19236
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8278048127148533e-05,
+      "loss": 0.9937,
+      "step": 19237
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8269119157105227e-05,
+      "loss": 1.0897,
+      "step": 19238
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8260190187061924e-05,
+      "loss": 0.9364,
+      "step": 19239
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.825126121701862e-05,
+      "loss": 1.03,
+      "step": 19240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8242332246975312e-05,
+      "loss": 1.0305,
+      "step": 19241
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.823340327693201e-05,
+      "loss": 1.0514,
+      "step": 19242
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8224474306888703e-05,
+      "loss": 1.0456,
+      "step": 19243
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8215545336845394e-05,
+      "loss": 1.0492,
+      "step": 19244
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.820661636680209e-05,
+      "loss": 1.0599,
+      "step": 19245
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8197687396758788e-05,
+      "loss": 1.0299,
+      "step": 19246
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.818875842671548e-05,
+      "loss": 1.0144,
+      "step": 19247
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8179829456672176e-05,
+      "loss": 1.06,
+      "step": 19248
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.817090048662887e-05,
+      "loss": 0.9228,
+      "step": 19249
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.816197151658556e-05,
+      "loss": 1.0835,
+      "step": 19250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8153042546542257e-05,
+      "loss": 1.0065,
+      "step": 19251
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8144113576498955e-05,
+      "loss": 0.9278,
+      "step": 19252
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8135184606455645e-05,
+      "loss": 1.0864,
+      "step": 19253
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8126255636412342e-05,
+      "loss": 1.0557,
+      "step": 19254
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8117326666369036e-05,
+      "loss": 1.1448,
+      "step": 19255
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8108397696325727e-05,
+      "loss": 1.0172,
+      "step": 19256
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8099468726282424e-05,
+      "loss": 0.9347,
+      "step": 19257
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.809053975623912e-05,
+      "loss": 1.1166,
+      "step": 19258
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8081610786195812e-05,
+      "loss": 1.0524,
+      "step": 19259
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.807268181615251e-05,
+      "loss": 0.9918,
+      "step": 19260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8063752846109203e-05,
+      "loss": 1.0332,
+      "step": 19261
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.80548238760659e-05,
+      "loss": 1.1989,
+      "step": 19262
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.804589490602259e-05,
+      "loss": 1.0236,
+      "step": 19263
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8036965935979288e-05,
+      "loss": 1.0284,
+      "step": 19264
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8028036965935982e-05,
+      "loss": 1.0441,
+      "step": 19265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8019107995892672e-05,
+      "loss": 0.9094,
+      "step": 19266
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.801017902584937e-05,
+      "loss": 0.9377,
+      "step": 19267
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8001250055806067e-05,
+      "loss": 0.9932,
+      "step": 19268
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7992321085762757e-05,
+      "loss": 1.0718,
+      "step": 19269
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7983392115719454e-05,
+      "loss": 0.8602,
+      "step": 19270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.797446314567615e-05,
+      "loss": 0.9416,
+      "step": 19271
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.796553417563284e-05,
+      "loss": 1.0876,
+      "step": 19272
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7956605205589536e-05,
+      "loss": 1.1269,
+      "step": 19273
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7947676235546233e-05,
+      "loss": 0.9693,
+      "step": 19274
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7938747265502924e-05,
+      "loss": 0.9761,
+      "step": 19275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.792981829545962e-05,
+      "loss": 0.9593,
+      "step": 19276
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7920889325416315e-05,
+      "loss": 1.0003,
+      "step": 19277
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7911960355373005e-05,
+      "loss": 0.9907,
+      "step": 19278
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7903031385329703e-05,
+      "loss": 1.0562,
+      "step": 19279
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.78941024152864e-05,
+      "loss": 0.9794,
+      "step": 19280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.788517344524309e-05,
+      "loss": 1.1061,
+      "step": 19281
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7876244475199788e-05,
+      "loss": 1.0679,
+      "step": 19282
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.786731550515648e-05,
+      "loss": 1.0195,
+      "step": 19283
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7858386535113172e-05,
+      "loss": 0.9152,
+      "step": 19284
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.784945756506987e-05,
+      "loss": 1.0201,
+      "step": 19285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7840528595026567e-05,
+      "loss": 1.0217,
+      "step": 19286
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7831599624983264e-05,
+      "loss": 1.0682,
+      "step": 19287
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7822670654939954e-05,
+      "loss": 0.9937,
+      "step": 19288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7813741684896648e-05,
+      "loss": 0.9735,
+      "step": 19289
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7804812714853345e-05,
+      "loss": 0.9812,
+      "step": 19290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7795883744810036e-05,
+      "loss": 1.1211,
+      "step": 19291
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7786954774766733e-05,
+      "loss": 1.1701,
+      "step": 19292
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.777802580472343e-05,
+      "loss": 1.0068,
+      "step": 19293
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.776909683468012e-05,
+      "loss": 0.9559,
+      "step": 19294
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7760167864636815e-05,
+      "loss": 0.898,
+      "step": 19295
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7751238894593512e-05,
+      "loss": 1.0187,
+      "step": 19296
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7742309924550202e-05,
+      "loss": 1.0628,
+      "step": 19297
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.77333809545069e-05,
+      "loss": 1.0757,
+      "step": 19298
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7724451984463594e-05,
+      "loss": 0.9297,
+      "step": 19299
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7715523014420287e-05,
+      "loss": 0.9716,
+      "step": 19300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.770659404437698e-05,
+      "loss": 1.1093,
+      "step": 19301
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.769766507433368e-05,
+      "loss": 1.0541,
+      "step": 19302
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.768873610429037e-05,
+      "loss": 1.0247,
+      "step": 19303
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7679807134247066e-05,
+      "loss": 1.0768,
+      "step": 19304
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.767087816420376e-05,
+      "loss": 1.0083,
+      "step": 19305
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7661949194160454e-05,
+      "loss": 0.9832,
+      "step": 19306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7653020224117148e-05,
+      "loss": 0.9719,
+      "step": 19307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7644091254073845e-05,
+      "loss": 0.8929,
+      "step": 19308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7635162284030536e-05,
+      "loss": 1.0216,
+      "step": 19309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7626233313987233e-05,
+      "loss": 1.098,
+      "step": 19310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7617304343943927e-05,
+      "loss": 1.0444,
+      "step": 19311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7608375373900624e-05,
+      "loss": 0.9426,
+      "step": 19312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7599446403857315e-05,
+      "loss": 0.8985,
+      "step": 19313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7590517433814012e-05,
+      "loss": 1.2034,
+      "step": 19314
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.758158846377071e-05,
+      "loss": 1.1348,
+      "step": 19315
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.75726594937274e-05,
+      "loss": 1.0553,
+      "step": 19316
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7563730523684093e-05,
+      "loss": 0.9724,
+      "step": 19317
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.755480155364079e-05,
+      "loss": 1.0259,
+      "step": 19318
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.754587258359748e-05,
+      "loss": 1.0837,
+      "step": 19319
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.753694361355418e-05,
+      "loss": 0.9551,
+      "step": 19320
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7528014643510876e-05,
+      "loss": 1.1301,
+      "step": 19321
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7519085673467566e-05,
+      "loss": 1.0158,
+      "step": 19322
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.751015670342426e-05,
+      "loss": 1.0279,
+      "step": 19323
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7501227733380957e-05,
+      "loss": 1.0207,
+      "step": 19324
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7492298763337648e-05,
+      "loss": 0.8783,
+      "step": 19325
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7483369793294345e-05,
+      "loss": 1.0049,
+      "step": 19326
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7474440823251042e-05,
+      "loss": 1.1163,
+      "step": 19327
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7465511853207733e-05,
+      "loss": 1.0483,
+      "step": 19328
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7456582883164427e-05,
+      "loss": 1.044,
+      "step": 19329
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7447653913121124e-05,
+      "loss": 0.9729,
+      "step": 19330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7438724943077814e-05,
+      "loss": 0.9584,
+      "step": 19331
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.742979597303451e-05,
+      "loss": 1.0608,
+      "step": 19332
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.742086700299121e-05,
+      "loss": 1.0147,
+      "step": 19333
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.74119380329479e-05,
+      "loss": 0.9522,
+      "step": 19334
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7403009062904593e-05,
+      "loss": 1.0726,
+      "step": 19335
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.739408009286129e-05,
+      "loss": 1.0222,
+      "step": 19336
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7385151122817988e-05,
+      "loss": 1.0314,
+      "step": 19337
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7376222152774678e-05,
+      "loss": 1.0031,
+      "step": 19338
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7367293182731375e-05,
+      "loss": 1.0966,
+      "step": 19339
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.735836421268807e-05,
+      "loss": 1.0836,
+      "step": 19340
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.734943524264476e-05,
+      "loss": 1.2058,
+      "step": 19341
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7340506272601457e-05,
+      "loss": 0.9507,
+      "step": 19342
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7331577302558154e-05,
+      "loss": 0.989,
+      "step": 19343
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7322648332514845e-05,
+      "loss": 0.9106,
+      "step": 19344
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7313719362471542e-05,
+      "loss": 0.9549,
+      "step": 19345
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7304790392428236e-05,
+      "loss": 0.9779,
+      "step": 19346
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7295861422384926e-05,
+      "loss": 0.9924,
+      "step": 19347
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7286932452341624e-05,
+      "loss": 1.1764,
+      "step": 19348
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.727800348229832e-05,
+      "loss": 0.9994,
+      "step": 19349
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.726907451225501e-05,
+      "loss": 1.1168,
+      "step": 19350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7260145542211705e-05,
+      "loss": 1.0635,
+      "step": 19351
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7251216572168402e-05,
+      "loss": 1.2317,
+      "step": 19352
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7242287602125093e-05,
+      "loss": 1.0917,
+      "step": 19353
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.723335863208179e-05,
+      "loss": 1.0467,
+      "step": 19354
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7224429662038487e-05,
+      "loss": 1.0606,
+      "step": 19355
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7215500691995178e-05,
+      "loss": 1.1177,
+      "step": 19356
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7206571721951872e-05,
+      "loss": 0.9489,
+      "step": 19357
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.719764275190857e-05,
+      "loss": 1.124,
+      "step": 19358
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.718871378186526e-05,
+      "loss": 0.9819,
+      "step": 19359
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7179784811821957e-05,
+      "loss": 1.0588,
+      "step": 19360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7170855841778654e-05,
+      "loss": 1.0649,
+      "step": 19361
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7161926871735348e-05,
+      "loss": 1.2181,
+      "step": 19362
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.715299790169204e-05,
+      "loss": 1.0633,
+      "step": 19363
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7144068931648736e-05,
+      "loss": 1.0195,
+      "step": 19364
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7135139961605433e-05,
+      "loss": 1.0824,
+      "step": 19365
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7126210991562123e-05,
+      "loss": 0.9565,
+      "step": 19366
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.711728202151882e-05,
+      "loss": 1.0191,
+      "step": 19367
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7108353051475515e-05,
+      "loss": 1.0398,
+      "step": 19368
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7099424081432205e-05,
+      "loss": 1.1313,
+      "step": 19369
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7090495111388902e-05,
+      "loss": 1.069,
+      "step": 19370
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.70815661413456e-05,
+      "loss": 1.0345,
+      "step": 19371
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.707263717130229e-05,
+      "loss": 1.0039,
+      "step": 19372
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7063708201258987e-05,
+      "loss": 1.0888,
+      "step": 19373
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.705477923121568e-05,
+      "loss": 1.1028,
+      "step": 19374
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.704585026117237e-05,
+      "loss": 0.9788,
+      "step": 19375
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.703692129112907e-05,
+      "loss": 0.9998,
+      "step": 19376
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7027992321085766e-05,
+      "loss": 1.1281,
+      "step": 19377
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7019063351042457e-05,
+      "loss": 1.1389,
+      "step": 19378
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7010134380999154e-05,
+      "loss": 0.9228,
+      "step": 19379
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7001205410955848e-05,
+      "loss": 1.0751,
+      "step": 19380
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6992276440912538e-05,
+      "loss": 1.0152,
+      "step": 19381
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6983347470869235e-05,
+      "loss": 0.9862,
+      "step": 19382
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6974418500825933e-05,
+      "loss": 1.0478,
+      "step": 19383
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6965489530782623e-05,
+      "loss": 1.0165,
+      "step": 19384
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.695656056073932e-05,
+      "loss": 1.0843,
+      "step": 19385
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6947631590696014e-05,
+      "loss": 0.8924,
+      "step": 19386
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.693870262065271e-05,
+      "loss": 1.0663,
+      "step": 19387
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6929773650609402e-05,
+      "loss": 1.0826,
+      "step": 19388
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.69208446805661e-05,
+      "loss": 1.0908,
+      "step": 19389
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6911915710522793e-05,
+      "loss": 0.981,
+      "step": 19390
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6902986740479487e-05,
+      "loss": 1.1372,
+      "step": 19391
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.689405777043618e-05,
+      "loss": 1.201,
+      "step": 19392
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6885128800392878e-05,
+      "loss": 1.0777,
+      "step": 19393
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.687619983034957e-05,
+      "loss": 0.953,
+      "step": 19394
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6867270860306266e-05,
+      "loss": 0.9481,
+      "step": 19395
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.685834189026296e-05,
+      "loss": 1.0054,
+      "step": 19396
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6849412920219654e-05,
+      "loss": 0.912,
+      "step": 19397
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6840483950176348e-05,
+      "loss": 1.0136,
+      "step": 19398
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6831554980133045e-05,
+      "loss": 1.0609,
+      "step": 19399
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6822626010089735e-05,
+      "loss": 1.0743,
+      "step": 19400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6813697040046433e-05,
+      "loss": 1.0677,
+      "step": 19401
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6804768070003126e-05,
+      "loss": 1.0611,
+      "step": 19402
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6795839099959817e-05,
+      "loss": 1.0109,
+      "step": 19403
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6786910129916514e-05,
+      "loss": 0.8564,
+      "step": 19404
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.677798115987321e-05,
+      "loss": 1.0191,
+      "step": 19405
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6769052189829902e-05,
+      "loss": 1.0958,
+      "step": 19406
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.67601232197866e-05,
+      "loss": 1.0292,
+      "step": 19407
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6751194249743293e-05,
+      "loss": 1.079,
+      "step": 19408
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.674226527969999e-05,
+      "loss": 0.98,
+      "step": 19409
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.673333630965668e-05,
+      "loss": 1.0475,
+      "step": 19410
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6724407339613378e-05,
+      "loss": 0.9232,
+      "step": 19411
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6715478369570075e-05,
+      "loss": 1.0774,
+      "step": 19412
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6706549399526766e-05,
+      "loss": 1.0037,
+      "step": 19413
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.669762042948346e-05,
+      "loss": 0.9562,
+      "step": 19414
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6688691459440157e-05,
+      "loss": 1.0486,
+      "step": 19415
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6679762489396847e-05,
+      "loss": 0.962,
+      "step": 19416
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6670833519353545e-05,
+      "loss": 0.9862,
+      "step": 19417
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6661904549310242e-05,
+      "loss": 0.9857,
+      "step": 19418
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6652975579266932e-05,
+      "loss": 1.1261,
+      "step": 19419
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6644046609223626e-05,
+      "loss": 0.968,
+      "step": 19420
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6635117639180323e-05,
+      "loss": 1.1276,
+      "step": 19421
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6626188669137014e-05,
+      "loss": 1.0173,
+      "step": 19422
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.661725969909371e-05,
+      "loss": 0.9853,
+      "step": 19423
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.660833072905041e-05,
+      "loss": 1.0496,
+      "step": 19424
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.65994017590071e-05,
+      "loss": 1.1618,
+      "step": 19425
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6590472788963793e-05,
+      "loss": 1.0245,
+      "step": 19426
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.658154381892049e-05,
+      "loss": 1.1036,
+      "step": 19427
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.657261484887718e-05,
+      "loss": 0.9655,
+      "step": 19428
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6563685878833878e-05,
+      "loss": 0.9588,
+      "step": 19429
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6554756908790575e-05,
+      "loss": 1.0009,
+      "step": 19430
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6545827938747266e-05,
+      "loss": 1.0071,
+      "step": 19431
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.653689896870396e-05,
+      "loss": 1.0879,
+      "step": 19432
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6527969998660657e-05,
+      "loss": 0.9343,
+      "step": 19433
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6519041028617354e-05,
+      "loss": 1.2188,
+      "step": 19434
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6510112058574044e-05,
+      "loss": 1.0461,
+      "step": 19435
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6501183088530738e-05,
+      "loss": 0.9537,
+      "step": 19436
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6492254118487436e-05,
+      "loss": 1.0052,
+      "step": 19437
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6483325148444126e-05,
+      "loss": 1.0186,
+      "step": 19438
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6474396178400823e-05,
+      "loss": 0.9236,
+      "step": 19439
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.646546720835752e-05,
+      "loss": 1.1122,
+      "step": 19440
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.645653823831421e-05,
+      "loss": 0.9786,
+      "step": 19441
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6447609268270905e-05,
+      "loss": 1.0409,
+      "step": 19442
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6438680298227602e-05,
+      "loss": 1.0595,
+      "step": 19443
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6429751328184293e-05,
+      "loss": 1.0341,
+      "step": 19444
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.642082235814099e-05,
+      "loss": 1.0403,
+      "step": 19445
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6411893388097687e-05,
+      "loss": 1.0221,
+      "step": 19446
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6402964418054378e-05,
+      "loss": 1.1434,
+      "step": 19447
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.639403544801107e-05,
+      "loss": 0.8595,
+      "step": 19448
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.638510647796777e-05,
+      "loss": 1.017,
+      "step": 19449
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.637617750792446e-05,
+      "loss": 1.0456,
+      "step": 19450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6367248537881156e-05,
+      "loss": 1.1274,
+      "step": 19451
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6358319567837854e-05,
+      "loss": 1.047,
+      "step": 19452
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6349390597794544e-05,
+      "loss": 1.0629,
+      "step": 19453
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6340461627751238e-05,
+      "loss": 1.0023,
+      "step": 19454
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6331532657707935e-05,
+      "loss": 1.0083,
+      "step": 19455
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6322603687664626e-05,
+      "loss": 1.1067,
+      "step": 19456
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6313674717621323e-05,
+      "loss": 0.9901,
+      "step": 19457
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.630474574757802e-05,
+      "loss": 1.004,
+      "step": 19458
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6295816777534714e-05,
+      "loss": 1.0084,
+      "step": 19459
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6286887807491405e-05,
+      "loss": 1.0466,
+      "step": 19460
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6277958837448102e-05,
+      "loss": 0.9991,
+      "step": 19461
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.62690298674048e-05,
+      "loss": 1.1049,
+      "step": 19462
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.626010089736149e-05,
+      "loss": 1.0229,
+      "step": 19463
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6251171927318187e-05,
+      "loss": 1.1353,
+      "step": 19464
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.624224295727488e-05,
+      "loss": 1.0094,
+      "step": 19465
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.623331398723157e-05,
+      "loss": 1.0339,
+      "step": 19466
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.622438501718827e-05,
+      "loss": 1.2353,
+      "step": 19467
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6215456047144966e-05,
+      "loss": 1.0569,
+      "step": 19468
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6206527077101656e-05,
+      "loss": 1.0396,
+      "step": 19469
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6197598107058354e-05,
+      "loss": 1.0608,
+      "step": 19470
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6188669137015047e-05,
+      "loss": 0.9655,
+      "step": 19471
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6179740166971738e-05,
+      "loss": 1.1408,
+      "step": 19472
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6170811196928435e-05,
+      "loss": 1.0177,
+      "step": 19473
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6161882226885132e-05,
+      "loss": 0.9477,
+      "step": 19474
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6152953256841823e-05,
+      "loss": 0.9613,
+      "step": 19475
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.614402428679852e-05,
+      "loss": 1.0407,
+      "step": 19476
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6135095316755214e-05,
+      "loss": 0.9432,
+      "step": 19477
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6126166346711904e-05,
+      "loss": 1.0764,
+      "step": 19478
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6117237376668602e-05,
+      "loss": 1.1459,
+      "step": 19479
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.61083084066253e-05,
+      "loss": 1.0221,
+      "step": 19480
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.609937943658199e-05,
+      "loss": 1.1963,
+      "step": 19481
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6090450466538687e-05,
+      "loss": 1.0754,
+      "step": 19482
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.608152149649538e-05,
+      "loss": 1.0713,
+      "step": 19483
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6072592526452078e-05,
+      "loss": 1.0453,
+      "step": 19484
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6063663556408768e-05,
+      "loss": 0.9622,
+      "step": 19485
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6054734586365466e-05,
+      "loss": 1.0031,
+      "step": 19486
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.604580561632216e-05,
+      "loss": 1.1045,
+      "step": 19487
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.603687664627885e-05,
+      "loss": 1.1335,
+      "step": 19488
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6027947676235547e-05,
+      "loss": 1.0491,
+      "step": 19489
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6019018706192244e-05,
+      "loss": 1.0132,
+      "step": 19490
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6010089736148935e-05,
+      "loss": 1.129,
+      "step": 19491
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6001160766105632e-05,
+      "loss": 1.0548,
+      "step": 19492
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5992231796062326e-05,
+      "loss": 1.0277,
+      "step": 19493
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5983302826019017e-05,
+      "loss": 0.9378,
+      "step": 19494
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5974373855975714e-05,
+      "loss": 1.101,
+      "step": 19495
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.596544488593241e-05,
+      "loss": 0.9954,
+      "step": 19496
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.59565159158891e-05,
+      "loss": 0.9291,
+      "step": 19497
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.59475869458458e-05,
+      "loss": 1.0001,
+      "step": 19498
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5938657975802493e-05,
+      "loss": 0.9516,
+      "step": 19499
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5929729005759183e-05,
+      "loss": 1.0156,
+      "step": 19500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.592080003571588e-05,
+      "loss": 0.9828,
+      "step": 19501
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5911871065672578e-05,
+      "loss": 0.9619,
+      "step": 19502
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5902942095629268e-05,
+      "loss": 1.0006,
+      "step": 19503
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5894013125585965e-05,
+      "loss": 1.015,
+      "step": 19504
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.588508415554266e-05,
+      "loss": 0.8544,
+      "step": 19505
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.587615518549935e-05,
+      "loss": 0.9847,
+      "step": 19506
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5867226215456047e-05,
+      "loss": 1.0411,
+      "step": 19507
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5858297245412744e-05,
+      "loss": 1.1998,
+      "step": 19508
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.584936827536944e-05,
+      "loss": 1.0307,
+      "step": 19509
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5840439305326132e-05,
+      "loss": 0.9311,
+      "step": 19510
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5831510335282826e-05,
+      "loss": 1.1023,
+      "step": 19511
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5822581365239523e-05,
+      "loss": 1.0413,
+      "step": 19512
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5813652395196214e-05,
+      "loss": 1.0293,
+      "step": 19513
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.580472342515291e-05,
+      "loss": 1.0269,
+      "step": 19514
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5795794455109608e-05,
+      "loss": 1.2178,
+      "step": 19515
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.57868654850663e-05,
+      "loss": 0.9091,
+      "step": 19516
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5777936515022992e-05,
+      "loss": 0.9085,
+      "step": 19517
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.576900754497969e-05,
+      "loss": 0.9849,
+      "step": 19518
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.576007857493638e-05,
+      "loss": 0.9574,
+      "step": 19519
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5751149604893077e-05,
+      "loss": 1.0493,
+      "step": 19520
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.574222063484977e-05,
+      "loss": 0.9741,
+      "step": 19521
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5733291664806465e-05,
+      "loss": 1.0587,
+      "step": 19522
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.572436269476316e-05,
+      "loss": 1.0732,
+      "step": 19523
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5715433724719856e-05,
+      "loss": 1.0168,
+      "step": 19524
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5706504754676547e-05,
+      "loss": 1.0016,
+      "step": 19525
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5697575784633244e-05,
+      "loss": 1.022,
+      "step": 19526
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5688646814589938e-05,
+      "loss": 1.0622,
+      "step": 19527
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5679717844546632e-05,
+      "loss": 1.0554,
+      "step": 19528
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5670788874503326e-05,
+      "loss": 0.9803,
+      "step": 19529
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5661859904460023e-05,
+      "loss": 0.9702,
+      "step": 19530
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.565293093441672e-05,
+      "loss": 1.0687,
+      "step": 19531
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.564400196437341e-05,
+      "loss": 0.9654,
+      "step": 19532
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5635072994330104e-05,
+      "loss": 1.0157,
+      "step": 19533
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5626144024286802e-05,
+      "loss": 0.9588,
+      "step": 19534
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5617215054243492e-05,
+      "loss": 1.0861,
+      "step": 19535
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.560828608420019e-05,
+      "loss": 1.0393,
+      "step": 19536
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5599357114156887e-05,
+      "loss": 1.0364,
+      "step": 19537
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5590428144113577e-05,
+      "loss": 1.0806,
+      "step": 19538
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.558149917407027e-05,
+      "loss": 0.9903,
+      "step": 19539
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.557257020402697e-05,
+      "loss": 0.998,
+      "step": 19540
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.556364123398366e-05,
+      "loss": 1.0639,
+      "step": 19541
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5554712263940356e-05,
+      "loss": 1.1519,
+      "step": 19542
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5545783293897053e-05,
+      "loss": 0.9544,
+      "step": 19543
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5536854323853744e-05,
+      "loss": 1.0162,
+      "step": 19544
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5527925353810438e-05,
+      "loss": 1.1299,
+      "step": 19545
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5518996383767135e-05,
+      "loss": 0.9502,
+      "step": 19546
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5510067413723825e-05,
+      "loss": 1.0729,
+      "step": 19547
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5501138443680523e-05,
+      "loss": 1.0681,
+      "step": 19548
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.549220947363722e-05,
+      "loss": 1.027,
+      "step": 19549
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.548328050359391e-05,
+      "loss": 1.0348,
+      "step": 19550
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5474351533550604e-05,
+      "loss": 0.952,
+      "step": 19551
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.54654225635073e-05,
+      "loss": 1.0303,
+      "step": 19552
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5456493593463992e-05,
+      "loss": 0.9388,
+      "step": 19553
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.544756462342069e-05,
+      "loss": 0.9955,
+      "step": 19554
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5438635653377387e-05,
+      "loss": 1.2051,
+      "step": 19555
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.542970668333408e-05,
+      "loss": 1.171,
+      "step": 19556
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.542077771329077e-05,
+      "loss": 1.1206,
+      "step": 19557
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5411848743247468e-05,
+      "loss": 0.9919,
+      "step": 19558
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5402919773204165e-05,
+      "loss": 0.9644,
+      "step": 19559
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5393990803160856e-05,
+      "loss": 0.9725,
+      "step": 19560
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5385061833117553e-05,
+      "loss": 1.0218,
+      "step": 19561
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5376132863074247e-05,
+      "loss": 0.9939,
+      "step": 19562
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5367203893030937e-05,
+      "loss": 1.0101,
+      "step": 19563
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5358274922987635e-05,
+      "loss": 1.0345,
+      "step": 19564
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5349345952944332e-05,
+      "loss": 1.056,
+      "step": 19565
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5340416982901022e-05,
+      "loss": 0.9818,
+      "step": 19566
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.533148801285772e-05,
+      "loss": 1.2187,
+      "step": 19567
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5322559042814414e-05,
+      "loss": 0.9673,
+      "step": 19568
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5313630072771104e-05,
+      "loss": 0.8484,
+      "step": 19569
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.53047011027278e-05,
+      "loss": 1.0324,
+      "step": 19570
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.52957721326845e-05,
+      "loss": 0.9859,
+      "step": 19571
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.528684316264119e-05,
+      "loss": 1.0207,
+      "step": 19572
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5277914192597883e-05,
+      "loss": 1.0391,
+      "step": 19573
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.526898522255458e-05,
+      "loss": 1.0843,
+      "step": 19574
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.526005625251127e-05,
+      "loss": 0.997,
+      "step": 19575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5251127282467968e-05,
+      "loss": 0.954,
+      "step": 19576
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5242198312424665e-05,
+      "loss": 0.9855,
+      "step": 19577
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5233269342381356e-05,
+      "loss": 1.1037,
+      "step": 19578
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.522434037233805e-05,
+      "loss": 0.9518,
+      "step": 19579
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5215411402294747e-05,
+      "loss": 0.9694,
+      "step": 19580
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5206482432251444e-05,
+      "loss": 1.0346,
+      "step": 19581
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5197553462208135e-05,
+      "loss": 0.9735,
+      "step": 19582
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5188624492164832e-05,
+      "loss": 1.0416,
+      "step": 19583
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5179695522121526e-05,
+      "loss": 1.1312,
+      "step": 19584
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5170766552078216e-05,
+      "loss": 1.1329,
+      "step": 19585
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5161837582034913e-05,
+      "loss": 1.0614,
+      "step": 19586
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.515290861199161e-05,
+      "loss": 0.9261,
+      "step": 19587
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.51439796419483e-05,
+      "loss": 0.926,
+      "step": 19588
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5135050671905e-05,
+      "loss": 0.989,
+      "step": 19589
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5126121701861692e-05,
+      "loss": 1.1536,
+      "step": 19590
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5117192731818383e-05,
+      "loss": 0.9437,
+      "step": 19591
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.510826376177508e-05,
+      "loss": 0.9637,
+      "step": 19592
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5099334791731777e-05,
+      "loss": 1.009,
+      "step": 19593
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5090405821688468e-05,
+      "loss": 0.998,
+      "step": 19594
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5081476851645165e-05,
+      "loss": 0.8408,
+      "step": 19595
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.507254788160186e-05,
+      "loss": 0.9752,
+      "step": 19596
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.506361891155855e-05,
+      "loss": 0.9064,
+      "step": 19597
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5054689941515247e-05,
+      "loss": 1.0152,
+      "step": 19598
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5045760971471944e-05,
+      "loss": 1.0682,
+      "step": 19599
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5036832001428634e-05,
+      "loss": 1.0308,
+      "step": 19600
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.502790303138533e-05,
+      "loss": 0.8217,
+      "step": 19601
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5018974061342025e-05,
+      "loss": 1.0879,
+      "step": 19602
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5010045091298716e-05,
+      "loss": 1.0015,
+      "step": 19603
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5001116121255413e-05,
+      "loss": 1.0404,
+      "step": 19604
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.499218715121211e-05,
+      "loss": 1.0263,
+      "step": 19605
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4983258181168804e-05,
+      "loss": 1.0487,
+      "step": 19606
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4974329211125498e-05,
+      "loss": 1.0213,
+      "step": 19607
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4965400241082192e-05,
+      "loss": 1.0124,
+      "step": 19608
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4956471271038886e-05,
+      "loss": 1.0037,
+      "step": 19609
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.494754230099558e-05,
+      "loss": 0.9984,
+      "step": 19610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4938613330952277e-05,
+      "loss": 0.9776,
+      "step": 19611
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.492968436090897e-05,
+      "loss": 1.1533,
+      "step": 19612
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4920755390865665e-05,
+      "loss": 1.1042,
+      "step": 19613
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.491182642082236e-05,
+      "loss": 0.9346,
+      "step": 19614
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4902897450779053e-05,
+      "loss": 1.0818,
+      "step": 19615
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.489396848073575e-05,
+      "loss": 1.1392,
+      "step": 19616
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4885039510692444e-05,
+      "loss": 0.9052,
+      "step": 19617
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4876110540649138e-05,
+      "loss": 0.9642,
+      "step": 19618
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.486718157060583e-05,
+      "loss": 1.0948,
+      "step": 19619
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4858252600562525e-05,
+      "loss": 0.9766,
+      "step": 19620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.484932363051922e-05,
+      "loss": 0.9703,
+      "step": 19621
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4840394660475916e-05,
+      "loss": 0.9778,
+      "step": 19622
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.483146569043261e-05,
+      "loss": 1.0275,
+      "step": 19623
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4822536720389304e-05,
+      "loss": 1.1522,
+      "step": 19624
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4813607750345998e-05,
+      "loss": 1.0617,
+      "step": 19625
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4804678780302692e-05,
+      "loss": 0.9681,
+      "step": 19626
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.479574981025939e-05,
+      "loss": 1.1293,
+      "step": 19627
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4786820840216083e-05,
+      "loss": 1.1324,
+      "step": 19628
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4777891870172777e-05,
+      "loss": 1.0574,
+      "step": 19629
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.476896290012947e-05,
+      "loss": 0.9742,
+      "step": 19630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4760033930086165e-05,
+      "loss": 1.0623,
+      "step": 19631
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.475110496004286e-05,
+      "loss": 0.9721,
+      "step": 19632
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4742175989999556e-05,
+      "loss": 1.1696,
+      "step": 19633
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.473324701995625e-05,
+      "loss": 0.9645,
+      "step": 19634
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4724318049912943e-05,
+      "loss": 1.1064,
+      "step": 19635
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4715389079869637e-05,
+      "loss": 0.9798,
+      "step": 19636
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.470646010982633e-05,
+      "loss": 0.9636,
+      "step": 19637
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.469753113978303e-05,
+      "loss": 0.9596,
+      "step": 19638
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4688602169739722e-05,
+      "loss": 0.9671,
+      "step": 19639
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4679673199696416e-05,
+      "loss": 0.9828,
+      "step": 19640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.467074422965311e-05,
+      "loss": 0.9675,
+      "step": 19641
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4661815259609804e-05,
+      "loss": 1.0424,
+      "step": 19642
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4652886289566498e-05,
+      "loss": 0.9062,
+      "step": 19643
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4643957319523195e-05,
+      "loss": 1.0849,
+      "step": 19644
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.463502834947989e-05,
+      "loss": 1.1462,
+      "step": 19645
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4626099379436583e-05,
+      "loss": 0.9407,
+      "step": 19646
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4617170409393277e-05,
+      "loss": 1.0413,
+      "step": 19647
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.460824143934997e-05,
+      "loss": 0.9837,
+      "step": 19648
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4599312469306668e-05,
+      "loss": 1.1131,
+      "step": 19649
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.459038349926336e-05,
+      "loss": 1.258,
+      "step": 19650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4581454529220056e-05,
+      "loss": 1.0035,
+      "step": 19651
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4572525559176753e-05,
+      "loss": 1.1055,
+      "step": 19652
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4563596589133443e-05,
+      "loss": 1.0884,
+      "step": 19653
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4554667619090137e-05,
+      "loss": 1.0145,
+      "step": 19654
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4545738649046834e-05,
+      "loss": 1.0278,
+      "step": 19655
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4536809679003528e-05,
+      "loss": 1.0113,
+      "step": 19656
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4527880708960222e-05,
+      "loss": 1.0191,
+      "step": 19657
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4518951738916916e-05,
+      "loss": 0.9598,
+      "step": 19658
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.451002276887361e-05,
+      "loss": 0.9999,
+      "step": 19659
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4501093798830307e-05,
+      "loss": 0.9801,
+      "step": 19660
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4492164828787e-05,
+      "loss": 1.0082,
+      "step": 19661
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4483235858743695e-05,
+      "loss": 1.0543,
+      "step": 19662
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4474306888700392e-05,
+      "loss": 1.0674,
+      "step": 19663
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4465377918657083e-05,
+      "loss": 0.9973,
+      "step": 19664
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4456448948613776e-05,
+      "loss": 1.1358,
+      "step": 19665
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4447519978570474e-05,
+      "loss": 1.0656,
+      "step": 19666
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4438591008527168e-05,
+      "loss": 0.962,
+      "step": 19667
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.442966203848386e-05,
+      "loss": 0.8933,
+      "step": 19668
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.442073306844056e-05,
+      "loss": 1.005,
+      "step": 19669
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.441180409839725e-05,
+      "loss": 1.0507,
+      "step": 19670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4402875128353943e-05,
+      "loss": 1.0748,
+      "step": 19671
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.439394615831064e-05,
+      "loss": 0.9855,
+      "step": 19672
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4385017188267334e-05,
+      "loss": 0.9593,
+      "step": 19673
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.437608821822403e-05,
+      "loss": 0.9496,
+      "step": 19674
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4367159248180725e-05,
+      "loss": 1.0817,
+      "step": 19675
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4358230278137416e-05,
+      "loss": 0.8798,
+      "step": 19676
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4349301308094113e-05,
+      "loss": 0.9425,
+      "step": 19677
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4340372338050807e-05,
+      "loss": 0.9333,
+      "step": 19678
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.43314433680075e-05,
+      "loss": 0.9198,
+      "step": 19679
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4322514397964198e-05,
+      "loss": 1.063,
+      "step": 19680
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4313585427920892e-05,
+      "loss": 1.0734,
+      "step": 19681
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4304656457877582e-05,
+      "loss": 0.9621,
+      "step": 19682
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.429572748783428e-05,
+      "loss": 1.0494,
+      "step": 19683
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4286798517790973e-05,
+      "loss": 1.0941,
+      "step": 19684
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.427786954774767e-05,
+      "loss": 1.077,
+      "step": 19685
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4268940577704365e-05,
+      "loss": 0.998,
+      "step": 19686
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4260011607661055e-05,
+      "loss": 0.9651,
+      "step": 19687
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4251082637617752e-05,
+      "loss": 1.1516,
+      "step": 19688
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4242153667574446e-05,
+      "loss": 0.9802,
+      "step": 19689
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.423322469753114e-05,
+      "loss": 1.0606,
+      "step": 19690
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4224295727487837e-05,
+      "loss": 0.9798,
+      "step": 19691
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.421536675744453e-05,
+      "loss": 1.0984,
+      "step": 19692
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4206437787401222e-05,
+      "loss": 1.1146,
+      "step": 19693
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.419750881735792e-05,
+      "loss": 0.9731,
+      "step": 19694
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4188579847314613e-05,
+      "loss": 1.0134,
+      "step": 19695
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.417965087727131e-05,
+      "loss": 0.9711,
+      "step": 19696
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4170721907228004e-05,
+      "loss": 1.0662,
+      "step": 19697
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4161792937184698e-05,
+      "loss": 0.9807,
+      "step": 19698
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.415286396714139e-05,
+      "loss": 0.9898,
+      "step": 19699
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4143934997098086e-05,
+      "loss": 0.9978,
+      "step": 19700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.413500602705478e-05,
+      "loss": 0.9637,
+      "step": 19701
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4126077057011477e-05,
+      "loss": 1.0164,
+      "step": 19702
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.411714808696817e-05,
+      "loss": 0.9189,
+      "step": 19703
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4108219116924864e-05,
+      "loss": 1.083,
+      "step": 19704
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4099290146881558e-05,
+      "loss": 0.9601,
+      "step": 19705
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4090361176838252e-05,
+      "loss": 1.1223,
+      "step": 19706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4081432206794946e-05,
+      "loss": 1.0401,
+      "step": 19707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4072503236751643e-05,
+      "loss": 0.9762,
+      "step": 19708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4063574266708337e-05,
+      "loss": 1.0457,
+      "step": 19709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.405464529666503e-05,
+      "loss": 0.9563,
+      "step": 19710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4045716326621725e-05,
+      "loss": 1.1536,
+      "step": 19711
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.403678735657842e-05,
+      "loss": 0.9239,
+      "step": 19712
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4027858386535116e-05,
+      "loss": 1.1542,
+      "step": 19713
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.401892941649181e-05,
+      "loss": 0.9622,
+      "step": 19714
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4010000446448504e-05,
+      "loss": 0.976,
+      "step": 19715
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4001071476405198e-05,
+      "loss": 1.0318,
+      "step": 19716
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.399214250636189e-05,
+      "loss": 1.1512,
+      "step": 19717
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3983213536318585e-05,
+      "loss": 1.0642,
+      "step": 19718
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3974284566275283e-05,
+      "loss": 0.9889,
+      "step": 19719
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3965355596231976e-05,
+      "loss": 0.9993,
+      "step": 19720
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.395642662618867e-05,
+      "loss": 0.9929,
+      "step": 19721
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3947497656145364e-05,
+      "loss": 0.9213,
+      "step": 19722
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3938568686102058e-05,
+      "loss": 1.1458,
+      "step": 19723
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3929639716058755e-05,
+      "loss": 0.9003,
+      "step": 19724
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.392071074601545e-05,
+      "loss": 1.039,
+      "step": 19725
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3911781775972143e-05,
+      "loss": 0.9796,
+      "step": 19726
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3902852805928837e-05,
+      "loss": 1.028,
+      "step": 19727
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.389392383588553e-05,
+      "loss": 1.0215,
+      "step": 19728
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3884994865842225e-05,
+      "loss": 0.98,
+      "step": 19729
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3876065895798922e-05,
+      "loss": 1.1059,
+      "step": 19730
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3867136925755616e-05,
+      "loss": 1.0737,
+      "step": 19731
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.385820795571231e-05,
+      "loss": 1.0001,
+      "step": 19732
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3849278985669004e-05,
+      "loss": 1.0197,
+      "step": 19733
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3840350015625697e-05,
+      "loss": 1.0934,
+      "step": 19734
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3831421045582395e-05,
+      "loss": 1.0306,
+      "step": 19735
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.382249207553909e-05,
+      "loss": 1.0834,
+      "step": 19736
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3813563105495782e-05,
+      "loss": 0.9323,
+      "step": 19737
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3804634135452476e-05,
+      "loss": 0.9911,
+      "step": 19738
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.379570516540917e-05,
+      "loss": 0.9754,
+      "step": 19739
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3786776195365864e-05,
+      "loss": 1.0548,
+      "step": 19740
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.377784722532256e-05,
+      "loss": 1.14,
+      "step": 19741
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3768918255279255e-05,
+      "loss": 1.042,
+      "step": 19742
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.375998928523595e-05,
+      "loss": 1.0383,
+      "step": 19743
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3751060315192643e-05,
+      "loss": 1.086,
+      "step": 19744
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3742131345149337e-05,
+      "loss": 1.0561,
+      "step": 19745
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3733202375106034e-05,
+      "loss": 1.004,
+      "step": 19746
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3724273405062728e-05,
+      "loss": 0.8665,
+      "step": 19747
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3715344435019422e-05,
+      "loss": 1.1013,
+      "step": 19748
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3706415464976116e-05,
+      "loss": 1.0364,
+      "step": 19749
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.369748649493281e-05,
+      "loss": 0.9876,
+      "step": 19750
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3688557524889503e-05,
+      "loss": 1.1047,
+      "step": 19751
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.36796285548462e-05,
+      "loss": 1.0708,
+      "step": 19752
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3670699584802894e-05,
+      "loss": 0.99,
+      "step": 19753
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.366177061475959e-05,
+      "loss": 1.0799,
+      "step": 19754
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3652841644716282e-05,
+      "loss": 1.0289,
+      "step": 19755
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3643912674672976e-05,
+      "loss": 1.0263,
+      "step": 19756
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3634983704629673e-05,
+      "loss": 1.0452,
+      "step": 19757
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3626054734586367e-05,
+      "loss": 0.9738,
+      "step": 19758
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.361712576454306e-05,
+      "loss": 1.1364,
+      "step": 19759
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.360819679449976e-05,
+      "loss": 1.0473,
+      "step": 19760
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.359926782445645e-05,
+      "loss": 0.9953,
+      "step": 19761
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3590338854413143e-05,
+      "loss": 1.0891,
+      "step": 19762
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.358140988436984e-05,
+      "loss": 1.0306,
+      "step": 19763
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3572480914326534e-05,
+      "loss": 1.0621,
+      "step": 19764
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3563551944283228e-05,
+      "loss": 0.993,
+      "step": 19765
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3554622974239925e-05,
+      "loss": 1.0707,
+      "step": 19766
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3545694004196615e-05,
+      "loss": 0.9892,
+      "step": 19767
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.353676503415331e-05,
+      "loss": 1.0464,
+      "step": 19768
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3527836064110007e-05,
+      "loss": 1.0226,
+      "step": 19769
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.35189070940667e-05,
+      "loss": 1.0535,
+      "step": 19770
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3509978124023398e-05,
+      "loss": 0.9913,
+      "step": 19771
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3501049153980088e-05,
+      "loss": 0.9753,
+      "step": 19772
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3492120183936782e-05,
+      "loss": 0.9653,
+      "step": 19773
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.348319121389348e-05,
+      "loss": 1.0237,
+      "step": 19774
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3474262243850173e-05,
+      "loss": 1.2069,
+      "step": 19775
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3465333273806867e-05,
+      "loss": 0.9519,
+      "step": 19776
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3456404303763564e-05,
+      "loss": 1.1103,
+      "step": 19777
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3447475333720255e-05,
+      "loss": 0.9761,
+      "step": 19778
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.343854636367695e-05,
+      "loss": 1.0376,
+      "step": 19779
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3429617393633646e-05,
+      "loss": 0.9787,
+      "step": 19780
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.342068842359034e-05,
+      "loss": 0.8974,
+      "step": 19781
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3411759453547037e-05,
+      "loss": 1.0164,
+      "step": 19782
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.340283048350373e-05,
+      "loss": 1.0613,
+      "step": 19783
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.339390151346042e-05,
+      "loss": 0.9359,
+      "step": 19784
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.338497254341712e-05,
+      "loss": 1.0276,
+      "step": 19785
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3376043573373812e-05,
+      "loss": 1.112,
+      "step": 19786
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3367114603330506e-05,
+      "loss": 1.0372,
+      "step": 19787
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3358185633287204e-05,
+      "loss": 1.098,
+      "step": 19788
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3349256663243897e-05,
+      "loss": 1.0167,
+      "step": 19789
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3340327693200588e-05,
+      "loss": 0.9788,
+      "step": 19790
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3331398723157285e-05,
+      "loss": 0.9833,
+      "step": 19791
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.332246975311398e-05,
+      "loss": 1.1268,
+      "step": 19792
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3313540783070673e-05,
+      "loss": 0.9578,
+      "step": 19793
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.330461181302737e-05,
+      "loss": 1.0392,
+      "step": 19794
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.329568284298406e-05,
+      "loss": 1.1966,
+      "step": 19795
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3286753872940758e-05,
+      "loss": 1.2237,
+      "step": 19796
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3277824902897452e-05,
+      "loss": 1.0748,
+      "step": 19797
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3268895932854146e-05,
+      "loss": 0.9355,
+      "step": 19798
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3259966962810843e-05,
+      "loss": 1.231,
+      "step": 19799
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3251037992767537e-05,
+      "loss": 1.0739,
+      "step": 19800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3242109022724227e-05,
+      "loss": 0.9648,
+      "step": 19801
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3233180052680925e-05,
+      "loss": 1.0653,
+      "step": 19802
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.322425108263762e-05,
+      "loss": 1.0411,
+      "step": 19803
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3215322112594312e-05,
+      "loss": 1.0174,
+      "step": 19804
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.320639314255101e-05,
+      "loss": 1.091,
+      "step": 19805
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3197464172507703e-05,
+      "loss": 0.966,
+      "step": 19806
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3188535202464397e-05,
+      "loss": 1.1701,
+      "step": 19807
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.317960623242109e-05,
+      "loss": 1.0089,
+      "step": 19808
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3170677262377785e-05,
+      "loss": 1.0333,
+      "step": 19809
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3161748292334482e-05,
+      "loss": 1.103,
+      "step": 19810
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3152819322291176e-05,
+      "loss": 1.2454,
+      "step": 19811
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.314389035224787e-05,
+      "loss": 1.0555,
+      "step": 19812
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3134961382204564e-05,
+      "loss": 1.0498,
+      "step": 19813
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3126032412161258e-05,
+      "loss": 0.8514,
+      "step": 19814
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.311710344211795e-05,
+      "loss": 0.9728,
+      "step": 19815
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.310817447207465e-05,
+      "loss": 1.0174,
+      "step": 19816
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3099245502031343e-05,
+      "loss": 1.1284,
+      "step": 19817
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3090316531988037e-05,
+      "loss": 1.2224,
+      "step": 19818
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.308138756194473e-05,
+      "loss": 0.9665,
+      "step": 19819
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3072458591901424e-05,
+      "loss": 1.0418,
+      "step": 19820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.306352962185812e-05,
+      "loss": 1.0422,
+      "step": 19821
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3054600651814815e-05,
+      "loss": 1.1163,
+      "step": 19822
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.304567168177151e-05,
+      "loss": 0.9732,
+      "step": 19823
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3036742711728203e-05,
+      "loss": 1.1362,
+      "step": 19824
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3027813741684897e-05,
+      "loss": 1.1621,
+      "step": 19825
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.301888477164159e-05,
+      "loss": 1.0402,
+      "step": 19826
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3009955801598288e-05,
+      "loss": 0.9962,
+      "step": 19827
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3001026831554982e-05,
+      "loss": 1.0437,
+      "step": 19828
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2992097861511676e-05,
+      "loss": 0.9735,
+      "step": 19829
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.298316889146837e-05,
+      "loss": 1.0791,
+      "step": 19830
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2974239921425064e-05,
+      "loss": 0.9613,
+      "step": 19831
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.296531095138176e-05,
+      "loss": 1.0065,
+      "step": 19832
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2956381981338455e-05,
+      "loss": 1.0626,
+      "step": 19833
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.294745301129515e-05,
+      "loss": 1.0212,
+      "step": 19834
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2938524041251843e-05,
+      "loss": 0.9714,
+      "step": 19835
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2929595071208536e-05,
+      "loss": 1.0658,
+      "step": 19836
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.292066610116523e-05,
+      "loss": 0.9644,
+      "step": 19837
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2911737131121927e-05,
+      "loss": 1.1071,
+      "step": 19838
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.290280816107862e-05,
+      "loss": 1.058,
+      "step": 19839
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2893879191035315e-05,
+      "loss": 1.0597,
+      "step": 19840
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.288495022099201e-05,
+      "loss": 1.0881,
+      "step": 19841
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2876021250948703e-05,
+      "loss": 0.9297,
+      "step": 19842
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.28670922809054e-05,
+      "loss": 1.1076,
+      "step": 19843
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2858163310862094e-05,
+      "loss": 1.0577,
+      "step": 19844
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2849234340818788e-05,
+      "loss": 0.9661,
+      "step": 19845
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2840305370775482e-05,
+      "loss": 1.1161,
+      "step": 19846
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2831376400732176e-05,
+      "loss": 0.9485,
+      "step": 19847
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.282244743068887e-05,
+      "loss": 0.9359,
+      "step": 19848
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2813518460645567e-05,
+      "loss": 1.0368,
+      "step": 19849
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.280458949060226e-05,
+      "loss": 1.0774,
+      "step": 19850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2795660520558955e-05,
+      "loss": 0.9721,
+      "step": 19851
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.278673155051565e-05,
+      "loss": 1.0647,
+      "step": 19852
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2777802580472342e-05,
+      "loss": 1.0976,
+      "step": 19853
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2768873610429036e-05,
+      "loss": 0.923,
+      "step": 19854
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2759944640385733e-05,
+      "loss": 1.003,
+      "step": 19855
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2751015670342427e-05,
+      "loss": 0.9678,
+      "step": 19856
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.274208670029912e-05,
+      "loss": 0.9296,
+      "step": 19857
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2733157730255815e-05,
+      "loss": 1.013,
+      "step": 19858
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.272422876021251e-05,
+      "loss": 1.0419,
+      "step": 19859
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2715299790169206e-05,
+      "loss": 1.009,
+      "step": 19860
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.27063708201259e-05,
+      "loss": 1.0639,
+      "step": 19861
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2697441850082594e-05,
+      "loss": 1.0656,
+      "step": 19862
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2688512880039288e-05,
+      "loss": 1.0659,
+      "step": 19863
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.267958390999598e-05,
+      "loss": 1.1039,
+      "step": 19864
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2670654939952675e-05,
+      "loss": 1.0623,
+      "step": 19865
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2661725969909373e-05,
+      "loss": 0.9582,
+      "step": 19866
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2652796999866067e-05,
+      "loss": 1.1754,
+      "step": 19867
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2643868029822764e-05,
+      "loss": 1.0175,
+      "step": 19868
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2634939059779454e-05,
+      "loss": 1.0332,
+      "step": 19869
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2626010089736148e-05,
+      "loss": 0.9958,
+      "step": 19870
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2617081119692845e-05,
+      "loss": 1.0075,
+      "step": 19871
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.260815214964954e-05,
+      "loss": 1.1096,
+      "step": 19872
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2599223179606233e-05,
+      "loss": 0.9386,
+      "step": 19873
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.259029420956293e-05,
+      "loss": 1.0955,
+      "step": 19874
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.258136523951962e-05,
+      "loss": 1.0258,
+      "step": 19875
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2572436269476315e-05,
+      "loss": 1.1326,
+      "step": 19876
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2563507299433012e-05,
+      "loss": 0.9974,
+      "step": 19877
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2554578329389706e-05,
+      "loss": 1.0361,
+      "step": 19878
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.25456493593464e-05,
+      "loss": 1.065,
+      "step": 19879
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2536720389303094e-05,
+      "loss": 0.8919,
+      "step": 19880
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2527791419259788e-05,
+      "loss": 1.177,
+      "step": 19881
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2518862449216485e-05,
+      "loss": 1.0547,
+      "step": 19882
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.250993347917318e-05,
+      "loss": 1.073,
+      "step": 19883
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2501004509129873e-05,
+      "loss": 1.001,
+      "step": 19884
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.249207553908657e-05,
+      "loss": 1.0921,
+      "step": 19885
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.248314656904326e-05,
+      "loss": 1.0397,
+      "step": 19886
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2474217598999954e-05,
+      "loss": 0.9268,
+      "step": 19887
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.246528862895665e-05,
+      "loss": 1.0887,
+      "step": 19888
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2456359658913345e-05,
+      "loss": 0.9789,
+      "step": 19889
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.244743068887004e-05,
+      "loss": 1.0867,
+      "step": 19890
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2438501718826736e-05,
+      "loss": 1.146,
+      "step": 19891
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2429572748783427e-05,
+      "loss": 1.0799,
+      "step": 19892
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2420643778740124e-05,
+      "loss": 0.9148,
+      "step": 19893
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2411714808696818e-05,
+      "loss": 0.999,
+      "step": 19894
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2402785838653512e-05,
+      "loss": 1.1094,
+      "step": 19895
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.239385686861021e-05,
+      "loss": 1.0811,
+      "step": 19896
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2384927898566903e-05,
+      "loss": 1.0248,
+      "step": 19897
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2375998928523593e-05,
+      "loss": 0.9899,
+      "step": 19898
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.236706995848029e-05,
+      "loss": 1.0326,
+      "step": 19899
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2358140988436985e-05,
+      "loss": 0.9814,
+      "step": 19900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.234921201839368e-05,
+      "loss": 1.0616,
+      "step": 19901
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2340283048350376e-05,
+      "loss": 1.0495,
+      "step": 19902
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.233135407830707e-05,
+      "loss": 1.0395,
+      "step": 19903
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2322425108263763e-05,
+      "loss": 1.1402,
+      "step": 19904
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2313496138220457e-05,
+      "loss": 1.1881,
+      "step": 19905
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.230456716817715e-05,
+      "loss": 1.0185,
+      "step": 19906
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.229563819813385e-05,
+      "loss": 1.1131,
+      "step": 19907
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2286709228090542e-05,
+      "loss": 1.1596,
+      "step": 19908
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2277780258047233e-05,
+      "loss": 0.8391,
+      "step": 19909
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.226885128800393e-05,
+      "loss": 1.0904,
+      "step": 19910
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2259922317960624e-05,
+      "loss": 1.0858,
+      "step": 19911
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2250993347917318e-05,
+      "loss": 1.0025,
+      "step": 19912
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2242064377874015e-05,
+      "loss": 1.0495,
+      "step": 19913
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.223313540783071e-05,
+      "loss": 1.1235,
+      "step": 19914
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.22242064377874e-05,
+      "loss": 1.1174,
+      "step": 19915
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2215277467744097e-05,
+      "loss": 1.0184,
+      "step": 19916
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.220634849770079e-05,
+      "loss": 1.0828,
+      "step": 19917
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2197419527657488e-05,
+      "loss": 0.9797,
+      "step": 19918
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.218849055761418e-05,
+      "loss": 1.0959,
+      "step": 19919
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2179561587570876e-05,
+      "loss": 0.9801,
+      "step": 19920
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.217063261752757e-05,
+      "loss": 1.0212,
+      "step": 19921
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2161703647484263e-05,
+      "loss": 1.0399,
+      "step": 19922
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2152774677440957e-05,
+      "loss": 1.1351,
+      "step": 19923
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2143845707397654e-05,
+      "loss": 0.9642,
+      "step": 19924
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2134916737354348e-05,
+      "loss": 0.8855,
+      "step": 19925
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2125987767311042e-05,
+      "loss": 1.1238,
+      "step": 19926
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2117058797267736e-05,
+      "loss": 0.9922,
+      "step": 19927
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.210812982722443e-05,
+      "loss": 1.0355,
+      "step": 19928
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2099200857181127e-05,
+      "loss": 1.1232,
+      "step": 19929
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.209027188713782e-05,
+      "loss": 0.9233,
+      "step": 19930
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2081342917094515e-05,
+      "loss": 1.0024,
+      "step": 19931
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.207241394705121e-05,
+      "loss": 1.008,
+      "step": 19932
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2063484977007903e-05,
+      "loss": 1.0172,
+      "step": 19933
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2054556006964596e-05,
+      "loss": 0.9406,
+      "step": 19934
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2045627036921294e-05,
+      "loss": 1.066,
+      "step": 19935
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2036698066877988e-05,
+      "loss": 1.0074,
+      "step": 19936
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.202776909683468e-05,
+      "loss": 0.944,
+      "step": 19937
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2018840126791375e-05,
+      "loss": 0.9835,
+      "step": 19938
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.200991115674807e-05,
+      "loss": 0.9581,
+      "step": 19939
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2000982186704763e-05,
+      "loss": 0.9839,
+      "step": 19940
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.199205321666146e-05,
+      "loss": 1.1368,
+      "step": 19941
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1983124246618154e-05,
+      "loss": 1.1563,
+      "step": 19942
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1974195276574848e-05,
+      "loss": 1.178,
+      "step": 19943
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1965266306531542e-05,
+      "loss": 1.0555,
+      "step": 19944
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1956337336488236e-05,
+      "loss": 1.1061,
+      "step": 19945
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1947408366444933e-05,
+      "loss": 1.0313,
+      "step": 19946
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1938479396401627e-05,
+      "loss": 1.096,
+      "step": 19947
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.192955042635832e-05,
+      "loss": 1.1215,
+      "step": 19948
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1920621456315015e-05,
+      "loss": 1.1934,
+      "step": 19949
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.191169248627171e-05,
+      "loss": 1.0128,
+      "step": 19950
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1902763516228402e-05,
+      "loss": 0.9719,
+      "step": 19951
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.18938345461851e-05,
+      "loss": 1.1958,
+      "step": 19952
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1884905576141794e-05,
+      "loss": 1.0603,
+      "step": 19953
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1875976606098487e-05,
+      "loss": 1.0197,
+      "step": 19954
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.186704763605518e-05,
+      "loss": 1.1016,
+      "step": 19955
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1858118666011875e-05,
+      "loss": 1.0266,
+      "step": 19956
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1849189695968572e-05,
+      "loss": 1.0848,
+      "step": 19957
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1840260725925266e-05,
+      "loss": 0.9736,
+      "step": 19958
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.183133175588196e-05,
+      "loss": 1.0295,
+      "step": 19959
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1822402785838654e-05,
+      "loss": 1.0558,
+      "step": 19960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1813473815795348e-05,
+      "loss": 1.1394,
+      "step": 19961
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1804544845752042e-05,
+      "loss": 0.9273,
+      "step": 19962
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.179561587570874e-05,
+      "loss": 0.9572,
+      "step": 19963
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1786686905665433e-05,
+      "loss": 0.9898,
+      "step": 19964
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1777757935622127e-05,
+      "loss": 1.0074,
+      "step": 19965
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.176882896557882e-05,
+      "loss": 1.0135,
+      "step": 19966
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1759899995535514e-05,
+      "loss": 0.9434,
+      "step": 19967
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1750971025492212e-05,
+      "loss": 1.0202,
+      "step": 19968
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1742042055448906e-05,
+      "loss": 1.0439,
+      "step": 19969
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.17331130854056e-05,
+      "loss": 1.1397,
+      "step": 19970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1724184115362293e-05,
+      "loss": 0.9411,
+      "step": 19971
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1715255145318987e-05,
+      "loss": 1.0188,
+      "step": 19972
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.170632617527568e-05,
+      "loss": 0.9801,
+      "step": 19973
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.169739720523238e-05,
+      "loss": 1.0154,
+      "step": 19974
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1688468235189072e-05,
+      "loss": 0.9969,
+      "step": 19975
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1679539265145766e-05,
+      "loss": 0.9446,
+      "step": 19976
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.167061029510246e-05,
+      "loss": 0.9774,
+      "step": 19977
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1661681325059154e-05,
+      "loss": 1.0518,
+      "step": 19978
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.165275235501585e-05,
+      "loss": 1.0152,
+      "step": 19979
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1643823384972545e-05,
+      "loss": 1.0716,
+      "step": 19980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.163489441492924e-05,
+      "loss": 1.1449,
+      "step": 19981
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1625965444885936e-05,
+      "loss": 0.9614,
+      "step": 19982
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1617036474842627e-05,
+      "loss": 0.9857,
+      "step": 19983
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.160810750479932e-05,
+      "loss": 1.0296,
+      "step": 19984
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1599178534756018e-05,
+      "loss": 0.9402,
+      "step": 19985
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.159024956471271e-05,
+      "loss": 0.9375,
+      "step": 19986
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1581320594669405e-05,
+      "loss": 0.9205,
+      "step": 19987
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1572391624626103e-05,
+      "loss": 1.1855,
+      "step": 19988
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1563462654582793e-05,
+      "loss": 0.9135,
+      "step": 19989
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.155453368453949e-05,
+      "loss": 1.1161,
+      "step": 19990
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1545604714496184e-05,
+      "loss": 1.0628,
+      "step": 19991
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1536675744452878e-05,
+      "loss": 0.9621,
+      "step": 19992
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1527746774409575e-05,
+      "loss": 0.9966,
+      "step": 19993
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1518817804366266e-05,
+      "loss": 1.0072,
+      "step": 19994
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.150988883432296e-05,
+      "loss": 1.1072,
+      "step": 19995
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1500959864279657e-05,
+      "loss": 1.049,
+      "step": 19996
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.149203089423635e-05,
+      "loss": 1.0383,
+      "step": 19997
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1483101924193045e-05,
+      "loss": 0.9372,
+      "step": 19998
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1474172954149742e-05,
+      "loss": 0.9657,
+      "step": 19999
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1465243984106432e-05,
+      "loss": 1.2065,
+      "step": 20000
+    },
+    {
+      "epoch": 2.68,
+      "eval_loss": 1.025498628616333,
+      "eval_runtime": 3047.9077,
+      "eval_samples_per_second": 6.17,
+      "eval_steps_per_second": 0.771,
+      "step": 20000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1456315014063126e-05,
+      "loss": 0.9989,
+      "step": 20001
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1447386044019824e-05,
+      "loss": 1.123,
+      "step": 20002
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1438457073976517e-05,
+      "loss": 0.9121,
+      "step": 20003
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1429528103933215e-05,
+      "loss": 1.0519,
+      "step": 20004
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.142059913388991e-05,
+      "loss": 1.1092,
+      "step": 20005
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.14116701638466e-05,
+      "loss": 1.0024,
+      "step": 20006
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1402741193803296e-05,
+      "loss": 0.9911,
+      "step": 20007
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.139381222375999e-05,
+      "loss": 0.9477,
+      "step": 20008
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1384883253716684e-05,
+      "loss": 0.9201,
+      "step": 20009
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.137595428367338e-05,
+      "loss": 1.0577,
+      "step": 20010
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1367025313630075e-05,
+      "loss": 1.0348,
+      "step": 20011
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1358096343586766e-05,
+      "loss": 1.0021,
+      "step": 20012
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1349167373543463e-05,
+      "loss": 1.0404,
+      "step": 20013
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1340238403500157e-05,
+      "loss": 0.9726,
+      "step": 20014
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1331309433456854e-05,
+      "loss": 1.063,
+      "step": 20015
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1322380463413548e-05,
+      "loss": 0.9935,
+      "step": 20016
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.131345149337024e-05,
+      "loss": 1.0106,
+      "step": 20017
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1304522523326936e-05,
+      "loss": 0.9733,
+      "step": 20018
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.129559355328363e-05,
+      "loss": 0.917,
+      "step": 20019
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1286664583240323e-05,
+      "loss": 1.0783,
+      "step": 20020
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.127773561319702e-05,
+      "loss": 1.0225,
+      "step": 20021
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1268806643153714e-05,
+      "loss": 0.9459,
+      "step": 20022
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1259877673110405e-05,
+      "loss": 1.0946,
+      "step": 20023
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1250948703067102e-05,
+      "loss": 1.2153,
+      "step": 20024
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1242019733023796e-05,
+      "loss": 1.1533,
+      "step": 20025
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1233090762980493e-05,
+      "loss": 0.9874,
+      "step": 20026
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1224161792937187e-05,
+      "loss": 1.0538,
+      "step": 20027
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.121523282289388e-05,
+      "loss": 1.1571,
+      "step": 20028
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1206303852850575e-05,
+      "loss": 1.0511,
+      "step": 20029
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.119737488280727e-05,
+      "loss": 1.1177,
+      "step": 20030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1188445912763963e-05,
+      "loss": 1.09,
+      "step": 20031
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.117951694272066e-05,
+      "loss": 1.0578,
+      "step": 20032
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1170587972677354e-05,
+      "loss": 1.061,
+      "step": 20033
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1161659002634048e-05,
+      "loss": 1.0317,
+      "step": 20034
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.115273003259074e-05,
+      "loss": 1.2422,
+      "step": 20035
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1143801062547435e-05,
+      "loss": 1.0463,
+      "step": 20036
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.113487209250413e-05,
+      "loss": 1.0785,
+      "step": 20037
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1125943122460827e-05,
+      "loss": 0.9889,
+      "step": 20038
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.111701415241752e-05,
+      "loss": 0.9324,
+      "step": 20039
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1108085182374214e-05,
+      "loss": 0.989,
+      "step": 20040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1099156212330908e-05,
+      "loss": 1.0539,
+      "step": 20041
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1090227242287602e-05,
+      "loss": 0.9682,
+      "step": 20042
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.10812982722443e-05,
+      "loss": 0.9632,
+      "step": 20043
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1072369302200993e-05,
+      "loss": 0.9485,
+      "step": 20044
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1063440332157687e-05,
+      "loss": 1.01,
+      "step": 20045
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.105451136211438e-05,
+      "loss": 1.016,
+      "step": 20046
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1045582392071075e-05,
+      "loss": 0.9893,
+      "step": 20047
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.103665342202777e-05,
+      "loss": 0.9785,
+      "step": 20048
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1027724451984466e-05,
+      "loss": 0.9676,
+      "step": 20049
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.101879548194116e-05,
+      "loss": 1.0149,
+      "step": 20050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1009866511897854e-05,
+      "loss": 1.0655,
+      "step": 20051
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1000937541854547e-05,
+      "loss": 1.0547,
+      "step": 20052
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.099200857181124e-05,
+      "loss": 1.0268,
+      "step": 20053
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.098307960176794e-05,
+      "loss": 0.9609,
+      "step": 20054
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0974150631724632e-05,
+      "loss": 1.0455,
+      "step": 20055
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0965221661681326e-05,
+      "loss": 0.9819,
+      "step": 20056
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.095629269163802e-05,
+      "loss": 0.982,
+      "step": 20057
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0947363721594714e-05,
+      "loss": 1.0311,
+      "step": 20058
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0938434751551408e-05,
+      "loss": 1.0564,
+      "step": 20059
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0929505781508105e-05,
+      "loss": 0.9424,
+      "step": 20060
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.09205768114648e-05,
+      "loss": 0.9185,
+      "step": 20061
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0911647841421493e-05,
+      "loss": 1.0267,
+      "step": 20062
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0902718871378187e-05,
+      "loss": 1.0285,
+      "step": 20063
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.089378990133488e-05,
+      "loss": 1.0702,
+      "step": 20064
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0884860931291578e-05,
+      "loss": 1.0397,
+      "step": 20065
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0875931961248272e-05,
+      "loss": 0.9321,
+      "step": 20066
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0867002991204966e-05,
+      "loss": 1.3147,
+      "step": 20067
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.085807402116166e-05,
+      "loss": 1.1403,
+      "step": 20068
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0849145051118353e-05,
+      "loss": 0.9959,
+      "step": 20069
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0840216081075047e-05,
+      "loss": 0.9703,
+      "step": 20070
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0831287111031745e-05,
+      "loss": 1.0352,
+      "step": 20071
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.082235814098844e-05,
+      "loss": 1.0063,
+      "step": 20072
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0813429170945132e-05,
+      "loss": 1.0257,
+      "step": 20073
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0804500200901826e-05,
+      "loss": 0.9968,
+      "step": 20074
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.079557123085852e-05,
+      "loss": 1.101,
+      "step": 20075
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0786642260815217e-05,
+      "loss": 1.0094,
+      "step": 20076
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.077771329077191e-05,
+      "loss": 0.9335,
+      "step": 20077
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0768784320728605e-05,
+      "loss": 1.0265,
+      "step": 20078
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.07598553506853e-05,
+      "loss": 0.8884,
+      "step": 20079
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0750926380641993e-05,
+      "loss": 0.9471,
+      "step": 20080
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0741997410598687e-05,
+      "loss": 0.9847,
+      "step": 20081
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0733068440555384e-05,
+      "loss": 0.9756,
+      "step": 20082
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0724139470512078e-05,
+      "loss": 0.9675,
+      "step": 20083
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.071521050046877e-05,
+      "loss": 0.9817,
+      "step": 20084
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0706281530425465e-05,
+      "loss": 0.9317,
+      "step": 20085
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.069735256038216e-05,
+      "loss": 0.9955,
+      "step": 20086
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0688423590338853e-05,
+      "loss": 1.0207,
+      "step": 20087
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.067949462029555e-05,
+      "loss": 0.9847,
+      "step": 20088
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0670565650252244e-05,
+      "loss": 1.159,
+      "step": 20089
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.066163668020894e-05,
+      "loss": 1.0806,
+      "step": 20090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0652707710165632e-05,
+      "loss": 0.8149,
+      "step": 20091
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0643778740122326e-05,
+      "loss": 0.9918,
+      "step": 20092
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0634849770079023e-05,
+      "loss": 0.981,
+      "step": 20093
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0625920800035717e-05,
+      "loss": 1.0846,
+      "step": 20094
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.061699182999241e-05,
+      "loss": 1.0955,
+      "step": 20095
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0608062859949108e-05,
+      "loss": 1.0787,
+      "step": 20096
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.05991338899058e-05,
+      "loss": 1.0142,
+      "step": 20097
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0590204919862493e-05,
+      "loss": 1.0027,
+      "step": 20098
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.058127594981919e-05,
+      "loss": 1.0049,
+      "step": 20099
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0572346979775884e-05,
+      "loss": 0.9579,
+      "step": 20100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.056341800973258e-05,
+      "loss": 0.9768,
+      "step": 20101
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.055448903968927e-05,
+      "loss": 0.9879,
+      "step": 20102
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0545560069645965e-05,
+      "loss": 0.9712,
+      "step": 20103
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0536631099602663e-05,
+      "loss": 1.0688,
+      "step": 20104
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0527702129559356e-05,
+      "loss": 0.9483,
+      "step": 20105
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.051877315951605e-05,
+      "loss": 1.0237,
+      "step": 20106
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0509844189472748e-05,
+      "loss": 0.9862,
+      "step": 20107
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0500915219429438e-05,
+      "loss": 1.0025,
+      "step": 20108
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0491986249386132e-05,
+      "loss": 1.0758,
+      "step": 20109
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.048305727934283e-05,
+      "loss": 1.094,
+      "step": 20110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0474128309299523e-05,
+      "loss": 1.1131,
+      "step": 20111
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.046519933925622e-05,
+      "loss": 0.9691,
+      "step": 20112
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0456270369212914e-05,
+      "loss": 1.0007,
+      "step": 20113
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0447341399169605e-05,
+      "loss": 0.9888,
+      "step": 20114
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0438412429126302e-05,
+      "loss": 0.9381,
+      "step": 20115
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0429483459082996e-05,
+      "loss": 1.0843,
+      "step": 20116
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.042055448903969e-05,
+      "loss": 1.038,
+      "step": 20117
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0411625518996387e-05,
+      "loss": 0.9963,
+      "step": 20118
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.040269654895308e-05,
+      "loss": 1.017,
+      "step": 20119
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.039376757890977e-05,
+      "loss": 1.1724,
+      "step": 20120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.038483860886647e-05,
+      "loss": 1.0653,
+      "step": 20121
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0375909638823162e-05,
+      "loss": 1.062,
+      "step": 20122
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0366980668779856e-05,
+      "loss": 1.08,
+      "step": 20123
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0358051698736553e-05,
+      "loss": 1.011,
+      "step": 20124
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0349122728693247e-05,
+      "loss": 1.1553,
+      "step": 20125
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.034019375864994e-05,
+      "loss": 1.1129,
+      "step": 20126
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0331264788606635e-05,
+      "loss": 0.9721,
+      "step": 20127
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.032233581856333e-05,
+      "loss": 1.0421,
+      "step": 20128
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0313406848520026e-05,
+      "loss": 1.0398,
+      "step": 20129
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.030447787847672e-05,
+      "loss": 0.9717,
+      "step": 20130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.029554890843341e-05,
+      "loss": 1.0464,
+      "step": 20131
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0286619938390108e-05,
+      "loss": 1.0469,
+      "step": 20132
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.02776909683468e-05,
+      "loss": 1.073,
+      "step": 20133
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0268761998303496e-05,
+      "loss": 0.949,
+      "step": 20134
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0259833028260193e-05,
+      "loss": 1.0791,
+      "step": 20135
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0250904058216887e-05,
+      "loss": 1.0286,
+      "step": 20136
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.024197508817358e-05,
+      "loss": 0.9847,
+      "step": 20137
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0233046118130274e-05,
+      "loss": 0.9708,
+      "step": 20138
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0224117148086968e-05,
+      "loss": 0.8674,
+      "step": 20139
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0215188178043666e-05,
+      "loss": 1.0917,
+      "step": 20140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.020625920800036e-05,
+      "loss": 0.9838,
+      "step": 20141
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0197330237957053e-05,
+      "loss": 0.947,
+      "step": 20142
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0188401267913747e-05,
+      "loss": 0.9678,
+      "step": 20143
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.017947229787044e-05,
+      "loss": 1.0165,
+      "step": 20144
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0170543327827135e-05,
+      "loss": 1.0498,
+      "step": 20145
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0161614357783832e-05,
+      "loss": 1.0036,
+      "step": 20146
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0152685387740526e-05,
+      "loss": 1.0189,
+      "step": 20147
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.014375641769722e-05,
+      "loss": 0.9821,
+      "step": 20148
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0134827447653914e-05,
+      "loss": 1.0292,
+      "step": 20149
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0125898477610608e-05,
+      "loss": 1.0532,
+      "step": 20150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0116969507567305e-05,
+      "loss": 1.0619,
+      "step": 20151
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0108040537524e-05,
+      "loss": 0.9882,
+      "step": 20152
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0099111567480693e-05,
+      "loss": 1.1626,
+      "step": 20153
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0090182597437386e-05,
+      "loss": 0.9024,
+      "step": 20154
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.008125362739408e-05,
+      "loss": 1.0248,
+      "step": 20155
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0072324657350774e-05,
+      "loss": 0.9876,
+      "step": 20156
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.006339568730747e-05,
+      "loss": 1.1142,
+      "step": 20157
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0054466717264165e-05,
+      "loss": 1.0417,
+      "step": 20158
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.004553774722086e-05,
+      "loss": 0.9668,
+      "step": 20159
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0036608777177553e-05,
+      "loss": 0.9591,
+      "step": 20160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0027679807134247e-05,
+      "loss": 1.0662,
+      "step": 20161
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0018750837090944e-05,
+      "loss": 0.9861,
+      "step": 20162
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0009821867047638e-05,
+      "loss": 0.8897,
+      "step": 20163
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0000892897004332e-05,
+      "loss": 1.0884,
+      "step": 20164
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9991963926961026e-05,
+      "loss": 0.9308,
+      "step": 20165
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.998303495691772e-05,
+      "loss": 1.0807,
+      "step": 20166
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9974105986874414e-05,
+      "loss": 1.0297,
+      "step": 20167
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.996517701683111e-05,
+      "loss": 1.0656,
+      "step": 20168
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9956248046787805e-05,
+      "loss": 1.0363,
+      "step": 20169
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.99473190767445e-05,
+      "loss": 0.9638,
+      "step": 20170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9938390106701192e-05,
+      "loss": 1.0232,
+      "step": 20171
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9929461136657886e-05,
+      "loss": 1.0483,
+      "step": 20172
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9920532166614584e-05,
+      "loss": 1.066,
+      "step": 20173
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9911603196571277e-05,
+      "loss": 0.9629,
+      "step": 20174
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.990267422652797e-05,
+      "loss": 1.0327,
+      "step": 20175
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9893745256484665e-05,
+      "loss": 1.0851,
+      "step": 20176
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.988481628644136e-05,
+      "loss": 0.9483,
+      "step": 20177
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9875887316398053e-05,
+      "loss": 0.8619,
+      "step": 20178
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.986695834635475e-05,
+      "loss": 0.9918,
+      "step": 20179
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9858029376311444e-05,
+      "loss": 1.0249,
+      "step": 20180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9849100406268138e-05,
+      "loss": 1.0392,
+      "step": 20181
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9840171436224832e-05,
+      "loss": 1.0252,
+      "step": 20182
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9831242466181526e-05,
+      "loss": 1.054,
+      "step": 20183
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.982231349613822e-05,
+      "loss": 1.017,
+      "step": 20184
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9813384526094917e-05,
+      "loss": 1.001,
+      "step": 20185
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.980445555605161e-05,
+      "loss": 1.0524,
+      "step": 20186
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9795526586008304e-05,
+      "loss": 1.1398,
+      "step": 20187
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9786597615965e-05,
+      "loss": 1.041,
+      "step": 20188
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9777668645921692e-05,
+      "loss": 1.0098,
+      "step": 20189
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.976873967587839e-05,
+      "loss": 1.0083,
+      "step": 20190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9759810705835083e-05,
+      "loss": 1.0023,
+      "step": 20191
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9750881735791777e-05,
+      "loss": 0.98,
+      "step": 20192
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.974195276574847e-05,
+      "loss": 1.0916,
+      "step": 20193
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9733023795705165e-05,
+      "loss": 1.1598,
+      "step": 20194
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.972409482566186e-05,
+      "loss": 0.9485,
+      "step": 20195
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9715165855618556e-05,
+      "loss": 0.9692,
+      "step": 20196
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.970623688557525e-05,
+      "loss": 0.9262,
+      "step": 20197
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9697307915531947e-05,
+      "loss": 1.027,
+      "step": 20198
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9688378945488638e-05,
+      "loss": 1.0221,
+      "step": 20199
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.967944997544533e-05,
+      "loss": 1.0615,
+      "step": 20200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.967052100540203e-05,
+      "loss": 1.0169,
+      "step": 20201
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9661592035358723e-05,
+      "loss": 0.9492,
+      "step": 20202
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9652663065315416e-05,
+      "loss": 0.9956,
+      "step": 20203
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9643734095272114e-05,
+      "loss": 1.0419,
+      "step": 20204
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9634805125228804e-05,
+      "loss": 1.1081,
+      "step": 20205
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9625876155185498e-05,
+      "loss": 1.0217,
+      "step": 20206
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9616947185142195e-05,
+      "loss": 1.0721,
+      "step": 20207
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.960801821509889e-05,
+      "loss": 0.941,
+      "step": 20208
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9599089245055583e-05,
+      "loss": 0.9169,
+      "step": 20209
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.959016027501228e-05,
+      "loss": 0.8718,
+      "step": 20210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.958123130496897e-05,
+      "loss": 1.1185,
+      "step": 20211
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9572302334925668e-05,
+      "loss": 1.073,
+      "step": 20212
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9563373364882362e-05,
+      "loss": 1.0699,
+      "step": 20213
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9554444394839056e-05,
+      "loss": 1.0697,
+      "step": 20214
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9545515424795753e-05,
+      "loss": 1.0452,
+      "step": 20215
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9536586454752444e-05,
+      "loss": 0.977,
+      "step": 20216
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9527657484709137e-05,
+      "loss": 0.9492,
+      "step": 20217
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9518728514665835e-05,
+      "loss": 1.1479,
+      "step": 20218
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.950979954462253e-05,
+      "loss": 0.993,
+      "step": 20219
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9500870574579222e-05,
+      "loss": 0.8854,
+      "step": 20220
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.949194160453592e-05,
+      "loss": 0.9901,
+      "step": 20221
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.948301263449261e-05,
+      "loss": 1.1591,
+      "step": 20222
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9474083664449307e-05,
+      "loss": 1.071,
+      "step": 20223
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9465154694406e-05,
+      "loss": 1.1336,
+      "step": 20224
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9456225724362695e-05,
+      "loss": 1.0602,
+      "step": 20225
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9447296754319392e-05,
+      "loss": 1.0177,
+      "step": 20226
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9438367784276086e-05,
+      "loss": 1.0488,
+      "step": 20227
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9429438814232777e-05,
+      "loss": 0.9571,
+      "step": 20228
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9420509844189474e-05,
+      "loss": 0.969,
+      "step": 20229
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9411580874146168e-05,
+      "loss": 1.0733,
+      "step": 20230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9402651904102862e-05,
+      "loss": 1.0327,
+      "step": 20231
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.939372293405956e-05,
+      "loss": 1.0663,
+      "step": 20232
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9384793964016253e-05,
+      "loss": 1.086,
+      "step": 20233
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9375864993972947e-05,
+      "loss": 1.0801,
+      "step": 20234
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.936693602392964e-05,
+      "loss": 1.0156,
+      "step": 20235
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9358007053886334e-05,
+      "loss": 1.0012,
+      "step": 20236
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9349078083843032e-05,
+      "loss": 1.0662,
+      "step": 20237
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9340149113799726e-05,
+      "loss": 1.2251,
+      "step": 20238
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9331220143756416e-05,
+      "loss": 0.9063,
+      "step": 20239
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9322291173713113e-05,
+      "loss": 1.1234,
+      "step": 20240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9313362203669807e-05,
+      "loss": 1.0261,
+      "step": 20241
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.93044332336265e-05,
+      "loss": 1.0994,
+      "step": 20242
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.92955042635832e-05,
+      "loss": 1.0046,
+      "step": 20243
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9286575293539892e-05,
+      "loss": 1.1557,
+      "step": 20244
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9277646323496583e-05,
+      "loss": 1.0257,
+      "step": 20245
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.926871735345328e-05,
+      "loss": 0.9882,
+      "step": 20246
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9259788383409974e-05,
+      "loss": 0.9865,
+      "step": 20247
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.925085941336667e-05,
+      "loss": 0.9555,
+      "step": 20248
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9241930443323365e-05,
+      "loss": 1.0772,
+      "step": 20249
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.923300147328006e-05,
+      "loss": 1.059,
+      "step": 20250
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9224072503236753e-05,
+      "loss": 1.0055,
+      "step": 20251
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9215143533193447e-05,
+      "loss": 1.05,
+      "step": 20252
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.920621456315014e-05,
+      "loss": 1.0171,
+      "step": 20253
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9197285593106838e-05,
+      "loss": 0.9677,
+      "step": 20254
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.918835662306353e-05,
+      "loss": 1.0518,
+      "step": 20255
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9179427653020225e-05,
+      "loss": 0.9146,
+      "step": 20256
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.917049868297692e-05,
+      "loss": 1.1832,
+      "step": 20257
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9161569712933613e-05,
+      "loss": 1.0374,
+      "step": 20258
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.915264074289031e-05,
+      "loss": 0.9558,
+      "step": 20259
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9143711772847004e-05,
+      "loss": 1.0134,
+      "step": 20260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9134782802803698e-05,
+      "loss": 0.9835,
+      "step": 20261
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9125853832760392e-05,
+      "loss": 0.9824,
+      "step": 20262
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9116924862717086e-05,
+      "loss": 1.0713,
+      "step": 20263
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.910799589267378e-05,
+      "loss": 0.9513,
+      "step": 20264
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9099066922630477e-05,
+      "loss": 1.0607,
+      "step": 20265
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.909013795258717e-05,
+      "loss": 1.0978,
+      "step": 20266
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9081208982543865e-05,
+      "loss": 1.0061,
+      "step": 20267
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.907228001250056e-05,
+      "loss": 0.9591,
+      "step": 20268
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9063351042457252e-05,
+      "loss": 1.1067,
+      "step": 20269
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9054422072413946e-05,
+      "loss": 0.9792,
+      "step": 20270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9045493102370644e-05,
+      "loss": 1.0273,
+      "step": 20271
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9036564132327337e-05,
+      "loss": 1.1354,
+      "step": 20272
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.902763516228403e-05,
+      "loss": 1.1527,
+      "step": 20273
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9018706192240725e-05,
+      "loss": 0.9548,
+      "step": 20274
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.900977722219742e-05,
+      "loss": 1.1101,
+      "step": 20275
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9000848252154116e-05,
+      "loss": 1.0325,
+      "step": 20276
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.899191928211081e-05,
+      "loss": 0.9703,
+      "step": 20277
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8982990312067504e-05,
+      "loss": 1.0389,
+      "step": 20278
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8974061342024198e-05,
+      "loss": 1.0623,
+      "step": 20279
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8965132371980892e-05,
+      "loss": 1.0379,
+      "step": 20280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8956203401937586e-05,
+      "loss": 1.0802,
+      "step": 20281
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8947274431894283e-05,
+      "loss": 1.0894,
+      "step": 20282
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8938345461850977e-05,
+      "loss": 0.9891,
+      "step": 20283
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.892941649180767e-05,
+      "loss": 1.0987,
+      "step": 20284
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8920487521764365e-05,
+      "loss": 1.1273,
+      "step": 20285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.891155855172106e-05,
+      "loss": 0.9719,
+      "step": 20286
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8902629581677756e-05,
+      "loss": 1.1467,
+      "step": 20287
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.889370061163445e-05,
+      "loss": 1.0982,
+      "step": 20288
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8884771641591143e-05,
+      "loss": 1.0852,
+      "step": 20289
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8875842671547837e-05,
+      "loss": 1.0674,
+      "step": 20290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.886691370150453e-05,
+      "loss": 1.1125,
+      "step": 20291
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8857984731461225e-05,
+      "loss": 0.9188,
+      "step": 20292
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8849055761417922e-05,
+      "loss": 1.1335,
+      "step": 20293
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8840126791374616e-05,
+      "loss": 1.0895,
+      "step": 20294
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8831197821331313e-05,
+      "loss": 0.9266,
+      "step": 20295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8822268851288004e-05,
+      "loss": 0.9514,
+      "step": 20296
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8813339881244698e-05,
+      "loss": 1.0873,
+      "step": 20297
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8804410911201395e-05,
+      "loss": 1.0378,
+      "step": 20298
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.879548194115809e-05,
+      "loss": 0.8593,
+      "step": 20299
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8786552971114783e-05,
+      "loss": 0.9475,
+      "step": 20300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8777624001071477e-05,
+      "loss": 0.9767,
+      "step": 20301
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.876869503102817e-05,
+      "loss": 0.9285,
+      "step": 20302
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8759766060984864e-05,
+      "loss": 1.1031,
+      "step": 20303
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.875083709094156e-05,
+      "loss": 1.1171,
+      "step": 20304
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8741908120898255e-05,
+      "loss": 1.0084,
+      "step": 20305
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.873297915085495e-05,
+      "loss": 0.9478,
+      "step": 20306
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8724050180811643e-05,
+      "loss": 0.9863,
+      "step": 20307
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8715121210768337e-05,
+      "loss": 1.0144,
+      "step": 20308
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8706192240725034e-05,
+      "loss": 1.0124,
+      "step": 20309
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8697263270681728e-05,
+      "loss": 0.9823,
+      "step": 20310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8688334300638422e-05,
+      "loss": 0.9978,
+      "step": 20311
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.867940533059512e-05,
+      "loss": 0.9257,
+      "step": 20312
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.867047636055181e-05,
+      "loss": 1.0006,
+      "step": 20313
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8661547390508504e-05,
+      "loss": 1.0534,
+      "step": 20314
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.86526184204652e-05,
+      "loss": 1.0889,
+      "step": 20315
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8643689450421895e-05,
+      "loss": 0.9271,
+      "step": 20316
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.863476048037859e-05,
+      "loss": 1.022,
+      "step": 20317
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8625831510335286e-05,
+      "loss": 1.0092,
+      "step": 20318
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8616902540291976e-05,
+      "loss": 1.0025,
+      "step": 20319
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8607973570248674e-05,
+      "loss": 1.1961,
+      "step": 20320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8599044600205368e-05,
+      "loss": 1.0648,
+      "step": 20321
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.859011563016206e-05,
+      "loss": 1.2252,
+      "step": 20322
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.858118666011876e-05,
+      "loss": 0.946,
+      "step": 20323
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.857225769007545e-05,
+      "loss": 1.0068,
+      "step": 20324
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8563328720032143e-05,
+      "loss": 0.9548,
+      "step": 20325
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.855439974998884e-05,
+      "loss": 1.005,
+      "step": 20326
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8545470779945534e-05,
+      "loss": 1.1851,
+      "step": 20327
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8536541809902228e-05,
+      "loss": 0.9565,
+      "step": 20328
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8527612839858925e-05,
+      "loss": 0.973,
+      "step": 20329
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8518683869815616e-05,
+      "loss": 1.1258,
+      "step": 20330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.850975489977231e-05,
+      "loss": 1.0328,
+      "step": 20331
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8500825929729007e-05,
+      "loss": 1.0454,
+      "step": 20332
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.84918969596857e-05,
+      "loss": 0.936,
+      "step": 20333
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8482967989642398e-05,
+      "loss": 1.0908,
+      "step": 20334
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8474039019599092e-05,
+      "loss": 0.9462,
+      "step": 20335
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8465110049555782e-05,
+      "loss": 1.2088,
+      "step": 20336
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.845618107951248e-05,
+      "loss": 0.9598,
+      "step": 20337
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8447252109469173e-05,
+      "loss": 1.075,
+      "step": 20338
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8438323139425867e-05,
+      "loss": 1.053,
+      "step": 20339
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8429394169382565e-05,
+      "loss": 1.0603,
+      "step": 20340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.842046519933926e-05,
+      "loss": 1.0929,
+      "step": 20341
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.841153622929595e-05,
+      "loss": 1.2702,
+      "step": 20342
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8402607259252646e-05,
+      "loss": 0.9876,
+      "step": 20343
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.839367828920934e-05,
+      "loss": 1.1384,
+      "step": 20344
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8384749319166037e-05,
+      "loss": 1.0203,
+      "step": 20345
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.837582034912273e-05,
+      "loss": 1.0654,
+      "step": 20346
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8366891379079425e-05,
+      "loss": 1.017,
+      "step": 20347
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.835796240903612e-05,
+      "loss": 1.0511,
+      "step": 20348
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8349033438992813e-05,
+      "loss": 1.1251,
+      "step": 20349
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8340104468949507e-05,
+      "loss": 1.0694,
+      "step": 20350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8331175498906204e-05,
+      "loss": 1.0806,
+      "step": 20351
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8322246528862898e-05,
+      "loss": 0.9748,
+      "step": 20352
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8313317558819588e-05,
+      "loss": 1.0134,
+      "step": 20353
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8304388588776286e-05,
+      "loss": 0.9909,
+      "step": 20354
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.829545961873298e-05,
+      "loss": 1.0753,
+      "step": 20355
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8286530648689673e-05,
+      "loss": 1.0284,
+      "step": 20356
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.827760167864637e-05,
+      "loss": 1.1114,
+      "step": 20357
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8268672708603064e-05,
+      "loss": 1.1988,
+      "step": 20358
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8259743738559758e-05,
+      "loss": 1.1289,
+      "step": 20359
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8250814768516452e-05,
+      "loss": 1.0504,
+      "step": 20360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8241885798473146e-05,
+      "loss": 1.1405,
+      "step": 20361
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8232956828429843e-05,
+      "loss": 1.152,
+      "step": 20362
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8224027858386537e-05,
+      "loss": 1.0497,
+      "step": 20363
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.821509888834323e-05,
+      "loss": 0.9316,
+      "step": 20364
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8206169918299925e-05,
+      "loss": 0.97,
+      "step": 20365
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.819724094825662e-05,
+      "loss": 1.0372,
+      "step": 20366
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8188311978213313e-05,
+      "loss": 1.052,
+      "step": 20367
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.817938300817001e-05,
+      "loss": 1.0391,
+      "step": 20368
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8170454038126704e-05,
+      "loss": 1.0143,
+      "step": 20369
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8161525068083398e-05,
+      "loss": 0.9423,
+      "step": 20370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.815259609804009e-05,
+      "loss": 1.085,
+      "step": 20371
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8143667127996785e-05,
+      "loss": 0.983,
+      "step": 20372
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8134738157953483e-05,
+      "loss": 0.995,
+      "step": 20373
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8125809187910176e-05,
+      "loss": 1.0134,
+      "step": 20374
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.811688021786687e-05,
+      "loss": 1.0457,
+      "step": 20375
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8107951247823564e-05,
+      "loss": 1.0438,
+      "step": 20376
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8099022277780258e-05,
+      "loss": 1.1372,
+      "step": 20377
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8090093307736952e-05,
+      "loss": 0.9833,
+      "step": 20378
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.808116433769365e-05,
+      "loss": 1.0108,
+      "step": 20379
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8072235367650343e-05,
+      "loss": 0.9054,
+      "step": 20380
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8063306397607037e-05,
+      "loss": 0.9922,
+      "step": 20381
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.805437742756373e-05,
+      "loss": 1.0104,
+      "step": 20382
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8045448457520425e-05,
+      "loss": 1.0554,
+      "step": 20383
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8036519487477122e-05,
+      "loss": 0.9865,
+      "step": 20384
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8027590517433816e-05,
+      "loss": 0.963,
+      "step": 20385
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.801866154739051e-05,
+      "loss": 1.2158,
+      "step": 20386
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8009732577347203e-05,
+      "loss": 1.0207,
+      "step": 20387
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8000803607303897e-05,
+      "loss": 1.0207,
+      "step": 20388
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.799187463726059e-05,
+      "loss": 0.9573,
+      "step": 20389
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.798294566721729e-05,
+      "loss": 0.9393,
+      "step": 20390
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7974016697173982e-05,
+      "loss": 0.9993,
+      "step": 20391
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7965087727130676e-05,
+      "loss": 1.0235,
+      "step": 20392
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.795615875708737e-05,
+      "loss": 1.0064,
+      "step": 20393
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7947229787044064e-05,
+      "loss": 1.0317,
+      "step": 20394
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.793830081700076e-05,
+      "loss": 0.9514,
+      "step": 20395
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7929371846957455e-05,
+      "loss": 1.0488,
+      "step": 20396
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.792044287691415e-05,
+      "loss": 1.0339,
+      "step": 20397
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7911513906870843e-05,
+      "loss": 1.0097,
+      "step": 20398
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7902584936827537e-05,
+      "loss": 1.132,
+      "step": 20399
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.789365596678423e-05,
+      "loss": 1.0063,
+      "step": 20400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7884726996740928e-05,
+      "loss": 0.9953,
+      "step": 20401
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.787579802669762e-05,
+      "loss": 0.911,
+      "step": 20402
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7866869056654316e-05,
+      "loss": 1.0339,
+      "step": 20403
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.785794008661101e-05,
+      "loss": 1.2052,
+      "step": 20404
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7849011116567703e-05,
+      "loss": 1.211,
+      "step": 20405
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.78400821465244e-05,
+      "loss": 1.0372,
+      "step": 20406
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7831153176481094e-05,
+      "loss": 0.9921,
+      "step": 20407
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7822224206437788e-05,
+      "loss": 1.0781,
+      "step": 20408
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7813295236394482e-05,
+      "loss": 1.1109,
+      "step": 20409
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7804366266351176e-05,
+      "loss": 1.0219,
+      "step": 20410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.779543729630787e-05,
+      "loss": 1.0046,
+      "step": 20411
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7786508326264567e-05,
+      "loss": 1.0024,
+      "step": 20412
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.777757935622126e-05,
+      "loss": 1.0177,
+      "step": 20413
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7768650386177955e-05,
+      "loss": 1.1101,
+      "step": 20414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.775972141613465e-05,
+      "loss": 0.9405,
+      "step": 20415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7750792446091343e-05,
+      "loss": 0.9374,
+      "step": 20416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7741863476048036e-05,
+      "loss": 0.9555,
+      "step": 20417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7732934506004734e-05,
+      "loss": 1.0554,
+      "step": 20418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7724005535961428e-05,
+      "loss": 0.9529,
+      "step": 20419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7715076565918125e-05,
+      "loss": 1.0611,
+      "step": 20420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7706147595874815e-05,
+      "loss": 1.0595,
+      "step": 20421
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.769721862583151e-05,
+      "loss": 1.1463,
+      "step": 20422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7688289655788206e-05,
+      "loss": 0.9964,
+      "step": 20423
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.76793606857449e-05,
+      "loss": 1.1266,
+      "step": 20424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7670431715701594e-05,
+      "loss": 1.0825,
+      "step": 20425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.766150274565829e-05,
+      "loss": 0.9753,
+      "step": 20426
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7652573775614982e-05,
+      "loss": 1.1056,
+      "step": 20427
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7643644805571676e-05,
+      "loss": 1.0942,
+      "step": 20428
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7634715835528373e-05,
+      "loss": 1.0464,
+      "step": 20429
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7625786865485067e-05,
+      "loss": 1.0775,
+      "step": 20430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7616857895441764e-05,
+      "loss": 1.0343,
+      "step": 20431
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7607928925398458e-05,
+      "loss": 0.9353,
+      "step": 20432
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.759899995535515e-05,
+      "loss": 0.9578,
+      "step": 20433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7590070985311846e-05,
+      "loss": 1.111,
+      "step": 20434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.758114201526854e-05,
+      "loss": 1.0109,
+      "step": 20435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7572213045225234e-05,
+      "loss": 1.0064,
+      "step": 20436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.756328407518193e-05,
+      "loss": 1.0254,
+      "step": 20437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.755435510513862e-05,
+      "loss": 0.928,
+      "step": 20438
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7545426135095315e-05,
+      "loss": 0.953,
+      "step": 20439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7536497165052012e-05,
+      "loss": 1.129,
+      "step": 20440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7527568195008706e-05,
+      "loss": 1.0157,
+      "step": 20441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7518639224965404e-05,
+      "loss": 1.1101,
+      "step": 20442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7509710254922097e-05,
+      "loss": 1.0274,
+      "step": 20443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7500781284878788e-05,
+      "loss": 1.1367,
+      "step": 20444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7491852314835485e-05,
+      "loss": 0.999,
+      "step": 20445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.748292334479218e-05,
+      "loss": 1.0649,
+      "step": 20446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7473994374748873e-05,
+      "loss": 1.0351,
+      "step": 20447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.746506540470557e-05,
+      "loss": 1.0823,
+      "step": 20448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7456136434662264e-05,
+      "loss": 1.0347,
+      "step": 20449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7447207464618954e-05,
+      "loss": 0.9303,
+      "step": 20450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7438278494575652e-05,
+      "loss": 1.0387,
+      "step": 20451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7429349524532346e-05,
+      "loss": 1.1365,
+      "step": 20452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.742042055448904e-05,
+      "loss": 1.2982,
+      "step": 20453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7411491584445737e-05,
+      "loss": 1.0964,
+      "step": 20454
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.740256261440243e-05,
+      "loss": 0.9778,
+      "step": 20455
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7393633644359124e-05,
+      "loss": 1.0564,
+      "step": 20456
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.738470467431582e-05,
+      "loss": 1.1232,
+      "step": 20457
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7375775704272512e-05,
+      "loss": 1.0032,
+      "step": 20458
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.736684673422921e-05,
+      "loss": 1.1632,
+      "step": 20459
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7357917764185903e-05,
+      "loss": 1.1779,
+      "step": 20460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7348988794142594e-05,
+      "loss": 1.1276,
+      "step": 20461
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.734005982409929e-05,
+      "loss": 0.934,
+      "step": 20462
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7331130854055985e-05,
+      "loss": 1.0769,
+      "step": 20463
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.732220188401268e-05,
+      "loss": 1.0451,
+      "step": 20464
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7313272913969376e-05,
+      "loss": 1.0054,
+      "step": 20465
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.730434394392607e-05,
+      "loss": 1.0155,
+      "step": 20466
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7295414973882764e-05,
+      "loss": 1.025,
+      "step": 20467
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7286486003839458e-05,
+      "loss": 0.8749,
+      "step": 20468
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.727755703379615e-05,
+      "loss": 1.0786,
+      "step": 20469
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.726862806375285e-05,
+      "loss": 1.055,
+      "step": 20470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7259699093709543e-05,
+      "loss": 1.0949,
+      "step": 20471
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7250770123666237e-05,
+      "loss": 0.9907,
+      "step": 20472
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.724184115362293e-05,
+      "loss": 1.0149,
+      "step": 20473
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7232912183579624e-05,
+      "loss": 0.9195,
+      "step": 20474
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7223983213536318e-05,
+      "loss": 1.0829,
+      "step": 20475
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7215054243493015e-05,
+      "loss": 1.0635,
+      "step": 20476
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.720612527344971e-05,
+      "loss": 1.084,
+      "step": 20477
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7197196303406403e-05,
+      "loss": 0.9805,
+      "step": 20478
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7188267333363097e-05,
+      "loss": 0.8383,
+      "step": 20479
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.717933836331979e-05,
+      "loss": 0.9942,
+      "step": 20480
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7170409393276488e-05,
+      "loss": 1.0114,
+      "step": 20481
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7161480423233182e-05,
+      "loss": 1.0238,
+      "step": 20482
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7152551453189876e-05,
+      "loss": 1.0791,
+      "step": 20483
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.714362248314657e-05,
+      "loss": 1.0385,
+      "step": 20484
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7134693513103264e-05,
+      "loss": 1.1451,
+      "step": 20485
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7125764543059957e-05,
+      "loss": 1.0493,
+      "step": 20486
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7116835573016655e-05,
+      "loss": 0.9493,
+      "step": 20487
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.710790660297335e-05,
+      "loss": 0.9992,
+      "step": 20488
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7098977632930042e-05,
+      "loss": 0.882,
+      "step": 20489
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7090048662886736e-05,
+      "loss": 1.0562,
+      "step": 20490
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.708111969284343e-05,
+      "loss": 0.9058,
+      "step": 20491
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7072190722800127e-05,
+      "loss": 0.9778,
+      "step": 20492
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.706326175275682e-05,
+      "loss": 1.0704,
+      "step": 20493
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7054332782713515e-05,
+      "loss": 1.0864,
+      "step": 20494
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.704540381267021e-05,
+      "loss": 1.0125,
+      "step": 20495
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7036474842626903e-05,
+      "loss": 0.9736,
+      "step": 20496
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7027545872583597e-05,
+      "loss": 0.9732,
+      "step": 20497
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7018616902540294e-05,
+      "loss": 1.0412,
+      "step": 20498
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7009687932496988e-05,
+      "loss": 1.1057,
+      "step": 20499
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7000758962453682e-05,
+      "loss": 1.0609,
+      "step": 20500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6991829992410376e-05,
+      "loss": 0.9787,
+      "step": 20501
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.698290102236707e-05,
+      "loss": 1.1612,
+      "step": 20502
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6973972052323767e-05,
+      "loss": 1.1638,
+      "step": 20503
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.696504308228046e-05,
+      "loss": 1.0949,
+      "step": 20504
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6956114112237155e-05,
+      "loss": 1.0551,
+      "step": 20505
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.694718514219385e-05,
+      "loss": 1.1674,
+      "step": 20506
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6938256172150542e-05,
+      "loss": 1.0682,
+      "step": 20507
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6929327202107236e-05,
+      "loss": 0.9919,
+      "step": 20508
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6920398232063933e-05,
+      "loss": 1.0437,
+      "step": 20509
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6911469262020627e-05,
+      "loss": 0.9462,
+      "step": 20510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.690254029197732e-05,
+      "loss": 0.9562,
+      "step": 20511
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6893611321934015e-05,
+      "loss": 0.972,
+      "step": 20512
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.688468235189071e-05,
+      "loss": 1.0312,
+      "step": 20513
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6875753381847403e-05,
+      "loss": 1.097,
+      "step": 20514
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.68668244118041e-05,
+      "loss": 1.0525,
+      "step": 20515
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6857895441760794e-05,
+      "loss": 0.9119,
+      "step": 20516
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.684896647171749e-05,
+      "loss": 1.0516,
+      "step": 20517
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.684003750167418e-05,
+      "loss": 0.9079,
+      "step": 20518
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6831108531630875e-05,
+      "loss": 1.0088,
+      "step": 20519
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6822179561587573e-05,
+      "loss": 1.033,
+      "step": 20520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6813250591544267e-05,
+      "loss": 1.0773,
+      "step": 20521
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.680432162150096e-05,
+      "loss": 0.969,
+      "step": 20522
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6795392651457654e-05,
+      "loss": 1.0065,
+      "step": 20523
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6786463681414348e-05,
+      "loss": 0.9697,
+      "step": 20524
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6777534711371042e-05,
+      "loss": 1.0418,
+      "step": 20525
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.676860574132774e-05,
+      "loss": 1.0482,
+      "step": 20526
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6759676771284433e-05,
+      "loss": 1.0663,
+      "step": 20527
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.675074780124113e-05,
+      "loss": 1.0844,
+      "step": 20528
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.674181883119782e-05,
+      "loss": 1.061,
+      "step": 20529
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6732889861154515e-05,
+      "loss": 1.0022,
+      "step": 20530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6723960891111212e-05,
+      "loss": 1.1088,
+      "step": 20531
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6715031921067906e-05,
+      "loss": 0.9809,
+      "step": 20532
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.67061029510246e-05,
+      "loss": 0.9835,
+      "step": 20533
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6697173980981297e-05,
+      "loss": 1.0841,
+      "step": 20534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6688245010937988e-05,
+      "loss": 0.9246,
+      "step": 20535
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.667931604089468e-05,
+      "loss": 0.9585,
+      "step": 20536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.667038707085138e-05,
+      "loss": 1.0677,
+      "step": 20537
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6661458100808073e-05,
+      "loss": 1.1289,
+      "step": 20538
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6652529130764766e-05,
+      "loss": 0.9001,
+      "step": 20539
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6643600160721464e-05,
+      "loss": 0.9694,
+      "step": 20540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6634671190678154e-05,
+      "loss": 1.071,
+      "step": 20541
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.662574222063485e-05,
+      "loss": 1.1281,
+      "step": 20542
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6616813250591545e-05,
+      "loss": 1.0237,
+      "step": 20543
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.660788428054824e-05,
+      "loss": 0.9519,
+      "step": 20544
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6598955310504936e-05,
+      "loss": 0.9614,
+      "step": 20545
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6590026340461627e-05,
+      "loss": 0.9698,
+      "step": 20546
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.658109737041832e-05,
+      "loss": 0.9646,
+      "step": 20547
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6572168400375018e-05,
+      "loss": 1.0422,
+      "step": 20548
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6563239430331712e-05,
+      "loss": 1.1246,
+      "step": 20549
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6554310460288406e-05,
+      "loss": 1.0943,
+      "step": 20550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6545381490245103e-05,
+      "loss": 1.104,
+      "step": 20551
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6536452520201793e-05,
+      "loss": 1.0671,
+      "step": 20552
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.652752355015849e-05,
+      "loss": 1.2678,
+      "step": 20553
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6518594580115185e-05,
+      "loss": 1.0475,
+      "step": 20554
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.650966561007188e-05,
+      "loss": 1.0161,
+      "step": 20555
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6500736640028576e-05,
+      "loss": 1.0105,
+      "step": 20556
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.649180766998527e-05,
+      "loss": 1.0573,
+      "step": 20557
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.648287869994196e-05,
+      "loss": 1.0829,
+      "step": 20558
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6473949729898657e-05,
+      "loss": 0.9282,
+      "step": 20559
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.646502075985535e-05,
+      "loss": 1.1065,
+      "step": 20560
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6456091789812045e-05,
+      "loss": 1.147,
+      "step": 20561
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6447162819768742e-05,
+      "loss": 1.0599,
+      "step": 20562
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6438233849725436e-05,
+      "loss": 1.2009,
+      "step": 20563
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.642930487968213e-05,
+      "loss": 1.0628,
+      "step": 20564
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6420375909638824e-05,
+      "loss": 1.0468,
+      "step": 20565
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6411446939595518e-05,
+      "loss": 1.0365,
+      "step": 20566
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6402517969552215e-05,
+      "loss": 1.0332,
+      "step": 20567
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.639358899950891e-05,
+      "loss": 1.0316,
+      "step": 20568
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6384660029465603e-05,
+      "loss": 1.0637,
+      "step": 20569
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6375731059422297e-05,
+      "loss": 0.9966,
+      "step": 20570
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.636680208937899e-05,
+      "loss": 1.0397,
+      "step": 20571
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6357873119335684e-05,
+      "loss": 1.2055,
+      "step": 20572
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.634894414929238e-05,
+      "loss": 0.9486,
+      "step": 20573
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6340015179249075e-05,
+      "loss": 0.98,
+      "step": 20574
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6331086209205766e-05,
+      "loss": 0.91,
+      "step": 20575
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6322157239162463e-05,
+      "loss": 1.0223,
+      "step": 20576
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6313228269119157e-05,
+      "loss": 1.1221,
+      "step": 20577
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6304299299075854e-05,
+      "loss": 0.9394,
+      "step": 20578
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6295370329032548e-05,
+      "loss": 0.9977,
+      "step": 20579
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6286441358989242e-05,
+      "loss": 1.0834,
+      "step": 20580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6277512388945936e-05,
+      "loss": 1.066,
+      "step": 20581
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.626858341890263e-05,
+      "loss": 0.9843,
+      "step": 20582
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6259654448859324e-05,
+      "loss": 1.0307,
+      "step": 20583
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.625072547881602e-05,
+      "loss": 1.1587,
+      "step": 20584
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6241796508772715e-05,
+      "loss": 1.2074,
+      "step": 20585
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.623286753872941e-05,
+      "loss": 1.1522,
+      "step": 20586
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6223938568686103e-05,
+      "loss": 1.0419,
+      "step": 20587
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6215009598642796e-05,
+      "loss": 1.2098,
+      "step": 20588
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6206080628599494e-05,
+      "loss": 1.0257,
+      "step": 20589
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6197151658556188e-05,
+      "loss": 0.9526,
+      "step": 20590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.618822268851288e-05,
+      "loss": 0.9678,
+      "step": 20591
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6179293718469575e-05,
+      "loss": 1.1202,
+      "step": 20592
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.617036474842627e-05,
+      "loss": 0.9781,
+      "step": 20593
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6161435778382963e-05,
+      "loss": 1.0016,
+      "step": 20594
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.615250680833966e-05,
+      "loss": 1.0902,
+      "step": 20595
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6143577838296354e-05,
+      "loss": 1.0687,
+      "step": 20596
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6134648868253048e-05,
+      "loss": 1.0745,
+      "step": 20597
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6125719898209742e-05,
+      "loss": 0.9664,
+      "step": 20598
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6116790928166436e-05,
+      "loss": 1.0911,
+      "step": 20599
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.610786195812313e-05,
+      "loss": 1.0662,
+      "step": 20600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6098932988079827e-05,
+      "loss": 1.1676,
+      "step": 20601
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.609000401803652e-05,
+      "loss": 1.0283,
+      "step": 20602
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6081075047993215e-05,
+      "loss": 1.0672,
+      "step": 20603
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.607214607794991e-05,
+      "loss": 1.0405,
+      "step": 20604
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6063217107906602e-05,
+      "loss": 1.1394,
+      "step": 20605
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.60542881378633e-05,
+      "loss": 1.0037,
+      "step": 20606
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6045359167819993e-05,
+      "loss": 1.0329,
+      "step": 20607
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6036430197776687e-05,
+      "loss": 1.0728,
+      "step": 20608
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.602750122773338e-05,
+      "loss": 0.9421,
+      "step": 20609
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6018572257690075e-05,
+      "loss": 0.9349,
+      "step": 20610
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.600964328764677e-05,
+      "loss": 1.0718,
+      "step": 20611
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6000714317603466e-05,
+      "loss": 0.881,
+      "step": 20612
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.599178534756016e-05,
+      "loss": 1.1207,
+      "step": 20613
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5982856377516854e-05,
+      "loss": 1.1135,
+      "step": 20614
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5973927407473548e-05,
+      "loss": 0.9205,
+      "step": 20615
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.596499843743024e-05,
+      "loss": 0.998,
+      "step": 20616
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.595606946738694e-05,
+      "loss": 0.9396,
+      "step": 20617
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5947140497343633e-05,
+      "loss": 1.029,
+      "step": 20618
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5938211527300327e-05,
+      "loss": 1.0027,
+      "step": 20619
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.592928255725702e-05,
+      "loss": 1.118,
+      "step": 20620
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5920353587213714e-05,
+      "loss": 0.895,
+      "step": 20621
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5911424617170408e-05,
+      "loss": 1.064,
+      "step": 20622
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5902495647127106e-05,
+      "loss": 1.021,
+      "step": 20623
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.58935666770838e-05,
+      "loss": 0.9451,
+      "step": 20624
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5884637707040493e-05,
+      "loss": 0.8808,
+      "step": 20625
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5875708736997187e-05,
+      "loss": 0.9309,
+      "step": 20626
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.586677976695388e-05,
+      "loss": 1.3074,
+      "step": 20627
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5857850796910578e-05,
+      "loss": 1.0917,
+      "step": 20628
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5848921826867272e-05,
+      "loss": 0.9385,
+      "step": 20629
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5839992856823966e-05,
+      "loss": 1.0963,
+      "step": 20630
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.583106388678066e-05,
+      "loss": 0.9871,
+      "step": 20631
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5822134916737354e-05,
+      "loss": 0.9985,
+      "step": 20632
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5813205946694048e-05,
+      "loss": 0.9867,
+      "step": 20633
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5804276976650745e-05,
+      "loss": 1.0163,
+      "step": 20634
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.579534800660744e-05,
+      "loss": 1.0398,
+      "step": 20635
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5786419036564133e-05,
+      "loss": 1.0179,
+      "step": 20636
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5777490066520826e-05,
+      "loss": 1.07,
+      "step": 20637
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.576856109647752e-05,
+      "loss": 1.175,
+      "step": 20638
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5759632126434218e-05,
+      "loss": 1.2251,
+      "step": 20639
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.575070315639091e-05,
+      "loss": 1.0999,
+      "step": 20640
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5741774186347605e-05,
+      "loss": 1.2344,
+      "step": 20641
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5732845216304303e-05,
+      "loss": 0.9304,
+      "step": 20642
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5723916246260993e-05,
+      "loss": 1.0103,
+      "step": 20643
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5714987276217687e-05,
+      "loss": 1.0511,
+      "step": 20644
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5706058306174384e-05,
+      "loss": 0.9557,
+      "step": 20645
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5697129336131078e-05,
+      "loss": 1.0194,
+      "step": 20646
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5688200366087772e-05,
+      "loss": 1.0633,
+      "step": 20647
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.567927139604447e-05,
+      "loss": 1.012,
+      "step": 20648
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.567034242600116e-05,
+      "loss": 0.9707,
+      "step": 20649
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5661413455957857e-05,
+      "loss": 1.0158,
+      "step": 20650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.565248448591455e-05,
+      "loss": 0.9608,
+      "step": 20651
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5643555515871245e-05,
+      "loss": 0.9893,
+      "step": 20652
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5634626545827942e-05,
+      "loss": 1.0171,
+      "step": 20653
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5625697575784636e-05,
+      "loss": 1.0357,
+      "step": 20654
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5616768605741326e-05,
+      "loss": 1.0897,
+      "step": 20655
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5607839635698024e-05,
+      "loss": 1.0978,
+      "step": 20656
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5598910665654717e-05,
+      "loss": 1.2585,
+      "step": 20657
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.558998169561141e-05,
+      "loss": 1.0509,
+      "step": 20658
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.558105272556811e-05,
+      "loss": 0.9489,
+      "step": 20659
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.55721237555248e-05,
+      "loss": 0.9594,
+      "step": 20660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5563194785481493e-05,
+      "loss": 1.0246,
+      "step": 20661
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.555426581543819e-05,
+      "loss": 0.9138,
+      "step": 20662
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5545336845394884e-05,
+      "loss": 1.0459,
+      "step": 20663
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.553640787535158e-05,
+      "loss": 1.0533,
+      "step": 20664
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5527478905308275e-05,
+      "loss": 1.217,
+      "step": 20665
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5518549935264966e-05,
+      "loss": 1.0756,
+      "step": 20666
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5509620965221663e-05,
+      "loss": 0.9977,
+      "step": 20667
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5500691995178357e-05,
+      "loss": 1.0467,
+      "step": 20668
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.549176302513505e-05,
+      "loss": 1.065,
+      "step": 20669
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5482834055091748e-05,
+      "loss": 1.0796,
+      "step": 20670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5473905085048442e-05,
+      "loss": 0.9425,
+      "step": 20671
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5464976115005132e-05,
+      "loss": 1.1064,
+      "step": 20672
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.545604714496183e-05,
+      "loss": 1.0267,
+      "step": 20673
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5447118174918523e-05,
+      "loss": 1.0659,
+      "step": 20674
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.543818920487522e-05,
+      "loss": 1.0746,
+      "step": 20675
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5429260234831914e-05,
+      "loss": 1.0321,
+      "step": 20676
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.542033126478861e-05,
+      "loss": 1.0091,
+      "step": 20677
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5411402294745302e-05,
+      "loss": 0.9383,
+      "step": 20678
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5402473324701996e-05,
+      "loss": 0.9787,
+      "step": 20679
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.539354435465869e-05,
+      "loss": 0.991,
+      "step": 20680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.0408,
+      "step": 20681
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.537568641457208e-05,
+      "loss": 0.8802,
+      "step": 20682
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.536675744452877e-05,
+      "loss": 1.0256,
+      "step": 20683
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.535782847448547e-05,
+      "loss": 0.9624,
+      "step": 20684
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5348899504442163e-05,
+      "loss": 0.9926,
+      "step": 20685
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5339970534398857e-05,
+      "loss": 1.0805,
+      "step": 20686
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5331041564355554e-05,
+      "loss": 1.0772,
+      "step": 20687
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5322112594312248e-05,
+      "loss": 0.8644,
+      "step": 20688
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.531318362426894e-05,
+      "loss": 0.9987,
+      "step": 20689
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5304254654225635e-05,
+      "loss": 0.9331,
+      "step": 20690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.529532568418233e-05,
+      "loss": 1.0652,
+      "step": 20691
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5286396714139027e-05,
+      "loss": 0.9337,
+      "step": 20692
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.527746774409572e-05,
+      "loss": 1.0361,
+      "step": 20693
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5268538774052414e-05,
+      "loss": 1.0633,
+      "step": 20694
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5259609804009108e-05,
+      "loss": 0.9935,
+      "step": 20695
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5250680833965802e-05,
+      "loss": 1.0866,
+      "step": 20696
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5241751863922496e-05,
+      "loss": 1.1423,
+      "step": 20697
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5232822893879191e-05,
+      "loss": 0.9884,
+      "step": 20698
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5223893923835885e-05,
+      "loss": 1.0971,
+      "step": 20699
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5214964953792583e-05,
+      "loss": 0.967,
+      "step": 20700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5206035983749275e-05,
+      "loss": 1.1698,
+      "step": 20701
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5197107013705969e-05,
+      "loss": 0.9578,
+      "step": 20702
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5188178043662666e-05,
+      "loss": 0.9188,
+      "step": 20703
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5179249073619358e-05,
+      "loss": 0.9552,
+      "step": 20704
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5170320103576052e-05,
+      "loss": 0.9908,
+      "step": 20705
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5161391133532749e-05,
+      "loss": 1.0406,
+      "step": 20706
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5152462163489441e-05,
+      "loss": 1.1363,
+      "step": 20707
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5143533193446135e-05,
+      "loss": 0.9718,
+      "step": 20708
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5134604223402832e-05,
+      "loss": 0.9768,
+      "step": 20709
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5125675253359525e-05,
+      "loss": 1.0765,
+      "step": 20710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5116746283316222e-05,
+      "loss": 1.1163,
+      "step": 20711
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5107817313272916e-05,
+      "loss": 0.919,
+      "step": 20712
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5098888343229608e-05,
+      "loss": 0.978,
+      "step": 20713
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5089959373186305e-05,
+      "loss": 0.8265,
+      "step": 20714
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5081030403142999e-05,
+      "loss": 1.0157,
+      "step": 20715
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5072101433099691e-05,
+      "loss": 1.0052,
+      "step": 20716
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5063172463056388e-05,
+      "loss": 1.135,
+      "step": 20717
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5054243493013082e-05,
+      "loss": 1.1079,
+      "step": 20718
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5045314522969775e-05,
+      "loss": 0.9982,
+      "step": 20719
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5036385552926472e-05,
+      "loss": 1.0613,
+      "step": 20720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5027456582883166e-05,
+      "loss": 1.0275,
+      "step": 20721
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5018527612839858e-05,
+      "loss": 1.1214,
+      "step": 20722
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5009598642796555e-05,
+      "loss": 0.9972,
+      "step": 20723
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5000669672753247e-05,
+      "loss": 0.9228,
+      "step": 20724
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4991740702709944e-05,
+      "loss": 1.011,
+      "step": 20725
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4982811732666638e-05,
+      "loss": 1.0301,
+      "step": 20726
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.497388276262333e-05,
+      "loss": 1.1384,
+      "step": 20727
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4964953792580028e-05,
+      "loss": 0.8911,
+      "step": 20728
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4956024822536722e-05,
+      "loss": 0.9908,
+      "step": 20729
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4947095852493414e-05,
+      "loss": 0.9314,
+      "step": 20730
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4938166882450111e-05,
+      "loss": 0.9551,
+      "step": 20731
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4929237912406805e-05,
+      "loss": 1.0716,
+      "step": 20732
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4920308942363497e-05,
+      "loss": 1.125,
+      "step": 20733
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4911379972320194e-05,
+      "loss": 1.0489,
+      "step": 20734
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4902451002276888e-05,
+      "loss": 1.0006,
+      "step": 20735
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4893522032233584e-05,
+      "loss": 0.9931,
+      "step": 20736
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4884593062190278e-05,
+      "loss": 1.0514,
+      "step": 20737
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4875664092146972e-05,
+      "loss": 0.9712,
+      "step": 20738
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4866735122103667e-05,
+      "loss": 1.0338,
+      "step": 20739
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4857806152060361e-05,
+      "loss": 1.0237,
+      "step": 20740
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4848877182017055e-05,
+      "loss": 0.9632,
+      "step": 20741
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.483994821197375e-05,
+      "loss": 1.0775,
+      "step": 20742
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4831019241930444e-05,
+      "loss": 1.2522,
+      "step": 20743
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4822090271887138e-05,
+      "loss": 1.0466,
+      "step": 20744
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4813161301843834e-05,
+      "loss": 0.9231,
+      "step": 20745
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4804232331800528e-05,
+      "loss": 1.1406,
+      "step": 20746
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4795303361757221e-05,
+      "loss": 0.9447,
+      "step": 20747
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4786374391713917e-05,
+      "loss": 1.0489,
+      "step": 20748
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4777445421670611e-05,
+      "loss": 1.0143,
+      "step": 20749
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4768516451627306e-05,
+      "loss": 1.1107,
+      "step": 20750
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4759587481584e-05,
+      "loss": 1.0541,
+      "step": 20751
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4750658511540694e-05,
+      "loss": 0.8806,
+      "step": 20752
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.474172954149739e-05,
+      "loss": 1.099,
+      "step": 20753
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4732800571454084e-05,
+      "loss": 0.9516,
+      "step": 20754
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4723871601410777e-05,
+      "loss": 0.9368,
+      "step": 20755
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4714942631367473e-05,
+      "loss": 1.0528,
+      "step": 20756
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4706013661324167e-05,
+      "loss": 1.0679,
+      "step": 20757
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.469708469128086e-05,
+      "loss": 1.1317,
+      "step": 20758
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4688155721237556e-05,
+      "loss": 1.0326,
+      "step": 20759
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.467922675119425e-05,
+      "loss": 0.9815,
+      "step": 20760
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4670297781150946e-05,
+      "loss": 1.1536,
+      "step": 20761
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.466136881110764e-05,
+      "loss": 1.0126,
+      "step": 20762
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4652439841064334e-05,
+      "loss": 0.9702,
+      "step": 20763
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4643510871021029e-05,
+      "loss": 0.9545,
+      "step": 20764
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4634581900977723e-05,
+      "loss": 1.0203,
+      "step": 20765
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4625652930934417e-05,
+      "loss": 1.0316,
+      "step": 20766
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4616723960891112e-05,
+      "loss": 1.0986,
+      "step": 20767
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4607794990847806e-05,
+      "loss": 0.9384,
+      "step": 20768
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.45988660208045e-05,
+      "loss": 0.9225,
+      "step": 20769
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4589937050761196e-05,
+      "loss": 0.9492,
+      "step": 20770
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.458100808071789e-05,
+      "loss": 1.0734,
+      "step": 20771
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4572079110674585e-05,
+      "loss": 0.9145,
+      "step": 20772
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4563150140631279e-05,
+      "loss": 1.1141,
+      "step": 20773
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4554221170587973e-05,
+      "loss": 1.1455,
+      "step": 20774
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4545292200544668e-05,
+      "loss": 1.0253,
+      "step": 20775
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4536363230501362e-05,
+      "loss": 0.907,
+      "step": 20776
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4527434260458056e-05,
+      "loss": 0.9902,
+      "step": 20777
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4518505290414752e-05,
+      "loss": 1.0113,
+      "step": 20778
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4509576320371446e-05,
+      "loss": 1.0191,
+      "step": 20779
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.450064735032814e-05,
+      "loss": 1.1156,
+      "step": 20780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4491718380284835e-05,
+      "loss": 1.0937,
+      "step": 20781
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4482789410241529e-05,
+      "loss": 1.0818,
+      "step": 20782
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4473860440198223e-05,
+      "loss": 1.163,
+      "step": 20783
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4464931470154918e-05,
+      "loss": 1.0099,
+      "step": 20784
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4456002500111612e-05,
+      "loss": 1.085,
+      "step": 20785
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4447073530068308e-05,
+      "loss": 0.9483,
+      "step": 20786
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4438144560025002e-05,
+      "loss": 1.0737,
+      "step": 20787
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4429215589981695e-05,
+      "loss": 0.9711,
+      "step": 20788
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4420286619938391e-05,
+      "loss": 1.0339,
+      "step": 20789
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4411357649895085e-05,
+      "loss": 1.0376,
+      "step": 20790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4402428679851779e-05,
+      "loss": 0.9068,
+      "step": 20791
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4393499709808474e-05,
+      "loss": 0.9847,
+      "step": 20792
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4384570739765168e-05,
+      "loss": 1.118,
+      "step": 20793
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4375641769721862e-05,
+      "loss": 1.0244,
+      "step": 20794
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4366712799678558e-05,
+      "loss": 0.8766,
+      "step": 20795
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4357783829635252e-05,
+      "loss": 0.9376,
+      "step": 20796
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4348854859591949e-05,
+      "loss": 1.0051,
+      "step": 20797
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4339925889548641e-05,
+      "loss": 1.07,
+      "step": 20798
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4330996919505335e-05,
+      "loss": 1.0659,
+      "step": 20799
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4322067949462032e-05,
+      "loss": 1.0486,
+      "step": 20800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4313138979418724e-05,
+      "loss": 1.0502,
+      "step": 20801
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4304210009375418e-05,
+      "loss": 1.1471,
+      "step": 20802
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4295281039332115e-05,
+      "loss": 1.0197,
+      "step": 20803
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4286352069288808e-05,
+      "loss": 1.0555,
+      "step": 20804
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4277423099245501e-05,
+      "loss": 1.1055,
+      "step": 20805
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4268494129202199e-05,
+      "loss": 1.1285,
+      "step": 20806
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.425956515915889e-05,
+      "loss": 1.1263,
+      "step": 20807
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4250636189115585e-05,
+      "loss": 1.0498,
+      "step": 20808
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.424170721907228e-05,
+      "loss": 1.0077,
+      "step": 20809
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4232778249028974e-05,
+      "loss": 1.222,
+      "step": 20810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4223849278985671e-05,
+      "loss": 1.0488,
+      "step": 20811
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4214920308942364e-05,
+      "loss": 1.0967,
+      "step": 20812
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4205991338899057e-05,
+      "loss": 1.2169,
+      "step": 20813
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4197062368855755e-05,
+      "loss": 0.9338,
+      "step": 20814
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4188133398812447e-05,
+      "loss": 1.1143,
+      "step": 20815
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.417920442876914e-05,
+      "loss": 1.2414,
+      "step": 20816
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4170275458725838e-05,
+      "loss": 1.0527,
+      "step": 20817
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.416134648868253e-05,
+      "loss": 1.019,
+      "step": 20818
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4152417518639224e-05,
+      "loss": 1.0179,
+      "step": 20819
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4143488548595921e-05,
+      "loss": 1.2093,
+      "step": 20820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4134559578552613e-05,
+      "loss": 1.0165,
+      "step": 20821
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.412563060850931e-05,
+      "loss": 1.1434,
+      "step": 20822
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4116701638466005e-05,
+      "loss": 1.0142,
+      "step": 20823
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4107772668422697e-05,
+      "loss": 0.882,
+      "step": 20824
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4098843698379394e-05,
+      "loss": 1.0763,
+      "step": 20825
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4089914728336088e-05,
+      "loss": 1.0696,
+      "step": 20826
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.408098575829278e-05,
+      "loss": 1.037,
+      "step": 20827
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4072056788249477e-05,
+      "loss": 0.9404,
+      "step": 20828
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4063127818206171e-05,
+      "loss": 0.9996,
+      "step": 20829
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4054198848162863e-05,
+      "loss": 1.0113,
+      "step": 20830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.404526987811956e-05,
+      "loss": 1.1148,
+      "step": 20831
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4036340908076254e-05,
+      "loss": 1.1326,
+      "step": 20832
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.402741193803295e-05,
+      "loss": 1.0054,
+      "step": 20833
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4018482967989644e-05,
+      "loss": 1.0248,
+      "step": 20834
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4009553997946336e-05,
+      "loss": 0.97,
+      "step": 20835
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4000625027903033e-05,
+      "loss": 1.0619,
+      "step": 20836
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3991696057859727e-05,
+      "loss": 0.912,
+      "step": 20837
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.398276708781642e-05,
+      "loss": 0.9534,
+      "step": 20838
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3973838117773117e-05,
+      "loss": 0.8668,
+      "step": 20839
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.396490914772981e-05,
+      "loss": 1.0453,
+      "step": 20840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3955980177686503e-05,
+      "loss": 1.1096,
+      "step": 20841
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.39470512076432e-05,
+      "loss": 0.977,
+      "step": 20842
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3938122237599894e-05,
+      "loss": 1.0848,
+      "step": 20843
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3929193267556586e-05,
+      "loss": 1.1036,
+      "step": 20844
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3920264297513283e-05,
+      "loss": 1.0914,
+      "step": 20845
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3911335327469977e-05,
+      "loss": 0.9557,
+      "step": 20846
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3902406357426673e-05,
+      "loss": 0.987,
+      "step": 20847
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3893477387383367e-05,
+      "loss": 1.0445,
+      "step": 20848
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.388454841734006e-05,
+      "loss": 1.0339,
+      "step": 20849
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3875619447296756e-05,
+      "loss": 1.5801,
+      "step": 20850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.386669047725345e-05,
+      "loss": 0.962,
+      "step": 20851
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3857761507210144e-05,
+      "loss": 0.9935,
+      "step": 20852
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.384883253716684e-05,
+      "loss": 0.9813,
+      "step": 20853
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3839903567123533e-05,
+      "loss": 1.1448,
+      "step": 20854
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3830974597080227e-05,
+      "loss": 1.0167,
+      "step": 20855
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3822045627036923e-05,
+      "loss": 1.0498,
+      "step": 20856
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3813116656993616e-05,
+      "loss": 0.9801,
+      "step": 20857
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3804187686950312e-05,
+      "loss": 1.0076,
+      "step": 20858
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3795258716907006e-05,
+      "loss": 1.062,
+      "step": 20859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.37863297468637e-05,
+      "loss": 0.9463,
+      "step": 20860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3777400776820395e-05,
+      "loss": 0.9789,
+      "step": 20861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.376847180677709e-05,
+      "loss": 1.3406,
+      "step": 20862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3759542836733783e-05,
+      "loss": 1.2548,
+      "step": 20863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3750613866690479e-05,
+      "loss": 0.9732,
+      "step": 20864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3741684896647172e-05,
+      "loss": 0.9295,
+      "step": 20865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3732755926603866e-05,
+      "loss": 1.0155,
+      "step": 20866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3723826956560562e-05,
+      "loss": 0.9735,
+      "step": 20867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3714897986517256e-05,
+      "loss": 0.9209,
+      "step": 20868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.370596901647395e-05,
+      "loss": 1.0177,
+      "step": 20869
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3697040046430645e-05,
+      "loss": 0.9554,
+      "step": 20870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3688111076387339e-05,
+      "loss": 1.0836,
+      "step": 20871
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3679182106344035e-05,
+      "loss": 1.1042,
+      "step": 20872
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3670253136300729e-05,
+      "loss": 0.9636,
+      "step": 20873
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3661324166257422e-05,
+      "loss": 0.9904,
+      "step": 20874
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3652395196214118e-05,
+      "loss": 1.0754,
+      "step": 20875
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3643466226170812e-05,
+      "loss": 1.1161,
+      "step": 20876
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3634537256127506e-05,
+      "loss": 1.0472,
+      "step": 20877
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3625608286084201e-05,
+      "loss": 1.0324,
+      "step": 20878
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3616679316040895e-05,
+      "loss": 0.8881,
+      "step": 20879
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3607750345997589e-05,
+      "loss": 0.9729,
+      "step": 20880
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3598821375954285e-05,
+      "loss": 0.9772,
+      "step": 20881
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3589892405910978e-05,
+      "loss": 0.9682,
+      "step": 20882
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3580963435867674e-05,
+      "loss": 1.1207,
+      "step": 20883
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3572034465824368e-05,
+      "loss": 1.1526,
+      "step": 20884
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3563105495781062e-05,
+      "loss": 0.9979,
+      "step": 20885
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3554176525737757e-05,
+      "loss": 0.9241,
+      "step": 20886
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3545247555694451e-05,
+      "loss": 1.0784,
+      "step": 20887
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3536318585651145e-05,
+      "loss": 0.9938,
+      "step": 20888
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.352738961560784e-05,
+      "loss": 0.9556,
+      "step": 20889
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3518460645564534e-05,
+      "loss": 0.9662,
+      "step": 20890
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3509531675521228e-05,
+      "loss": 1.0754,
+      "step": 20891
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3500602705477924e-05,
+      "loss": 1.0005,
+      "step": 20892
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3491673735434618e-05,
+      "loss": 0.975,
+      "step": 20893
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3482744765391312e-05,
+      "loss": 0.978,
+      "step": 20894
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3473815795348007e-05,
+      "loss": 0.9847,
+      "step": 20895
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3464886825304701e-05,
+      "loss": 1.0364,
+      "step": 20896
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3455957855261397e-05,
+      "loss": 1.0153,
+      "step": 20897
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.344702888521809e-05,
+      "loss": 1.1693,
+      "step": 20898
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3438099915174784e-05,
+      "loss": 0.9792,
+      "step": 20899
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.342917094513148e-05,
+      "loss": 1.1486,
+      "step": 20900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3420241975088174e-05,
+      "loss": 0.9262,
+      "step": 20901
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3411313005044868e-05,
+      "loss": 1.027,
+      "step": 20902
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3402384035001563e-05,
+      "loss": 1.0496,
+      "step": 20903
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3393455064958257e-05,
+      "loss": 1.0305,
+      "step": 20904
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3384526094914951e-05,
+      "loss": 0.9817,
+      "step": 20905
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3375597124871646e-05,
+      "loss": 1.0031,
+      "step": 20906
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.336666815482834e-05,
+      "loss": 1.0214,
+      "step": 20907
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3357739184785038e-05,
+      "loss": 0.9605,
+      "step": 20908
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.334881021474173e-05,
+      "loss": 0.9896,
+      "step": 20909
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3339881244698424e-05,
+      "loss": 0.9969,
+      "step": 20910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3330952274655121e-05,
+      "loss": 0.8951,
+      "step": 20911
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3322023304611813e-05,
+      "loss": 0.9909,
+      "step": 20912
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3313094334568507e-05,
+      "loss": 1.1163,
+      "step": 20913
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3304165364525204e-05,
+      "loss": 1.0262,
+      "step": 20914
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3295236394481896e-05,
+      "loss": 1.1032,
+      "step": 20915
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.328630742443859e-05,
+      "loss": 0.9856,
+      "step": 20916
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3277378454395288e-05,
+      "loss": 1.0372,
+      "step": 20917
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.326844948435198e-05,
+      "loss": 0.9673,
+      "step": 20918
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3259520514308677e-05,
+      "loss": 1.0205,
+      "step": 20919
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3250591544265369e-05,
+      "loss": 1.1286,
+      "step": 20920
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3241662574222063e-05,
+      "loss": 1.0158,
+      "step": 20921
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.323273360417876e-05,
+      "loss": 0.9533,
+      "step": 20922
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3223804634135452e-05,
+      "loss": 1.0885,
+      "step": 20923
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3214875664092146e-05,
+      "loss": 0.989,
+      "step": 20924
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3205946694048844e-05,
+      "loss": 1.0263,
+      "step": 20925
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3197017724005536e-05,
+      "loss": 1.0849,
+      "step": 20926
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.318808875396223e-05,
+      "loss": 1.046,
+      "step": 20927
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3179159783918927e-05,
+      "loss": 0.9472,
+      "step": 20928
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3170230813875619e-05,
+      "loss": 1.0015,
+      "step": 20929
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3161301843832313e-05,
+      "loss": 0.9986,
+      "step": 20930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.315237287378901e-05,
+      "loss": 1.0845,
+      "step": 20931
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3143443903745702e-05,
+      "loss": 0.941,
+      "step": 20932
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.31345149337024e-05,
+      "loss": 1.1124,
+      "step": 20933
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3125585963659093e-05,
+      "loss": 1.1546,
+      "step": 20934
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3116656993615786e-05,
+      "loss": 0.9835,
+      "step": 20935
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3107728023572483e-05,
+      "loss": 1.0225,
+      "step": 20936
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3098799053529177e-05,
+      "loss": 1.0995,
+      "step": 20937
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3089870083485869e-05,
+      "loss": 0.9665,
+      "step": 20938
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3080941113442566e-05,
+      "loss": 1.086,
+      "step": 20939
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.307201214339926e-05,
+      "loss": 1.0115,
+      "step": 20940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3063083173355952e-05,
+      "loss": 0.97,
+      "step": 20941
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.305415420331265e-05,
+      "loss": 1.0039,
+      "step": 20942
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3045225233269343e-05,
+      "loss": 0.9958,
+      "step": 20943
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3036296263226039e-05,
+      "loss": 0.9142,
+      "step": 20944
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3027367293182733e-05,
+      "loss": 1.1137,
+      "step": 20945
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3018438323139425e-05,
+      "loss": 1.0847,
+      "step": 20946
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3009509353096122e-05,
+      "loss": 0.9373,
+      "step": 20947
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3000580383052816e-05,
+      "loss": 0.9028,
+      "step": 20948
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.2991651413009508e-05,
+      "loss": 0.9633,
+      "step": 20949
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2982722442966206e-05,
+      "loss": 0.9616,
+      "step": 20950
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.29737934729229e-05,
+      "loss": 1.0499,
+      "step": 20951
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2964864502879592e-05,
+      "loss": 1.0844,
+      "step": 20952
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2955935532836289e-05,
+      "loss": 0.9232,
+      "step": 20953
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2947006562792983e-05,
+      "loss": 0.9534,
+      "step": 20954
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2938077592749675e-05,
+      "loss": 1.0964,
+      "step": 20955
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2929148622706372e-05,
+      "loss": 0.9148,
+      "step": 20956
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2920219652663066e-05,
+      "loss": 1.0317,
+      "step": 20957
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2911290682619762e-05,
+      "loss": 0.9896,
+      "step": 20958
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2902361712576455e-05,
+      "loss": 1.0201,
+      "step": 20959
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.289343274253315e-05,
+      "loss": 1.1774,
+      "step": 20960
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2884503772489845e-05,
+      "loss": 1.1147,
+      "step": 20961
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2875574802446539e-05,
+      "loss": 0.9178,
+      "step": 20962
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2866645832403233e-05,
+      "loss": 1.0381,
+      "step": 20963
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2857716862359928e-05,
+      "loss": 0.9633,
+      "step": 20964
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2848787892316622e-05,
+      "loss": 1.0604,
+      "step": 20965
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2839858922273316e-05,
+      "loss": 1.0087,
+      "step": 20966
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2830929952230011e-05,
+      "loss": 1.071,
+      "step": 20967
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2822000982186705e-05,
+      "loss": 0.9806,
+      "step": 20968
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2813072012143401e-05,
+      "loss": 0.9977,
+      "step": 20969
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2804143042100095e-05,
+      "loss": 0.9574,
+      "step": 20970
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2795214072056789e-05,
+      "loss": 1.0426,
+      "step": 20971
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2786285102013484e-05,
+      "loss": 1.0281,
+      "step": 20972
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2777356131970178e-05,
+      "loss": 0.9884,
+      "step": 20973
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2768427161926872e-05,
+      "loss": 1.1354,
+      "step": 20974
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2759498191883567e-05,
+      "loss": 0.9426,
+      "step": 20975
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2750569221840261e-05,
+      "loss": 1.081,
+      "step": 20976
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2741640251796955e-05,
+      "loss": 0.9168,
+      "step": 20977
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.273271128175365e-05,
+      "loss": 1.005,
+      "step": 20978
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2723782311710345e-05,
+      "loss": 0.9673,
+      "step": 20979
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.271485334166704e-05,
+      "loss": 0.9891,
+      "step": 20980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2705924371623734e-05,
+      "loss": 0.9566,
+      "step": 20981
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2696995401580428e-05,
+      "loss": 1.0718,
+      "step": 20982
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2688066431537124e-05,
+      "loss": 1.0995,
+      "step": 20983
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2679137461493817e-05,
+      "loss": 0.9405,
+      "step": 20984
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2670208491450511e-05,
+      "loss": 0.953,
+      "step": 20985
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2661279521407207e-05,
+      "loss": 1.1664,
+      "step": 20986
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.26523505513639e-05,
+      "loss": 0.9227,
+      "step": 20987
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2643421581320595e-05,
+      "loss": 1.1467,
+      "step": 20988
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.263449261127729e-05,
+      "loss": 0.9216,
+      "step": 20989
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2625563641233984e-05,
+      "loss": 0.9932,
+      "step": 20990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2616634671190678e-05,
+      "loss": 1.0141,
+      "step": 20991
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2607705701147373e-05,
+      "loss": 0.9946,
+      "step": 20992
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2598776731104067e-05,
+      "loss": 0.9626,
+      "step": 20993
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2589847761060763e-05,
+      "loss": 1.0831,
+      "step": 20994
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2580918791017457e-05,
+      "loss": 1.0171,
+      "step": 20995
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.257198982097415e-05,
+      "loss": 1.0526,
+      "step": 20996
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2563060850930846e-05,
+      "loss": 1.1036,
+      "step": 20997
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.255413188088754e-05,
+      "loss": 0.9989,
+      "step": 20998
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2545202910844234e-05,
+      "loss": 1.0628,
+      "step": 20999
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.253627394080093e-05,
+      "loss": 1.1176,
+      "step": 21000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2527344970757623e-05,
+      "loss": 0.911,
+      "step": 21001
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2518416000714317e-05,
+      "loss": 1.0131,
+      "step": 21002
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2509487030671013e-05,
+      "loss": 1.0189,
+      "step": 21003
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2500558060627707e-05,
+      "loss": 1.1084,
+      "step": 21004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2491629090584402e-05,
+      "loss": 1.1381,
+      "step": 21005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2482700120541096e-05,
+      "loss": 1.018,
+      "step": 21006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.247377115049779e-05,
+      "loss": 1.1128,
+      "step": 21007
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2464842180454485e-05,
+      "loss": 1.0064,
+      "step": 21008
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.245591321041118e-05,
+      "loss": 0.9957,
+      "step": 21009
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2446984240367875e-05,
+      "loss": 1.0986,
+      "step": 21010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2438055270324569e-05,
+      "loss": 1.045,
+      "step": 21011
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2429126300281263e-05,
+      "loss": 1.1646,
+      "step": 21012
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2420197330237958e-05,
+      "loss": 1.1509,
+      "step": 21013
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2411268360194652e-05,
+      "loss": 1.2339,
+      "step": 21014
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2402339390151346e-05,
+      "loss": 1.036,
+      "step": 21015
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2393410420108041e-05,
+      "loss": 0.9157,
+      "step": 21016
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2384481450064735e-05,
+      "loss": 0.9592,
+      "step": 21017
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.237555248002143e-05,
+      "loss": 0.9531,
+      "step": 21018
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2366623509978125e-05,
+      "loss": 0.9634,
+      "step": 21019
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2357694539934819e-05,
+      "loss": 0.8767,
+      "step": 21020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2348765569891514e-05,
+      "loss": 1.0818,
+      "step": 21021
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2339836599848208e-05,
+      "loss": 1.1869,
+      "step": 21022
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2330907629804902e-05,
+      "loss": 0.9552,
+      "step": 21023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2321978659761598e-05,
+      "loss": 1.1758,
+      "step": 21024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2313049689718291e-05,
+      "loss": 0.9961,
+      "step": 21025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2304120719674985e-05,
+      "loss": 1.0282,
+      "step": 21026
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.229519174963168e-05,
+      "loss": 1.0038,
+      "step": 21027
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2286262779588376e-05,
+      "loss": 0.8241,
+      "step": 21028
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2277333809545069e-05,
+      "loss": 0.9225,
+      "step": 21029
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2268404839501764e-05,
+      "loss": 1.0064,
+      "step": 21030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2259475869458458e-05,
+      "loss": 1.2035,
+      "step": 21031
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2250546899415154e-05,
+      "loss": 1.1674,
+      "step": 21032
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2241617929371847e-05,
+      "loss": 0.9383,
+      "step": 21033
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2232688959328541e-05,
+      "loss": 0.9508,
+      "step": 21034
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2223759989285237e-05,
+      "loss": 1.0154,
+      "step": 21035
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.221483101924193e-05,
+      "loss": 1.0111,
+      "step": 21036
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2205902049198625e-05,
+      "loss": 0.9493,
+      "step": 21037
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.219697307915532e-05,
+      "loss": 1.028,
+      "step": 21038
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2188044109112016e-05,
+      "loss": 0.9918,
+      "step": 21039
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2179115139068708e-05,
+      "loss": 0.9595,
+      "step": 21040
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2170186169025403e-05,
+      "loss": 1.0603,
+      "step": 21041
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2161257198982099e-05,
+      "loss": 1.1122,
+      "step": 21042
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2152328228938791e-05,
+      "loss": 1.1165,
+      "step": 21043
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2143399258895487e-05,
+      "loss": 0.9636,
+      "step": 21044
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2134470288852182e-05,
+      "loss": 0.9311,
+      "step": 21045
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2125541318808876e-05,
+      "loss": 0.9882,
+      "step": 21046
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.211661234876557e-05,
+      "loss": 1.0322,
+      "step": 21047
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2107683378722266e-05,
+      "loss": 0.9392,
+      "step": 21048
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.209875440867896e-05,
+      "loss": 0.9578,
+      "step": 21049
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2089825438635655e-05,
+      "loss": 0.9829,
+      "step": 21050
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2080896468592349e-05,
+      "loss": 1.1518,
+      "step": 21051
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2071967498549043e-05,
+      "loss": 1.0046,
+      "step": 21052
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2063038528505738e-05,
+      "loss": 1.0996,
+      "step": 21053
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2054109558462432e-05,
+      "loss": 1.1094,
+      "step": 21054
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2045180588419126e-05,
+      "loss": 1.1724,
+      "step": 21055
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2036251618375822e-05,
+      "loss": 0.89,
+      "step": 21056
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2027322648332516e-05,
+      "loss": 1.0843,
+      "step": 21057
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.201839367828921e-05,
+      "loss": 1.0777,
+      "step": 21058
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2009464708245905e-05,
+      "loss": 0.9791,
+      "step": 21059
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2000535738202599e-05,
+      "loss": 1.0668,
+      "step": 21060
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1991606768159293e-05,
+      "loss": 0.9746,
+      "step": 21061
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1982677798115988e-05,
+      "loss": 1.1819,
+      "step": 21062
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1973748828072682e-05,
+      "loss": 1.056,
+      "step": 21063
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1964819858029378e-05,
+      "loss": 1.0666,
+      "step": 21064
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1955890887986072e-05,
+      "loss": 0.9928,
+      "step": 21065
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1946961917942765e-05,
+      "loss": 1.0274,
+      "step": 21066
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1938032947899461e-05,
+      "loss": 1.0612,
+      "step": 21067
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1929103977856155e-05,
+      "loss": 0.9875,
+      "step": 21068
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1920175007812849e-05,
+      "loss": 1.0093,
+      "step": 21069
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1911246037769544e-05,
+      "loss": 1.0526,
+      "step": 21070
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1902317067726238e-05,
+      "loss": 0.9829,
+      "step": 21071
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1893388097682932e-05,
+      "loss": 1.0182,
+      "step": 21072
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1884459127639628e-05,
+      "loss": 1.019,
+      "step": 21073
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1875530157596321e-05,
+      "loss": 0.9289,
+      "step": 21074
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1866601187553017e-05,
+      "loss": 1.0074,
+      "step": 21075
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1857672217509711e-05,
+      "loss": 0.9444,
+      "step": 21076
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1848743247466405e-05,
+      "loss": 0.9379,
+      "step": 21077
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.18398142774231e-05,
+      "loss": 1.0213,
+      "step": 21078
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1830885307379794e-05,
+      "loss": 0.9804,
+      "step": 21079
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1821956337336488e-05,
+      "loss": 1.1754,
+      "step": 21080
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1813027367293184e-05,
+      "loss": 1.2206,
+      "step": 21081
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.180409839724988e-05,
+      "loss": 1.0393,
+      "step": 21082
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1795169427206571e-05,
+      "loss": 0.9571,
+      "step": 21083
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1786240457163267e-05,
+      "loss": 1.0471,
+      "step": 21084
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1777311487119962e-05,
+      "loss": 1.0662,
+      "step": 21085
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1768382517076655e-05,
+      "loss": 1.0219,
+      "step": 21086
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.175945354703335e-05,
+      "loss": 0.9967,
+      "step": 21087
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1750524576990044e-05,
+      "loss": 1.0182,
+      "step": 21088
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.174159560694674e-05,
+      "loss": 1.0967,
+      "step": 21089
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1732666636903433e-05,
+      "loss": 1.1287,
+      "step": 21090
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1723737666860127e-05,
+      "loss": 1.0882,
+      "step": 21091
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1714808696816823e-05,
+      "loss": 1.0217,
+      "step": 21092
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1705879726773518e-05,
+      "loss": 0.9191,
+      "step": 21093
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.169695075673021e-05,
+      "loss": 1.0998,
+      "step": 21094
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1688021786686906e-05,
+      "loss": 0.9711,
+      "step": 21095
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1679092816643602e-05,
+      "loss": 0.99,
+      "step": 21096
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1670163846600294e-05,
+      "loss": 1.0386,
+      "step": 21097
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.166123487655699e-05,
+      "loss": 1.109,
+      "step": 21098
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1652305906513685e-05,
+      "loss": 1.1773,
+      "step": 21099
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1643376936470379e-05,
+      "loss": 1.0817,
+      "step": 21100
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1634447966427073e-05,
+      "loss": 0.9911,
+      "step": 21101
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1625518996383768e-05,
+      "loss": 1.0546,
+      "step": 21102
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1616590026340462e-05,
+      "loss": 1.1679,
+      "step": 21103
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1607661056297156e-05,
+      "loss": 0.9516,
+      "step": 21104
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1598732086253852e-05,
+      "loss": 1.0347,
+      "step": 21105
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1589803116210546e-05,
+      "loss": 1.2361,
+      "step": 21106
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1580874146167241e-05,
+      "loss": 1.0421,
+      "step": 21107
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1571945176123935e-05,
+      "loss": 1.0424,
+      "step": 21108
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1563016206080629e-05,
+      "loss": 1.0559,
+      "step": 21109
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1554087236037324e-05,
+      "loss": 1.0685,
+      "step": 21110
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1545158265994018e-05,
+      "loss": 1.0167,
+      "step": 21111
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1536229295950712e-05,
+      "loss": 0.988,
+      "step": 21112
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1527300325907408e-05,
+      "loss": 0.9463,
+      "step": 21113
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1518371355864102e-05,
+      "loss": 1.2576,
+      "step": 21114
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1509442385820795e-05,
+      "loss": 1.0806,
+      "step": 21115
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1500513415777491e-05,
+      "loss": 0.8892,
+      "step": 21116
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1491584445734185e-05,
+      "loss": 1.0806,
+      "step": 21117
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.148265547569088e-05,
+      "loss": 1.0299,
+      "step": 21118
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1473726505647574e-05,
+      "loss": 0.9462,
+      "step": 21119
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1464797535604268e-05,
+      "loss": 1.1053,
+      "step": 21120
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1455868565560964e-05,
+      "loss": 0.9816,
+      "step": 21121
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1446939595517658e-05,
+      "loss": 1.0122,
+      "step": 21122
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1438010625474351e-05,
+      "loss": 0.9542,
+      "step": 21123
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1429081655431047e-05,
+      "loss": 1.1066,
+      "step": 21124
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1420152685387741e-05,
+      "loss": 0.9924,
+      "step": 21125
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1411223715344435e-05,
+      "loss": 1.1374,
+      "step": 21126
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.140229474530113e-05,
+      "loss": 1.0277,
+      "step": 21127
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1393365775257824e-05,
+      "loss": 1.0198,
+      "step": 21128
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1384436805214518e-05,
+      "loss": 1.0685,
+      "step": 21129
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1375507835171214e-05,
+      "loss": 0.9613,
+      "step": 21130
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1366578865127908e-05,
+      "loss": 1.0052,
+      "step": 21131
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1357649895084603e-05,
+      "loss": 0.9506,
+      "step": 21132
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1348720925041297e-05,
+      "loss": 1.0799,
+      "step": 21133
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.133979195499799e-05,
+      "loss": 1.0089,
+      "step": 21134
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1330862984954686e-05,
+      "loss": 0.9758,
+      "step": 21135
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1321934014911382e-05,
+      "loss": 1.0786,
+      "step": 21136
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1313005044868074e-05,
+      "loss": 1.0314,
+      "step": 21137
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.130407607482477e-05,
+      "loss": 1.0318,
+      "step": 21138
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1295147104781465e-05,
+      "loss": 1.197,
+      "step": 21139
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1286218134738157e-05,
+      "loss": 1.0187,
+      "step": 21140
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1277289164694853e-05,
+      "loss": 1.0811,
+      "step": 21141
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1268360194651547e-05,
+      "loss": 0.9919,
+      "step": 21142
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1259431224608242e-05,
+      "loss": 1.0946,
+      "step": 21143
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1250502254564936e-05,
+      "loss": 1.0898,
+      "step": 21144
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.124157328452163e-05,
+      "loss": 0.9851,
+      "step": 21145
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1232644314478326e-05,
+      "loss": 1.051,
+      "step": 21146
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.122371534443502e-05,
+      "loss": 1.0548,
+      "step": 21147
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1214786374391713e-05,
+      "loss": 1.0996,
+      "step": 21148
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1205857404348409e-05,
+      "loss": 1.069,
+      "step": 21149
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1196928434305105e-05,
+      "loss": 1.021,
+      "step": 21150
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1187999464261797e-05,
+      "loss": 0.9768,
+      "step": 21151
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1179070494218492e-05,
+      "loss": 0.9813,
+      "step": 21152
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1170141524175188e-05,
+      "loss": 1.081,
+      "step": 21153
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1161212554131882e-05,
+      "loss": 1.0689,
+      "step": 21154
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1152283584088576e-05,
+      "loss": 0.8857,
+      "step": 21155
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1143354614045271e-05,
+      "loss": 1.12,
+      "step": 21156
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1134425644001965e-05,
+      "loss": 0.9244,
+      "step": 21157
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1125496673958659e-05,
+      "loss": 1.0244,
+      "step": 21158
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1116567703915354e-05,
+      "loss": 1.1297,
+      "step": 21159
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1107638733872048e-05,
+      "loss": 0.9751,
+      "step": 21160
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1098709763828744e-05,
+      "loss": 1.1316,
+      "step": 21161
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1089780793785438e-05,
+      "loss": 1.2406,
+      "step": 21162
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1080851823742132e-05,
+      "loss": 1.0663,
+      "step": 21163
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1071922853698827e-05,
+      "loss": 1.2341,
+      "step": 21164
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1062993883655521e-05,
+      "loss": 1.1327,
+      "step": 21165
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1054064913612215e-05,
+      "loss": 1.0935,
+      "step": 21166
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.104513594356891e-05,
+      "loss": 1.0743,
+      "step": 21167
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1036206973525604e-05,
+      "loss": 1.0502,
+      "step": 21168
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1027278003482298e-05,
+      "loss": 1.0702,
+      "step": 21169
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1018349033438994e-05,
+      "loss": 1.0381,
+      "step": 21170
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1009420063395688e-05,
+      "loss": 1.0272,
+      "step": 21171
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1000491093352382e-05,
+      "loss": 1.0654,
+      "step": 21172
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0991562123309077e-05,
+      "loss": 1.1103,
+      "step": 21173
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0982633153265771e-05,
+      "loss": 0.8457,
+      "step": 21174
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0973704183222467e-05,
+      "loss": 1.0083,
+      "step": 21175
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.096477521317916e-05,
+      "loss": 1.0411,
+      "step": 21176
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0955846243135854e-05,
+      "loss": 1.1142,
+      "step": 21177
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.094691727309255e-05,
+      "loss": 1.0603,
+      "step": 21178
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0937988303049244e-05,
+      "loss": 0.9434,
+      "step": 21179
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0929059333005938e-05,
+      "loss": 0.9574,
+      "step": 21180
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0920130362962633e-05,
+      "loss": 1.1008,
+      "step": 21181
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0911201392919327e-05,
+      "loss": 1.2401,
+      "step": 21182
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0902272422876021e-05,
+      "loss": 1.0135,
+      "step": 21183
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0893343452832716e-05,
+      "loss": 0.9898,
+      "step": 21184
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.088441448278941e-05,
+      "loss": 1.0251,
+      "step": 21185
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0875485512746106e-05,
+      "loss": 1.1758,
+      "step": 21186
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.08665565427028e-05,
+      "loss": 1.1314,
+      "step": 21187
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0857627572659494e-05,
+      "loss": 1.121,
+      "step": 21188
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.084869860261619e-05,
+      "loss": 1.0709,
+      "step": 21189
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0839769632572883e-05,
+      "loss": 0.9428,
+      "step": 21190
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0830840662529577e-05,
+      "loss": 0.9428,
+      "step": 21191
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0821911692486272e-05,
+      "loss": 1.0875,
+      "step": 21192
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0812982722442968e-05,
+      "loss": 1.1004,
+      "step": 21193
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.080405375239966e-05,
+      "loss": 1.1271,
+      "step": 21194
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0795124782356356e-05,
+      "loss": 1.0741,
+      "step": 21195
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0786195812313051e-05,
+      "loss": 1.1262,
+      "step": 21196
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0777266842269745e-05,
+      "loss": 0.8944,
+      "step": 21197
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0768337872226439e-05,
+      "loss": 1.1408,
+      "step": 21198
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0759408902183133e-05,
+      "loss": 1.0115,
+      "step": 21199
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0750479932139828e-05,
+      "loss": 1.0168,
+      "step": 21200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0741550962096522e-05,
+      "loss": 1.0723,
+      "step": 21201
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0732621992053216e-05,
+      "loss": 1.0701,
+      "step": 21202
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0723693022009912e-05,
+      "loss": 1.1063,
+      "step": 21203
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0714764051966607e-05,
+      "loss": 1.1401,
+      "step": 21204
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.07058350819233e-05,
+      "loss": 0.9555,
+      "step": 21205
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0696906111879995e-05,
+      "loss": 1.1354,
+      "step": 21206
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.068797714183669e-05,
+      "loss": 1.0604,
+      "step": 21207
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0679048171793383e-05,
+      "loss": 1.0815,
+      "step": 21208
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0670119201750078e-05,
+      "loss": 1.0649,
+      "step": 21209
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0661190231706774e-05,
+      "loss": 0.9215,
+      "step": 21210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0652261261663468e-05,
+      "loss": 1.0673,
+      "step": 21211
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0643332291620162e-05,
+      "loss": 1.0313,
+      "step": 21212
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0634403321576857e-05,
+      "loss": 1.0256,
+      "step": 21213
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0625474351533551e-05,
+      "loss": 0.944,
+      "step": 21214
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0616545381490247e-05,
+      "loss": 1.0862,
+      "step": 21215
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.060761641144694e-05,
+      "loss": 1.0456,
+      "step": 21216
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0598687441403634e-05,
+      "loss": 1.0605,
+      "step": 21217
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.058975847136033e-05,
+      "loss": 1.1757,
+      "step": 21218
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0580829501317024e-05,
+      "loss": 0.9494,
+      "step": 21219
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0571900531273718e-05,
+      "loss": 1.0053,
+      "step": 21220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0562971561230413e-05,
+      "loss": 0.9519,
+      "step": 21221
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0554042591187107e-05,
+      "loss": 1.0874,
+      "step": 21222
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0545113621143801e-05,
+      "loss": 1.1115,
+      "step": 21223
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0536184651100497e-05,
+      "loss": 1.1121,
+      "step": 21224
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.052725568105719e-05,
+      "loss": 0.9231,
+      "step": 21225
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0518326711013884e-05,
+      "loss": 1.0715,
+      "step": 21226
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.050939774097058e-05,
+      "loss": 0.8924,
+      "step": 21227
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0500468770927274e-05,
+      "loss": 0.9907,
+      "step": 21228
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.049153980088397e-05,
+      "loss": 0.953,
+      "step": 21229
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0482610830840663e-05,
+      "loss": 0.9967,
+      "step": 21230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0473681860797357e-05,
+      "loss": 0.9476,
+      "step": 21231
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0464752890754053e-05,
+      "loss": 1.0984,
+      "step": 21232
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0455823920710746e-05,
+      "loss": 1.1295,
+      "step": 21233
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.044689495066744e-05,
+      "loss": 1.0104,
+      "step": 21234
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0437965980624136e-05,
+      "loss": 1.0338,
+      "step": 21235
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.042903701058083e-05,
+      "loss": 1.0413,
+      "step": 21236
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0420108040537524e-05,
+      "loss": 0.8805,
+      "step": 21237
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.041117907049422e-05,
+      "loss": 1.0542,
+      "step": 21238
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0402250100450913e-05,
+      "loss": 0.9674,
+      "step": 21239
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0393321130407609e-05,
+      "loss": 0.9852,
+      "step": 21240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0384392160364303e-05,
+      "loss": 1.0295,
+      "step": 21241
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0375463190320996e-05,
+      "loss": 0.9247,
+      "step": 21242
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0366534220277692e-05,
+      "loss": 1.0059,
+      "step": 21243
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0357605250234386e-05,
+      "loss": 0.9695,
+      "step": 21244
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.034867628019108e-05,
+      "loss": 1.1043,
+      "step": 21245
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0339747310147775e-05,
+      "loss": 1.0362,
+      "step": 21246
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.033081834010447e-05,
+      "loss": 1.0071,
+      "step": 21247
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0321889370061163e-05,
+      "loss": 1.1299,
+      "step": 21248
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0312960400017859e-05,
+      "loss": 1.1128,
+      "step": 21249
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0304031429974554e-05,
+      "loss": 0.9951,
+      "step": 21250
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0295102459931246e-05,
+      "loss": 0.9872,
+      "step": 21251
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0286173489887942e-05,
+      "loss": 1.032,
+      "step": 21252
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0277244519844636e-05,
+      "loss": 0.9373,
+      "step": 21253
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0268315549801331e-05,
+      "loss": 1.0831,
+      "step": 21254
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0259386579758025e-05,
+      "loss": 1.0602,
+      "step": 21255
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0250457609714719e-05,
+      "loss": 0.8949,
+      "step": 21256
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0241528639671415e-05,
+      "loss": 0.9995,
+      "step": 21257
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.023259966962811e-05,
+      "loss": 1.0293,
+      "step": 21258
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0223670699584802e-05,
+      "loss": 0.9498,
+      "step": 21259
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0214741729541498e-05,
+      "loss": 0.9939,
+      "step": 21260
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0205812759498193e-05,
+      "loss": 1.1234,
+      "step": 21261
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0196883789454886e-05,
+      "loss": 0.901,
+      "step": 21262
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0187954819411581e-05,
+      "loss": 1.1585,
+      "step": 21263
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0179025849368277e-05,
+      "loss": 0.9407,
+      "step": 21264
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.017009687932497e-05,
+      "loss": 1.2054,
+      "step": 21265
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0161167909281664e-05,
+      "loss": 0.9326,
+      "step": 21266
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.015223893923836e-05,
+      "loss": 0.9692,
+      "step": 21267
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0143309969195054e-05,
+      "loss": 0.918,
+      "step": 21268
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0134380999151748e-05,
+      "loss": 0.9328,
+      "step": 21269
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0125452029108443e-05,
+      "loss": 1.0815,
+      "step": 21270
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0116523059065137e-05,
+      "loss": 1.0051,
+      "step": 21271
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0107594089021833e-05,
+      "loss": 1.1259,
+      "step": 21272
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0098665118978527e-05,
+      "loss": 0.9188,
+      "step": 21273
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.008973614893522e-05,
+      "loss": 1.0481,
+      "step": 21274
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0080807178891916e-05,
+      "loss": 0.9882,
+      "step": 21275
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.007187820884861e-05,
+      "loss": 0.8856,
+      "step": 21276
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0062949238805304e-05,
+      "loss": 0.9568,
+      "step": 21277
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0054020268762e-05,
+      "loss": 1.0382,
+      "step": 21278
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0045091298718693e-05,
+      "loss": 1.0171,
+      "step": 21279
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0036162328675387e-05,
+      "loss": 1.0216,
+      "step": 21280
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0027233358632083e-05,
+      "loss": 0.9375,
+      "step": 21281
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0018304388588777e-05,
+      "loss": 1.1473,
+      "step": 21282
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0009375418545472e-05,
+      "loss": 1.1995,
+      "step": 21283
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0000446448502166e-05,
+      "loss": 1.1148,
+      "step": 21284
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.99151747845886e-06,
+      "loss": 1.0227,
+      "step": 21285
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.982588508415555e-06,
+      "loss": 1.1391,
+      "step": 21286
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.97365953837225e-06,
+      "loss": 0.9268,
+      "step": 21287
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.964730568328943e-06,
+      "loss": 0.8893,
+      "step": 21288
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.955801598285639e-06,
+      "loss": 0.9928,
+      "step": 21289
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.946872628242333e-06,
+      "loss": 1.0273,
+      "step": 21290
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.937943658199026e-06,
+      "loss": 1.0759,
+      "step": 21291
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.929014688155722e-06,
+      "loss": 1.1142,
+      "step": 21292
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.920085718112416e-06,
+      "loss": 1.1378,
+      "step": 21293
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.91115674806911e-06,
+      "loss": 1.0081,
+      "step": 21294
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.902227778025805e-06,
+      "loss": 0.9031,
+      "step": 21295
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.8932988079825e-06,
+      "loss": 1.1447,
+      "step": 21296
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.884369837939195e-06,
+      "loss": 1.1611,
+      "step": 21297
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.875440867895889e-06,
+      "loss": 1.0139,
+      "step": 21298
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.866511897852582e-06,
+      "loss": 1.1722,
+      "step": 21299
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.857582927809278e-06,
+      "loss": 0.9761,
+      "step": 21300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.848653957765974e-06,
+      "loss": 0.9504,
+      "step": 21301
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.839724987722666e-06,
+      "loss": 1.0121,
+      "step": 21302
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.830796017679361e-06,
+      "loss": 0.9589,
+      "step": 21303
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.821867047636057e-06,
+      "loss": 1.0846,
+      "step": 21304
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.812938077592749e-06,
+      "loss": 0.8908,
+      "step": 21305
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.804009107549445e-06,
+      "loss": 1.0063,
+      "step": 21306
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.79508013750614e-06,
+      "loss": 0.9603,
+      "step": 21307
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.786151167462834e-06,
+      "loss": 1.0922,
+      "step": 21308
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.777222197419528e-06,
+      "loss": 1.0227,
+      "step": 21309
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.768293227376222e-06,
+      "loss": 0.9977,
+      "step": 21310
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.759364257332917e-06,
+      "loss": 0.9851,
+      "step": 21311
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.750435287289611e-06,
+      "loss": 0.9683,
+      "step": 21312
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.741506317246305e-06,
+      "loss": 1.0184,
+      "step": 21313
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.732577347203e-06,
+      "loss": 0.9356,
+      "step": 21314
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.723648377159696e-06,
+      "loss": 1.2651,
+      "step": 21315
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.714719407116388e-06,
+      "loss": 0.9844,
+      "step": 21316
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.705790437073084e-06,
+      "loss": 1.0671,
+      "step": 21317
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.69686146702978e-06,
+      "loss": 1.0782,
+      "step": 21318
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.687932496986473e-06,
+      "loss": 1.1821,
+      "step": 21319
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.679003526943167e-06,
+      "loss": 1.0329,
+      "step": 21320
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.670074556899863e-06,
+      "loss": 0.9011,
+      "step": 21321
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.661145586856557e-06,
+      "loss": 1.0495,
+      "step": 21322
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.65221661681325e-06,
+      "loss": 0.9467,
+      "step": 21323
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.643287646769946e-06,
+      "loss": 0.9974,
+      "step": 21324
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.63435867672664e-06,
+      "loss": 0.905,
+      "step": 21325
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.625429706683336e-06,
+      "loss": 0.9039,
+      "step": 21326
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.61650073664003e-06,
+      "loss": 1.3412,
+      "step": 21327
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.607571766596723e-06,
+      "loss": 0.9577,
+      "step": 21328
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.598642796553419e-06,
+      "loss": 0.9549,
+      "step": 21329
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.589713826510113e-06,
+      "loss": 1.1387,
+      "step": 21330
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.580784856466807e-06,
+      "loss": 0.9801,
+      "step": 21331
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.571855886423502e-06,
+      "loss": 0.9418,
+      "step": 21332
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.562926916380196e-06,
+      "loss": 1.0782,
+      "step": 21333
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.55399794633689e-06,
+      "loss": 1.109,
+      "step": 21334
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.545068976293585e-06,
+      "loss": 1.0757,
+      "step": 21335
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.53614000625028e-06,
+      "loss": 1.0946,
+      "step": 21336
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.527211036206973e-06,
+      "loss": 0.9584,
+      "step": 21337
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.518282066163669e-06,
+      "loss": 0.9522,
+      "step": 21338
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.509353096120363e-06,
+      "loss": 0.9857,
+      "step": 21339
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.500424126077058e-06,
+      "loss": 0.9845,
+      "step": 21340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.491495156033752e-06,
+      "loss": 1.1861,
+      "step": 21341
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.482566185990446e-06,
+      "loss": 0.9712,
+      "step": 21342
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.473637215947141e-06,
+      "loss": 0.9844,
+      "step": 21343
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.464708245903835e-06,
+      "loss": 1.0527,
+      "step": 21344
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.45577927586053e-06,
+      "loss": 1.0202,
+      "step": 21345
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.446850305817225e-06,
+      "loss": 0.9046,
+      "step": 21346
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.437921335773919e-06,
+      "loss": 1.0554,
+      "step": 21347
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.428992365730613e-06,
+      "loss": 0.9551,
+      "step": 21348
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.420063395687308e-06,
+      "loss": 0.9678,
+      "step": 21349
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.411134425644002e-06,
+      "loss": 1.0267,
+      "step": 21350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.402205455600697e-06,
+      "loss": 0.9084,
+      "step": 21351
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.393276485557391e-06,
+      "loss": 1.0566,
+      "step": 21352
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.384347515514085e-06,
+      "loss": 1.0021,
+      "step": 21353
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.37541854547078e-06,
+      "loss": 1.0745,
+      "step": 21354
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.366489575427475e-06,
+      "loss": 1.2269,
+      "step": 21355
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.357560605384169e-06,
+      "loss": 1.043,
+      "step": 21356
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.348631635340864e-06,
+      "loss": 1.0084,
+      "step": 21357
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.33970266529756e-06,
+      "loss": 0.9938,
+      "step": 21358
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.330773695254252e-06,
+      "loss": 0.9898,
+      "step": 21359
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.321844725210947e-06,
+      "loss": 1.0373,
+      "step": 21360
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.312915755167643e-06,
+      "loss": 1.0477,
+      "step": 21361
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.303986785124337e-06,
+      "loss": 0.9598,
+      "step": 21362
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.29505781508103e-06,
+      "loss": 1.1235,
+      "step": 21363
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.286128845037725e-06,
+      "loss": 1.0835,
+      "step": 21364
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.27719987499442e-06,
+      "loss": 0.9482,
+      "step": 21365
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.268270904951114e-06,
+      "loss": 1.1626,
+      "step": 21366
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.259341934907808e-06,
+      "loss": 0.9482,
+      "step": 21367
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.250412964864503e-06,
+      "loss": 1.0622,
+      "step": 21368
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.241483994821199e-06,
+      "loss": 0.9258,
+      "step": 21369
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.232555024777891e-06,
+      "loss": 1.0216,
+      "step": 21370
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.223626054734587e-06,
+      "loss": 1.118,
+      "step": 21371
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.214697084691282e-06,
+      "loss": 0.8737,
+      "step": 21372
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.205768114647974e-06,
+      "loss": 1.1388,
+      "step": 21373
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.19683914460467e-06,
+      "loss": 1.1207,
+      "step": 21374
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.187910174561366e-06,
+      "loss": 1.1099,
+      "step": 21375
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.17898120451806e-06,
+      "loss": 0.9991,
+      "step": 21376
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.170052234474753e-06,
+      "loss": 0.9845,
+      "step": 21377
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.161123264431449e-06,
+      "loss": 0.9781,
+      "step": 21378
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.152194294388143e-06,
+      "loss": 0.9783,
+      "step": 21379
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.143265324344837e-06,
+      "loss": 1.0782,
+      "step": 21380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.134336354301532e-06,
+      "loss": 0.9962,
+      "step": 21381
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.125407384258226e-06,
+      "loss": 0.953,
+      "step": 21382
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.116478414214922e-06,
+      "loss": 1.0134,
+      "step": 21383
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.107549444171615e-06,
+      "loss": 0.9514,
+      "step": 21384
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.09862047412831e-06,
+      "loss": 1.014,
+      "step": 21385
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.089691504085005e-06,
+      "loss": 0.9919,
+      "step": 21386
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.080762534041699e-06,
+      "loss": 0.9731,
+      "step": 21387
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.071833563998393e-06,
+      "loss": 1.0576,
+      "step": 21388
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.062904593955088e-06,
+      "loss": 0.9727,
+      "step": 21389
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.053975623911782e-06,
+      "loss": 1.1632,
+      "step": 21390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.045046653868476e-06,
+      "loss": 1.0132,
+      "step": 21391
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.036117683825172e-06,
+      "loss": 0.8987,
+      "step": 21392
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.027188713781865e-06,
+      "loss": 1.0882,
+      "step": 21393
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.018259743738561e-06,
+      "loss": 0.9521,
+      "step": 21394
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.009330773695255e-06,
+      "loss": 1.2213,
+      "step": 21395
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.000401803651949e-06,
+      "loss": 1.0139,
+      "step": 21396
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.991472833608644e-06,
+      "loss": 0.9931,
+      "step": 21397
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.982543863565338e-06,
+      "loss": 1.1218,
+      "step": 21398
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.973614893522032e-06,
+      "loss": 0.8904,
+      "step": 21399
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.964685923478728e-06,
+      "loss": 1.0011,
+      "step": 21400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.955756953435421e-06,
+      "loss": 1.1268,
+      "step": 21401
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.946827983392115e-06,
+      "loss": 1.034,
+      "step": 21402
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.93789901334881e-06,
+      "loss": 1.0205,
+      "step": 21403
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.928970043305505e-06,
+      "loss": 1.0462,
+      "step": 21404
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.9200410732622e-06,
+      "loss": 0.9948,
+      "step": 21405
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.911112103218894e-06,
+      "loss": 0.9208,
+      "step": 21406
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.902183133175588e-06,
+      "loss": 1.0454,
+      "step": 21407
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.893254163132284e-06,
+      "loss": 1.0476,
+      "step": 21408
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.884325193088977e-06,
+      "loss": 0.9839,
+      "step": 21409
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.875396223045671e-06,
+      "loss": 1.0173,
+      "step": 21410
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.866467253002367e-06,
+      "loss": 0.9835,
+      "step": 21411
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.857538282959062e-06,
+      "loss": 1.0644,
+      "step": 21412
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.848609312915755e-06,
+      "loss": 1.0855,
+      "step": 21413
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.83968034287245e-06,
+      "loss": 1.1988,
+      "step": 21414
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.830751372829146e-06,
+      "loss": 1.0818,
+      "step": 21415
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.821822402785838e-06,
+      "loss": 0.9481,
+      "step": 21416
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.812893432742533e-06,
+      "loss": 0.9819,
+      "step": 21417
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.803964462699229e-06,
+      "loss": 0.9647,
+      "step": 21418
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.795035492655923e-06,
+      "loss": 1.0139,
+      "step": 21419
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.786106522612617e-06,
+      "loss": 1.0544,
+      "step": 21420
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.77717755256931e-06,
+      "loss": 0.986,
+      "step": 21421
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.768248582526006e-06,
+      "loss": 1.1359,
+      "step": 21422
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.759319612482702e-06,
+      "loss": 1.0092,
+      "step": 21423
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.750390642439394e-06,
+      "loss": 1.0032,
+      "step": 21424
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.74146167239609e-06,
+      "loss": 0.9677,
+      "step": 21425
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.732532702352785e-06,
+      "loss": 0.8829,
+      "step": 21426
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.723603732309477e-06,
+      "loss": 1.0614,
+      "step": 21427
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.714674762266173e-06,
+      "loss": 0.9314,
+      "step": 21428
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.705745792222868e-06,
+      "loss": 1.0072,
+      "step": 21429
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.696816822179562e-06,
+      "loss": 0.9892,
+      "step": 21430
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.687887852136256e-06,
+      "loss": 0.9991,
+      "step": 21431
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.678958882092952e-06,
+      "loss": 0.9899,
+      "step": 21432
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.670029912049646e-06,
+      "loss": 1.0225,
+      "step": 21433
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.66110094200634e-06,
+      "loss": 1.0013,
+      "step": 21434
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.652171971963035e-06,
+      "loss": 1.0423,
+      "step": 21435
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.643243001919729e-06,
+      "loss": 1.0138,
+      "step": 21436
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.634314031876424e-06,
+      "loss": 1.0002,
+      "step": 21437
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.625385061833118e-06,
+      "loss": 0.974,
+      "step": 21438
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.616456091789812e-06,
+      "loss": 0.9566,
+      "step": 21439
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.607527121746508e-06,
+      "loss": 1.0184,
+      "step": 21440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.598598151703202e-06,
+      "loss": 1.0351,
+      "step": 21441
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.589669181659895e-06,
+      "loss": 1.1242,
+      "step": 21442
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.580740211616591e-06,
+      "loss": 0.899,
+      "step": 21443
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.571811241573285e-06,
+      "loss": 1.0914,
+      "step": 21444
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.562882271529979e-06,
+      "loss": 1.0146,
+      "step": 21445
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.553953301486674e-06,
+      "loss": 1.0954,
+      "step": 21446
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.545024331443368e-06,
+      "loss": 1.0515,
+      "step": 21447
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.536095361400064e-06,
+      "loss": 0.9667,
+      "step": 21448
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.527166391356758e-06,
+      "loss": 1.0006,
+      "step": 21449
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.518237421313451e-06,
+      "loss": 0.9969,
+      "step": 21450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.509308451270147e-06,
+      "loss": 0.9476,
+      "step": 21451
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.500379481226841e-06,
+      "loss": 0.9484,
+      "step": 21452
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.491450511183535e-06,
+      "loss": 1.0439,
+      "step": 21453
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.48252154114023e-06,
+      "loss": 1.1509,
+      "step": 21454
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.473592571096924e-06,
+      "loss": 1.0386,
+      "step": 21455
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.464663601053618e-06,
+      "loss": 0.9619,
+      "step": 21456
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.455734631010314e-06,
+      "loss": 1.0157,
+      "step": 21457
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.446805660967007e-06,
+      "loss": 1.0123,
+      "step": 21458
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.437876690923701e-06,
+      "loss": 1.0049,
+      "step": 21459
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.428947720880397e-06,
+      "loss": 1.06,
+      "step": 21460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.42001875083709e-06,
+      "loss": 0.9627,
+      "step": 21461
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.411089780793786e-06,
+      "loss": 1.0588,
+      "step": 21462
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.40216081075048e-06,
+      "loss": 0.9618,
+      "step": 21463
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.393231840707174e-06,
+      "loss": 0.9687,
+      "step": 21464
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.38430287066387e-06,
+      "loss": 0.925,
+      "step": 21465
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.375373900620565e-06,
+      "loss": 0.9852,
+      "step": 21466
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.366444930577257e-06,
+      "loss": 1.0784,
+      "step": 21467
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.357515960533953e-06,
+      "loss": 1.0466,
+      "step": 21468
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.348586990490649e-06,
+      "loss": 0.9746,
+      "step": 21469
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.33965802044734e-06,
+      "loss": 0.9737,
+      "step": 21470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.330729050404036e-06,
+      "loss": 1.1241,
+      "step": 21471
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.321800080360732e-06,
+      "loss": 0.9738,
+      "step": 21472
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.312871110317426e-06,
+      "loss": 0.9672,
+      "step": 21473
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.30394214027412e-06,
+      "loss": 1.1713,
+      "step": 21474
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.295013170230813e-06,
+      "loss": 1.0364,
+      "step": 21475
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.286084200187509e-06,
+      "loss": 0.9838,
+      "step": 21476
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.277155230144203e-06,
+      "loss": 1.0434,
+      "step": 21477
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.268226260100897e-06,
+      "loss": 0.985,
+      "step": 21478
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.259297290057592e-06,
+      "loss": 1.0282,
+      "step": 21479
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.250368320014288e-06,
+      "loss": 1.0134,
+      "step": 21480
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.24143934997098e-06,
+      "loss": 0.9678,
+      "step": 21481
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.232510379927676e-06,
+      "loss": 1.1037,
+      "step": 21482
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.223581409884371e-06,
+      "loss": 0.9703,
+      "step": 21483
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.214652439841065e-06,
+      "loss": 0.9528,
+      "step": 21484
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.205723469797759e-06,
+      "loss": 1.028,
+      "step": 21485
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.196794499754454e-06,
+      "loss": 1.0234,
+      "step": 21486
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.187865529711148e-06,
+      "loss": 1.0362,
+      "step": 21487
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.178936559667842e-06,
+      "loss": 0.9535,
+      "step": 21488
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.170007589624538e-06,
+      "loss": 0.9793,
+      "step": 21489
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.161078619581232e-06,
+      "loss": 1.0394,
+      "step": 21490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.152149649537927e-06,
+      "loss": 0.9637,
+      "step": 21491
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.143220679494621e-06,
+      "loss": 0.9228,
+      "step": 21492
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.134291709451315e-06,
+      "loss": 1.1979,
+      "step": 21493
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.12536273940801e-06,
+      "loss": 1.1556,
+      "step": 21494
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.116433769364704e-06,
+      "loss": 1.0332,
+      "step": 21495
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.107504799321398e-06,
+      "loss": 1.031,
+      "step": 21496
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.098575829278094e-06,
+      "loss": 1.0353,
+      "step": 21497
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.089646859234788e-06,
+      "loss": 0.9515,
+      "step": 21498
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.080717889191482e-06,
+      "loss": 0.9936,
+      "step": 21499
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.071788919148177e-06,
+      "loss": 0.9929,
+      "step": 21500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.062859949104871e-06,
+      "loss": 1.2575,
+      "step": 21501
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053930979061565e-06,
+      "loss": 1.1719,
+      "step": 21502
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.04500200901826e-06,
+      "loss": 1.1035,
+      "step": 21503
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.036073038974954e-06,
+      "loss": 1.0548,
+      "step": 21504
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02714406893165e-06,
+      "loss": 1.0068,
+      "step": 21505
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.018215098888344e-06,
+      "loss": 1.1677,
+      "step": 21506
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.009286128845038e-06,
+      "loss": 1.1403,
+      "step": 21507
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.000357158801733e-06,
+      "loss": 1.1169,
+      "step": 21508
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.991428188758427e-06,
+      "loss": 0.9988,
+      "step": 21509
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.98249921871512e-06,
+      "loss": 0.9841,
+      "step": 21510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.973570248671816e-06,
+      "loss": 1.0083,
+      "step": 21511
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.96464127862851e-06,
+      "loss": 0.9714,
+      "step": 21512
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.955712308585204e-06,
+      "loss": 0.9866,
+      "step": 21513
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.9467833385419e-06,
+      "loss": 1.0298,
+      "step": 21514
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.937854368498594e-06,
+      "loss": 1.0539,
+      "step": 21515
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.928925398455289e-06,
+      "loss": 1.0198,
+      "step": 21516
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.919996428411983e-06,
+      "loss": 1.2451,
+      "step": 21517
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.911067458368677e-06,
+      "loss": 0.9309,
+      "step": 21518
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.902138488325372e-06,
+      "loss": 1.0512,
+      "step": 21519
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.893209518282066e-06,
+      "loss": 1.0593,
+      "step": 21520
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.88428054823876e-06,
+      "loss": 1.0036,
+      "step": 21521
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.875351578195456e-06,
+      "loss": 1.1087,
+      "step": 21522
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.866422608152151e-06,
+      "loss": 1.0227,
+      "step": 21523
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.857493638108843e-06,
+      "loss": 0.9889,
+      "step": 21524
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.848564668065539e-06,
+      "loss": 0.9439,
+      "step": 21525
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.839635698022235e-06,
+      "loss": 1.037,
+      "step": 21526
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.830706727978928e-06,
+      "loss": 1.0757,
+      "step": 21527
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.821777757935622e-06,
+      "loss": 0.9669,
+      "step": 21528
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.812848787892318e-06,
+      "loss": 1.0785,
+      "step": 21529
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.803919817849012e-06,
+      "loss": 1.0106,
+      "step": 21530
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.794990847805706e-06,
+      "loss": 1.2019,
+      "step": 21531
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.7860618777624e-06,
+      "loss": 1.063,
+      "step": 21532
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.777132907719095e-06,
+      "loss": 0.9941,
+      "step": 21533
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.76820393767579e-06,
+      "loss": 0.9668,
+      "step": 21534
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.759274967632483e-06,
+      "loss": 1.0993,
+      "step": 21535
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.750345997589178e-06,
+      "loss": 1.033,
+      "step": 21536
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.741417027545874e-06,
+      "loss": 1.1346,
+      "step": 21537
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.732488057502566e-06,
+      "loss": 0.9404,
+      "step": 21538
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.723559087459262e-06,
+      "loss": 0.9828,
+      "step": 21539
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.714630117415957e-06,
+      "loss": 1.1481,
+      "step": 21540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.705701147372651e-06,
+      "loss": 0.9372,
+      "step": 21541
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.696772177329345e-06,
+      "loss": 1.024,
+      "step": 21542
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.68784320728604e-06,
+      "loss": 0.9929,
+      "step": 21543
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.678914237242734e-06,
+      "loss": 0.9416,
+      "step": 21544
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.669985267199428e-06,
+      "loss": 0.9775,
+      "step": 21545
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.661056297156124e-06,
+      "loss": 1.0746,
+      "step": 21546
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.652127327112818e-06,
+      "loss": 1.0308,
+      "step": 21547
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.643198357069513e-06,
+      "loss": 0.9704,
+      "step": 21548
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.634269387026207e-06,
+      "loss": 1.0773,
+      "step": 21549
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.625340416982901e-06,
+      "loss": 1.0381,
+      "step": 21550
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.616411446939596e-06,
+      "loss": 1.0911,
+      "step": 21551
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.607482476896291e-06,
+      "loss": 1.0273,
+      "step": 21552
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.598553506852984e-06,
+      "loss": 1.1008,
+      "step": 21553
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.589624536809679e-06,
+      "loss": 1.0034,
+      "step": 21554
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.5806955667663746e-06,
+      "loss": 0.876,
+      "step": 21555
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.571766596723068e-06,
+      "loss": 0.9993,
+      "step": 21556
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.562837626679762e-06,
+      "loss": 0.9177,
+      "step": 21557
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.553908656636458e-06,
+      "loss": 0.9337,
+      "step": 21558
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.544979686593153e-06,
+      "loss": 0.9589,
+      "step": 21559
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.536050716549846e-06,
+      "loss": 0.8805,
+      "step": 21560
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.527121746506541e-06,
+      "loss": 0.9637,
+      "step": 21561
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.518192776463236e-06,
+      "loss": 1.0722,
+      "step": 21562
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.509263806419929e-06,
+      "loss": 1.0546,
+      "step": 21563
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.500334836376624e-06,
+      "loss": 1.0134,
+      "step": 21564
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.491405866333319e-06,
+      "loss": 0.9153,
+      "step": 21565
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.482476896290014e-06,
+      "loss": 1.1297,
+      "step": 21566
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.473547926246707e-06,
+      "loss": 0.9827,
+      "step": 21567
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.4646189562034025e-06,
+      "loss": 0.9168,
+      "step": 21568
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.455689986160097e-06,
+      "loss": 1.0777,
+      "step": 21569
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.446761016116792e-06,
+      "loss": 1.1459,
+      "step": 21570
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.437832046073486e-06,
+      "loss": 1.079,
+      "step": 21571
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.4289030760301805e-06,
+      "loss": 1.0986,
+      "step": 21572
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.419974105986875e-06,
+      "loss": 0.96,
+      "step": 21573
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.411045135943569e-06,
+      "loss": 1.2985,
+      "step": 21574
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.402116165900264e-06,
+      "loss": 1.0395,
+      "step": 21575
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.3931871958569585e-06,
+      "loss": 1.1172,
+      "step": 21576
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.384258225813653e-06,
+      "loss": 0.8893,
+      "step": 21577
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.375329255770347e-06,
+      "loss": 0.9233,
+      "step": 21578
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.366400285727042e-06,
+      "loss": 1.0512,
+      "step": 21579
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.3574713156837365e-06,
+      "loss": 1.0236,
+      "step": 21580
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.34854234564043e-06,
+      "loss": 1.091,
+      "step": 21581
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.339613375597125e-06,
+      "loss": 1.0343,
+      "step": 21582
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.33068440555382e-06,
+      "loss": 0.9523,
+      "step": 21583
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.3217554355105145e-06,
+      "loss": 0.9943,
+      "step": 21584
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.312826465467208e-06,
+      "loss": 1.0938,
+      "step": 21585
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.303897495423903e-06,
+      "loss": 0.9532,
+      "step": 21586
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.294968525380598e-06,
+      "loss": 1.1736,
+      "step": 21587
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.2860395553372926e-06,
+      "loss": 1.0003,
+      "step": 21588
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.2771105852939864e-06,
+      "loss": 0.9861,
+      "step": 21589
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.268181615250681e-06,
+      "loss": 0.99,
+      "step": 21590
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.259252645207376e-06,
+      "loss": 1.0651,
+      "step": 21591
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.25032367516407e-06,
+      "loss": 0.9326,
+      "step": 21592
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.2413947051207644e-06,
+      "loss": 0.9866,
+      "step": 21593
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.232465735077459e-06,
+      "loss": 0.9192,
+      "step": 21594
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.223536765034154e-06,
+      "loss": 0.8656,
+      "step": 21595
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.214607794990848e-06,
+      "loss": 1.1131,
+      "step": 21596
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.2056788249475425e-06,
+      "loss": 0.9928,
+      "step": 21597
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.196749854904237e-06,
+      "loss": 1.027,
+      "step": 21598
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.187820884860931e-06,
+      "loss": 1.038,
+      "step": 21599
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.178891914817626e-06,
+      "loss": 1.0504,
+      "step": 21600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.1699629447743205e-06,
+      "loss": 1.0474,
+      "step": 21601
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.161033974731016e-06,
+      "loss": 1.065,
+      "step": 21602
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.152105004687709e-06,
+      "loss": 0.9064,
+      "step": 21603
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.143176034644404e-06,
+      "loss": 1.0623,
+      "step": 21604
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.134247064601099e-06,
+      "loss": 1.0361,
+      "step": 21605
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.125318094557792e-06,
+      "loss": 0.9241,
+      "step": 21606
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.116389124514487e-06,
+      "loss": 1.1855,
+      "step": 21607
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.107460154471182e-06,
+      "loss": 1.0354,
+      "step": 21608
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.098531184427877e-06,
+      "loss": 0.9771,
+      "step": 21609
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.08960221438457e-06,
+      "loss": 0.9731,
+      "step": 21610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.080673244341265e-06,
+      "loss": 0.9592,
+      "step": 21611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.071744274297961e-06,
+      "loss": 1.0957,
+      "step": 21612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.062815304254655e-06,
+      "loss": 1.1164,
+      "step": 21613
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.053886334211348e-06,
+      "loss": 0.9561,
+      "step": 21614
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.044957364168044e-06,
+      "loss": 0.906,
+      "step": 21615
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.036028394124739e-06,
+      "loss": 1.3805,
+      "step": 21616
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.027099424081432e-06,
+      "loss": 0.9601,
+      "step": 21617
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.018170454038127e-06,
+      "loss": 1.0084,
+      "step": 21618
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.009241483994822e-06,
+      "loss": 1.0336,
+      "step": 21619
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.000312513951517e-06,
+      "loss": 0.9665,
+      "step": 21620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.99138354390821e-06,
+      "loss": 1.1261,
+      "step": 21621
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.982454573864905e-06,
+      "loss": 1.0135,
+      "step": 21622
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.9735256038216e-06,
+      "loss": 1.1352,
+      "step": 21623
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.964596633778293e-06,
+      "loss": 1.03,
+      "step": 21624
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.9556676637349886e-06,
+      "loss": 1.072,
+      "step": 21625
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.946738693691683e-06,
+      "loss": 1.0591,
+      "step": 21626
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.937809723648378e-06,
+      "loss": 1.019,
+      "step": 21627
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.928880753605072e-06,
+      "loss": 0.9581,
+      "step": 21628
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.9199517835617666e-06,
+      "loss": 1.0043,
+      "step": 21629
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.911022813518461e-06,
+      "loss": 0.8725,
+      "step": 21630
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.902093843475156e-06,
+      "loss": 1.0496,
+      "step": 21631
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.89316487343185e-06,
+      "loss": 1.05,
+      "step": 21632
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.884235903388545e-06,
+      "loss": 1.0009,
+      "step": 21633
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.875306933345239e-06,
+      "loss": 1.0547,
+      "step": 21634
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.866377963301933e-06,
+      "loss": 1.0255,
+      "step": 21635
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.857448993258628e-06,
+      "loss": 0.9711,
+      "step": 21636
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.848520023215323e-06,
+      "loss": 1.0997,
+      "step": 21637
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.839591053172017e-06,
+      "loss": 0.9496,
+      "step": 21638
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.830662083128711e-06,
+      "loss": 0.8991,
+      "step": 21639
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.821733113085406e-06,
+      "loss": 0.9813,
+      "step": 21640
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.812804143042101e-06,
+      "loss": 1.1215,
+      "step": 21641
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.8038751729987945e-06,
+      "loss": 0.9357,
+      "step": 21642
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.794946202955489e-06,
+      "loss": 1.0261,
+      "step": 21643
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.786017232912184e-06,
+      "loss": 0.9513,
+      "step": 21644
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.777088262868879e-06,
+      "loss": 1.0236,
+      "step": 21645
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.7681592928255725e-06,
+      "loss": 0.9876,
+      "step": 21646
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.759230322782267e-06,
+      "loss": 1.0119,
+      "step": 21647
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.750301352738962e-06,
+      "loss": 1.0367,
+      "step": 21648
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.741372382695656e-06,
+      "loss": 1.1807,
+      "step": 21649
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.7324434126523505e-06,
+      "loss": 0.9647,
+      "step": 21650
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.723514442609045e-06,
+      "loss": 0.9385,
+      "step": 21651
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.71458547256574e-06,
+      "loss": 1.0856,
+      "step": 21652
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.705656502522434e-06,
+      "loss": 0.9895,
+      "step": 21653
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.6967275324791285e-06,
+      "loss": 1.1214,
+      "step": 21654
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.687798562435823e-06,
+      "loss": 0.9798,
+      "step": 21655
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.678869592392519e-06,
+      "loss": 1.0482,
+      "step": 21656
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.669940622349212e-06,
+      "loss": 1.1425,
+      "step": 21657
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.6610116523059065e-06,
+      "loss": 1.1493,
+      "step": 21658
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.652082682262602e-06,
+      "loss": 1.0255,
+      "step": 21659
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.643153712219295e-06,
+      "loss": 1.2065,
+      "step": 21660
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.63422474217599e-06,
+      "loss": 1.1194,
+      "step": 21661
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.6252957721326846e-06,
+      "loss": 1.0665,
+      "step": 21662
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.61636680208938e-06,
+      "loss": 0.9645,
+      "step": 21663
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.607437832046073e-06,
+      "loss": 1.0049,
+      "step": 21664
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.598508862002768e-06,
+      "loss": 0.9675,
+      "step": 21665
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.5895798919594634e-06,
+      "loss": 1.0366,
+      "step": 21666
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.5806509219161564e-06,
+      "loss": 0.917,
+      "step": 21667
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.571721951872851e-06,
+      "loss": 0.9519,
+      "step": 21668
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.562792981829547e-06,
+      "loss": 1.0596,
+      "step": 21669
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.5538640117862414e-06,
+      "loss": 1.036,
+      "step": 21670
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.5449350417429345e-06,
+      "loss": 0.9567,
+      "step": 21671
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.53600607169963e-06,
+      "loss": 1.0933,
+      "step": 21672
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.527077101656325e-06,
+      "loss": 0.9341,
+      "step": 21673
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.5181481316130195e-06,
+      "loss": 1.0285,
+      "step": 21674
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.5092191615697125e-06,
+      "loss": 1.0296,
+      "step": 21675
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.500290191526408e-06,
+      "loss": 0.9807,
+      "step": 21676
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.491361221483103e-06,
+      "loss": 1.0548,
+      "step": 21677
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.482432251439796e-06,
+      "loss": 1.1017,
+      "step": 21678
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.473503281396491e-06,
+      "loss": 1.0643,
+      "step": 21679
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.464574311353186e-06,
+      "loss": 1.0092,
+      "step": 21680
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.455645341309881e-06,
+      "loss": 0.9953,
+      "step": 21681
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.446716371266575e-06,
+      "loss": 1.1141,
+      "step": 21682
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.437787401223269e-06,
+      "loss": 1.0144,
+      "step": 21683
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.428858431179964e-06,
+      "loss": 1.0834,
+      "step": 21684
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.419929461136658e-06,
+      "loss": 0.9701,
+      "step": 21685
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.411000491093353e-06,
+      "loss": 0.9935,
+      "step": 21686
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.402071521050047e-06,
+      "loss": 1.0233,
+      "step": 21687
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.393142551006742e-06,
+      "loss": 0.9654,
+      "step": 21688
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.384213580963436e-06,
+      "loss": 1.1902,
+      "step": 21689
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.375284610920131e-06,
+      "loss": 1.0687,
+      "step": 21690
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.366355640876825e-06,
+      "loss": 1.0173,
+      "step": 21691
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.35742667083352e-06,
+      "loss": 0.9114,
+      "step": 21692
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.348497700790214e-06,
+      "loss": 0.9908,
+      "step": 21693
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.339568730746909e-06,
+      "loss": 1.0508,
+      "step": 21694
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.330639760703603e-06,
+      "loss": 1.3252,
+      "step": 21695
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.321710790660297e-06,
+      "loss": 0.9525,
+      "step": 21696
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.312781820616992e-06,
+      "loss": 1.0602,
+      "step": 21697
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.303852850573687e-06,
+      "loss": 0.9475,
+      "step": 21698
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.294923880530381e-06,
+      "loss": 0.9262,
+      "step": 21699
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.285994910487075e-06,
+      "loss": 1.0332,
+      "step": 21700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.27706594044377e-06,
+      "loss": 1.0719,
+      "step": 21701
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.268136970400465e-06,
+      "loss": 1.08,
+      "step": 21702
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.259208000357159e-06,
+      "loss": 1.1754,
+      "step": 21703
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.250279030313853e-06,
+      "loss": 1.0508,
+      "step": 21704
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.241350060270548e-06,
+      "loss": 0.9753,
+      "step": 21705
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.232421090227243e-06,
+      "loss": 0.9246,
+      "step": 21706
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.2234921201839374e-06,
+      "loss": 1.2668,
+      "step": 21707
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.214563150140631e-06,
+      "loss": 0.9077,
+      "step": 21708
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.205634180097326e-06,
+      "loss": 1.0731,
+      "step": 21709
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.196705210054021e-06,
+      "loss": 1.0902,
+      "step": 21710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.187776240010715e-06,
+      "loss": 1.0457,
+      "step": 21711
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.178847269967409e-06,
+      "loss": 0.9745,
+      "step": 21712
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.169918299924104e-06,
+      "loss": 1.0567,
+      "step": 21713
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.160989329880799e-06,
+      "loss": 0.9943,
+      "step": 21714
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.152060359837493e-06,
+      "loss": 0.9865,
+      "step": 21715
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.143131389794188e-06,
+      "loss": 1.0327,
+      "step": 21716
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.134202419750882e-06,
+      "loss": 1.1252,
+      "step": 21717
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.125273449707577e-06,
+      "loss": 1.0935,
+      "step": 21718
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.116344479664271e-06,
+      "loss": 1.0405,
+      "step": 21719
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.107415509620965e-06,
+      "loss": 0.9725,
+      "step": 21720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.09848653957766e-06,
+      "loss": 0.8989,
+      "step": 21721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.089557569534354e-06,
+      "loss": 0.995,
+      "step": 21722
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.0806285994910495e-06,
+      "loss": 1.111,
+      "step": 21723
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.071699629447743e-06,
+      "loss": 1.0319,
+      "step": 21724
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.062770659404438e-06,
+      "loss": 1.0304,
+      "step": 21725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.053841689361133e-06,
+      "loss": 1.0161,
+      "step": 21726
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.0449127193178275e-06,
+      "loss": 1.0264,
+      "step": 21727
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.035983749274521e-06,
+      "loss": 1.0038,
+      "step": 21728
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.027054779231216e-06,
+      "loss": 0.9602,
+      "step": 21729
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.018125809187911e-06,
+      "loss": 1.0322,
+      "step": 21730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.009196839144605e-06,
+      "loss": 1.0247,
+      "step": 21731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.000267869101299e-06,
+      "loss": 0.9864,
+      "step": 21732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.991338899057994e-06,
+      "loss": 0.9755,
+      "step": 21733
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.982409929014689e-06,
+      "loss": 1.0009,
+      "step": 21734
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.973480958971383e-06,
+      "loss": 0.8258,
+      "step": 21735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.964551988928077e-06,
+      "loss": 1.0873,
+      "step": 21736
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.955623018884772e-06,
+      "loss": 1.0082,
+      "step": 21737
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.946694048841466e-06,
+      "loss": 1.0088,
+      "step": 21738
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.937765078798161e-06,
+      "loss": 0.9564,
+      "step": 21739
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.9288361087548554e-06,
+      "loss": 1.0617,
+      "step": 21740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.91990713871155e-06,
+      "loss": 1.1674,
+      "step": 21741
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.910978168668244e-06,
+      "loss": 0.9889,
+      "step": 21742
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.90204919862494e-06,
+      "loss": 0.9502,
+      "step": 21743
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.8931202285816334e-06,
+      "loss": 1.0224,
+      "step": 21744
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.884191258538327e-06,
+      "loss": 1.1352,
+      "step": 21745
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.875262288495022e-06,
+      "loss": 1.0026,
+      "step": 21746
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.866333318451717e-06,
+      "loss": 1.046,
+      "step": 21747
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.8574043484084115e-06,
+      "loss": 1.1261,
+      "step": 21748
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.848475378365105e-06,
+      "loss": 0.9738,
+      "step": 21749
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.839546408321801e-06,
+      "loss": 1.0066,
+      "step": 21750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.830617438278495e-06,
+      "loss": 1.1029,
+      "step": 21751
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.8216884682351895e-06,
+      "loss": 1.0075,
+      "step": 21752
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.812759498191884e-06,
+      "loss": 1.0698,
+      "step": 21753
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.803830528148578e-06,
+      "loss": 1.0577,
+      "step": 21754
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.794901558105273e-06,
+      "loss": 1.1217,
+      "step": 21755
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.7859725880619675e-06,
+      "loss": 1.0615,
+      "step": 21756
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.777043618018662e-06,
+      "loss": 1.158,
+      "step": 21757
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.768114647975356e-06,
+      "loss": 0.9121,
+      "step": 21758
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.759185677932051e-06,
+      "loss": 1.0092,
+      "step": 21759
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.7502567078887455e-06,
+      "loss": 1.0472,
+      "step": 21760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.74132773784544e-06,
+      "loss": 1.0221,
+      "step": 21761
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.732398767802134e-06,
+      "loss": 1.1016,
+      "step": 21762
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.723469797758829e-06,
+      "loss": 1.01,
+      "step": 21763
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.7145408277155235e-06,
+      "loss": 1.1534,
+      "step": 21764
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.705611857672217e-06,
+      "loss": 0.9946,
+      "step": 21765
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.696682887628912e-06,
+      "loss": 1.0076,
+      "step": 21766
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.687753917585607e-06,
+      "loss": 1.0497,
+      "step": 21767
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.6788249475423015e-06,
+      "loss": 0.926,
+      "step": 21768
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.669895977498995e-06,
+      "loss": 0.8651,
+      "step": 21769
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.660967007455691e-06,
+      "loss": 0.9524,
+      "step": 21770
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.652038037412385e-06,
+      "loss": 1.1641,
+      "step": 21771
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.643109067369079e-06,
+      "loss": 0.9862,
+      "step": 21772
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.634180097325773e-06,
+      "loss": 1.0339,
+      "step": 21773
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.625251127282468e-06,
+      "loss": 1.0392,
+      "step": 21774
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.616322157239163e-06,
+      "loss": 1.0052,
+      "step": 21775
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.607393187195857e-06,
+      "loss": 1.0409,
+      "step": 21776
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.598464217152552e-06,
+      "loss": 1.1397,
+      "step": 21777
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.589535247109246e-06,
+      "loss": 0.9521,
+      "step": 21778
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.580606277065941e-06,
+      "loss": 0.9921,
+      "step": 21779
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.571677307022636e-06,
+      "loss": 1.2316,
+      "step": 21780
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.5627483369793295e-06,
+      "loss": 0.9959,
+      "step": 21781
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.553819366936024e-06,
+      "loss": 1.0824,
+      "step": 21782
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.544890396892719e-06,
+      "loss": 0.9917,
+      "step": 21783
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.535961426849414e-06,
+      "loss": 0.9453,
+      "step": 21784
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.5270324568061075e-06,
+      "loss": 1.1792,
+      "step": 21785
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.518103486762802e-06,
+      "loss": 1.0727,
+      "step": 21786
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.509174516719497e-06,
+      "loss": 1.1133,
+      "step": 21787
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.500245546676191e-06,
+      "loss": 1.0902,
+      "step": 21788
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4913165766328855e-06,
+      "loss": 0.978,
+      "step": 21789
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.48238760658958e-06,
+      "loss": 1.0049,
+      "step": 21790
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.473458636546275e-06,
+      "loss": 1.054,
+      "step": 21791
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.464529666502969e-06,
+      "loss": 1.0691,
+      "step": 21792
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4556006964596635e-06,
+      "loss": 0.9661,
+      "step": 21793
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.446671726416358e-06,
+      "loss": 0.9129,
+      "step": 21794
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.437742756373053e-06,
+      "loss": 0.9424,
+      "step": 21795
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.428813786329747e-06,
+      "loss": 0.9837,
+      "step": 21796
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.4198848162864415e-06,
+      "loss": 0.9484,
+      "step": 21797
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.410955846243136e-06,
+      "loss": 1.0919,
+      "step": 21798
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.40202687619983e-06,
+      "loss": 1.039,
+      "step": 21799
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.393097906156526e-06,
+      "loss": 1.0171,
+      "step": 21800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.3841689361132195e-06,
+      "loss": 1.0127,
+      "step": 21801
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.375239966069914e-06,
+      "loss": 1.0419,
+      "step": 21802
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.366310996026608e-06,
+      "loss": 0.9112,
+      "step": 21803
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.357382025983304e-06,
+      "loss": 1.1037,
+      "step": 21804
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.3484530559399975e-06,
+      "loss": 0.9506,
+      "step": 21805
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.339524085896691e-06,
+      "loss": 1.0108,
+      "step": 21806
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.330595115853387e-06,
+      "loss": 0.9746,
+      "step": 21807
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.321666145810081e-06,
+      "loss": 1.0165,
+      "step": 21808
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.3127371757667756e-06,
+      "loss": 1.0036,
+      "step": 21809
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.30380820572347e-06,
+      "loss": 0.9499,
+      "step": 21810
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.294879235680165e-06,
+      "loss": 0.9859,
+      "step": 21811
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.285950265636859e-06,
+      "loss": 1.0006,
+      "step": 21812
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.2770212955935536e-06,
+      "loss": 1.0658,
+      "step": 21813
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.268092325550248e-06,
+      "loss": 1.0283,
+      "step": 21814
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.259163355506942e-06,
+      "loss": 1.0231,
+      "step": 21815
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.250234385463637e-06,
+      "loss": 0.9981,
+      "step": 21816
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.241305415420332e-06,
+      "loss": 0.9196,
+      "step": 21817
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.232376445377026e-06,
+      "loss": 1.0449,
+      "step": 21818
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.22344747533372e-06,
+      "loss": 1.0225,
+      "step": 21819
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.214518505290415e-06,
+      "loss": 0.9408,
+      "step": 21820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.20558953524711e-06,
+      "loss": 0.9998,
+      "step": 21821
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.196660565203804e-06,
+      "loss": 1.2173,
+      "step": 21822
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.187731595160498e-06,
+      "loss": 1.001,
+      "step": 21823
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.178802625117193e-06,
+      "loss": 0.9516,
+      "step": 21824
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.169873655073888e-06,
+      "loss": 1.0221,
+      "step": 21825
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.1609446850305815e-06,
+      "loss": 0.9515,
+      "step": 21826
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.152015714987277e-06,
+      "loss": 1.0484,
+      "step": 21827
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.143086744943971e-06,
+      "loss": 0.9264,
+      "step": 21828
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.134157774900666e-06,
+      "loss": 0.9361,
+      "step": 21829
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.1252288048573595e-06,
+      "loss": 0.9897,
+      "step": 21830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.116299834814055e-06,
+      "loss": 0.9883,
+      "step": 21831
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.107370864770749e-06,
+      "loss": 1.0442,
+      "step": 21832
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.098441894727443e-06,
+      "loss": 1.0629,
+      "step": 21833
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.089512924684138e-06,
+      "loss": 0.952,
+      "step": 21834
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.080583954640832e-06,
+      "loss": 0.9734,
+      "step": 21835
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.071654984597527e-06,
+      "loss": 1.0081,
+      "step": 21836
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.062726014554222e-06,
+      "loss": 1.0968,
+      "step": 21837
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.053797044510916e-06,
+      "loss": 0.956,
+      "step": 21838
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.04486807446761e-06,
+      "loss": 1.1016,
+      "step": 21839
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.035939104424305e-06,
+      "loss": 1.1216,
+      "step": 21840
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.027010134381e-06,
+      "loss": 0.9579,
+      "step": 21841
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.0180811643376935e-06,
+      "loss": 0.9772,
+      "step": 21842
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.009152194294388e-06,
+      "loss": 1.0409,
+      "step": 21843
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.000223224251083e-06,
+      "loss": 0.9704,
+      "step": 21844
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.991294254207778e-06,
+      "loss": 1.0995,
+      "step": 21845
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.9823652841644716e-06,
+      "loss": 0.9863,
+      "step": 21846
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.973436314121166e-06,
+      "loss": 1.2603,
+      "step": 21847
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.964507344077861e-06,
+      "loss": 1.0108,
+      "step": 21848
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.955578374034555e-06,
+      "loss": 1.0118,
+      "step": 21849
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.94664940399125e-06,
+      "loss": 1.1795,
+      "step": 21850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.937720433947944e-06,
+      "loss": 1.0321,
+      "step": 21851
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.928791463904639e-06,
+      "loss": 1.0897,
+      "step": 21852
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.919862493861333e-06,
+      "loss": 0.9992,
+      "step": 21853
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.9109335238180284e-06,
+      "loss": 1.0567,
+      "step": 21854
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.902004553774722e-06,
+      "loss": 0.9332,
+      "step": 21855
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.893075583731417e-06,
+      "loss": 1.0784,
+      "step": 21856
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.884146613688111e-06,
+      "loss": 0.9046,
+      "step": 21857
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.875217643644806e-06,
+      "loss": 1.0033,
+      "step": 21858
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.8662886736015e-06,
+      "loss": 1.1345,
+      "step": 21859
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.857359703558194e-06,
+      "loss": 0.9366,
+      "step": 21860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.84843073351489e-06,
+      "loss": 1.0242,
+      "step": 21861
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.839501763471584e-06,
+      "loss": 1.131,
+      "step": 21862
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.830572793428278e-06,
+      "loss": 1.0427,
+      "step": 21863
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.821643823384973e-06,
+      "loss": 0.9988,
+      "step": 21864
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.812714853341668e-06,
+      "loss": 1.0027,
+      "step": 21865
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.803785883298362e-06,
+      "loss": 1.0952,
+      "step": 21866
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.794856913255056e-06,
+      "loss": 0.9652,
+      "step": 21867
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.785927943211751e-06,
+      "loss": 0.9825,
+      "step": 21868
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.776998973168445e-06,
+      "loss": 0.9993,
+      "step": 21869
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.76807000312514e-06,
+      "loss": 1.0023,
+      "step": 21870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.759141033081834e-06,
+      "loss": 0.933,
+      "step": 21871
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.750212063038529e-06,
+      "loss": 0.931,
+      "step": 21872
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.741283092995223e-06,
+      "loss": 0.9637,
+      "step": 21873
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.732354122951918e-06,
+      "loss": 0.9912,
+      "step": 21874
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.723425152908612e-06,
+      "loss": 0.9882,
+      "step": 21875
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.714496182865306e-06,
+      "loss": 1.1148,
+      "step": 21876
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.705567212822001e-06,
+      "loss": 1.1585,
+      "step": 21877
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.696638242778696e-06,
+      "loss": 1.0768,
+      "step": 21878
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.68770927273539e-06,
+      "loss": 1.1503,
+      "step": 21879
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.678780302692084e-06,
+      "loss": 1.0446,
+      "step": 21880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.66985133264878e-06,
+      "loss": 1.0449,
+      "step": 21881
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.660922362605474e-06,
+      "loss": 0.9908,
+      "step": 21882
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.651993392562168e-06,
+      "loss": 1.0714,
+      "step": 21883
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.643064422518862e-06,
+      "loss": 1.0543,
+      "step": 21884
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.634135452475557e-06,
+      "loss": 1.0722,
+      "step": 21885
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.625206482432252e-06,
+      "loss": 1.1825,
+      "step": 21886
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.616277512388946e-06,
+      "loss": 0.9941,
+      "step": 21887
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.607348542345641e-06,
+      "loss": 1.095,
+      "step": 21888
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.598419572302335e-06,
+      "loss": 1.0228,
+      "step": 21889
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.58949060225903e-06,
+      "loss": 1.02,
+      "step": 21890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5805616322157244e-06,
+      "loss": 0.9901,
+      "step": 21891
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.571632662172418e-06,
+      "loss": 1.1026,
+      "step": 21892
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.562703692129113e-06,
+      "loss": 0.9778,
+      "step": 21893
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.553774722085808e-06,
+      "loss": 1.1438,
+      "step": 21894
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5448457520425025e-06,
+      "loss": 1.0128,
+      "step": 21895
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.535916781999196e-06,
+      "loss": 1.1165,
+      "step": 21896
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.526987811955891e-06,
+      "loss": 1.0474,
+      "step": 21897
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.518058841912586e-06,
+      "loss": 1.0574,
+      "step": 21898
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5091298718692805e-06,
+      "loss": 1.1193,
+      "step": 21899
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.500200901825974e-06,
+      "loss": 0.9407,
+      "step": 21900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.491271931782669e-06,
+      "loss": 1.004,
+      "step": 21901
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.482342961739364e-06,
+      "loss": 1.0877,
+      "step": 21902
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.473413991696058e-06,
+      "loss": 1.0675,
+      "step": 21903
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.464485021652752e-06,
+      "loss": 1.9859,
+      "step": 21904
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.455556051609447e-06,
+      "loss": 1.0023,
+      "step": 21905
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.446627081566142e-06,
+      "loss": 0.9527,
+      "step": 21906
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.437698111522836e-06,
+      "loss": 1.0619,
+      "step": 21907
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.428769141479531e-06,
+      "loss": 1.2066,
+      "step": 21908
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.419840171436225e-06,
+      "loss": 1.1147,
+      "step": 21909
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.410911201392919e-06,
+      "loss": 1.3236,
+      "step": 21910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4019822313496145e-06,
+      "loss": 1.1262,
+      "step": 21911
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.393053261306308e-06,
+      "loss": 1.0136,
+      "step": 21912
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.384124291263003e-06,
+      "loss": 1.0119,
+      "step": 21913
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.375195321219697e-06,
+      "loss": 1.0143,
+      "step": 21914
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.3662663511763925e-06,
+      "loss": 0.9368,
+      "step": 21915
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.357337381133086e-06,
+      "loss": 0.948,
+      "step": 21916
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.348408411089781e-06,
+      "loss": 0.9389,
+      "step": 21917
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.339479441046476e-06,
+      "loss": 1.1279,
+      "step": 21918
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.33055047100317e-06,
+      "loss": 1.0148,
+      "step": 21919
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.321621500959864e-06,
+      "loss": 1.012,
+      "step": 21920
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.312692530916559e-06,
+      "loss": 0.9284,
+      "step": 21921
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.303763560873254e-06,
+      "loss": 1.0088,
+      "step": 21922
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.294834590829948e-06,
+      "loss": 1.0573,
+      "step": 21923
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2859056207866424e-06,
+      "loss": 1.1007,
+      "step": 21924
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.276976650743337e-06,
+      "loss": 1.0312,
+      "step": 21925
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.268047680700032e-06,
+      "loss": 1.1046,
+      "step": 21926
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.259118710656726e-06,
+      "loss": 1.0551,
+      "step": 21927
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2501897406134204e-06,
+      "loss": 0.9662,
+      "step": 21928
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.241260770570115e-06,
+      "loss": 1.0138,
+      "step": 21929
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.232331800526809e-06,
+      "loss": 1.0312,
+      "step": 21930
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.223402830483504e-06,
+      "loss": 1.0463,
+      "step": 21931
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2144738604401985e-06,
+      "loss": 0.9635,
+      "step": 21932
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.205544890396893e-06,
+      "loss": 1.1038,
+      "step": 21933
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.196615920353587e-06,
+      "loss": 0.9539,
+      "step": 21934
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.187686950310283e-06,
+      "loss": 1.112,
+      "step": 21935
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1787579802669765e-06,
+      "loss": 1.1399,
+      "step": 21936
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.16982901022367e-06,
+      "loss": 1.003,
+      "step": 21937
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.160900040180366e-06,
+      "loss": 0.9189,
+      "step": 21938
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.15197107013706e-06,
+      "loss": 1.0558,
+      "step": 21939
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1430421000937545e-06,
+      "loss": 0.9736,
+      "step": 21940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.134113130050448e-06,
+      "loss": 1.2155,
+      "step": 21941
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.125184160007144e-06,
+      "loss": 1.0065,
+      "step": 21942
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.116255189963838e-06,
+      "loss": 0.9913,
+      "step": 21943
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1073262199205325e-06,
+      "loss": 1.1333,
+      "step": 21944
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.098397249877227e-06,
+      "loss": 1.0788,
+      "step": 21945
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.089468279833921e-06,
+      "loss": 0.9764,
+      "step": 21946
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.080539309790616e-06,
+      "loss": 1.032,
+      "step": 21947
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0716103397473105e-06,
+      "loss": 1.047,
+      "step": 21948
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.062681369704005e-06,
+      "loss": 0.9478,
+      "step": 21949
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.053752399660699e-06,
+      "loss": 1.1125,
+      "step": 21950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.044823429617394e-06,
+      "loss": 0.9069,
+      "step": 21951
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0358944595740885e-06,
+      "loss": 0.9634,
+      "step": 21952
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.026965489530782e-06,
+      "loss": 0.9489,
+      "step": 21953
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.018036519487477e-06,
+      "loss": 1.0683,
+      "step": 21954
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.009107549444172e-06,
+      "loss": 1.1308,
+      "step": 21955
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0001785794008666e-06,
+      "loss": 0.9832,
+      "step": 21956
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.99124960935756e-06,
+      "loss": 1.0126,
+      "step": 21957
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.982320639314255e-06,
+      "loss": 0.9279,
+      "step": 21958
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.97339166927095e-06,
+      "loss": 0.7867,
+      "step": 21959
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9644626992276446e-06,
+      "loss": 1.0078,
+      "step": 21960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9555337291843384e-06,
+      "loss": 0.9058,
+      "step": 21961
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.946604759141033e-06,
+      "loss": 1.0144,
+      "step": 21962
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.937675789097728e-06,
+      "loss": 0.9183,
+      "step": 21963
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.928746819054422e-06,
+      "loss": 0.9383,
+      "step": 21964
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.919817849011117e-06,
+      "loss": 0.9996,
+      "step": 21965
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.910888878967811e-06,
+      "loss": 1.0612,
+      "step": 21966
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.901959908924506e-06,
+      "loss": 0.9818,
+      "step": 21967
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8930309388812e-06,
+      "loss": 1.0016,
+      "step": 21968
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.884101968837895e-06,
+      "loss": 0.8929,
+      "step": 21969
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.875172998794589e-06,
+      "loss": 0.9731,
+      "step": 21970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.866244028751283e-06,
+      "loss": 1.049,
+      "step": 21971
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.857315058707979e-06,
+      "loss": 1.1291,
+      "step": 21972
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8483860886646725e-06,
+      "loss": 1.0916,
+      "step": 21973
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.839457118621367e-06,
+      "loss": 0.9774,
+      "step": 21974
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.830528148578062e-06,
+      "loss": 0.9595,
+      "step": 21975
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.821599178534757e-06,
+      "loss": 0.9722,
+      "step": 21976
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8126702084914505e-06,
+      "loss": 1.1909,
+      "step": 21977
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8037412384481456e-06,
+      "loss": 1.0687,
+      "step": 21978
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.7948122684048395e-06,
+      "loss": 1.1054,
+      "step": 21979
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.785883298361534e-06,
+      "loss": 0.9595,
+      "step": 21980
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.776954328318229e-06,
+      "loss": 1.1005,
+      "step": 21981
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.768025358274923e-06,
+      "loss": 0.9049,
+      "step": 21982
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.759096388231618e-06,
+      "loss": 1.0461,
+      "step": 21983
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.750167418188312e-06,
+      "loss": 0.998,
+      "step": 21984
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.741238448145007e-06,
+      "loss": 1.0619,
+      "step": 21985
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.7323094781017012e-06,
+      "loss": 1.4395,
+      "step": 21986
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.723380508058396e-06,
+      "loss": 1.0413,
+      "step": 21987
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.7144515380150902e-06,
+      "loss": 1.109,
+      "step": 21988
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.7055225679717845e-06,
+      "loss": 1.257,
+      "step": 21989
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6965935979284793e-06,
+      "loss": 1.0962,
+      "step": 21990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6876646278851735e-06,
+      "loss": 1.0594,
+      "step": 21991
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6787356578418683e-06,
+      "loss": 1.3445,
+      "step": 21992
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6698066877985626e-06,
+      "loss": 0.9322,
+      "step": 21993
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6608777177552573e-06,
+      "loss": 1.0146,
+      "step": 21994
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6519487477119516e-06,
+      "loss": 1.1259,
+      "step": 21995
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6430197776686463e-06,
+      "loss": 0.9351,
+      "step": 21996
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6340908076253406e-06,
+      "loss": 0.9783,
+      "step": 21997
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.625161837582035e-06,
+      "loss": 0.9799,
+      "step": 21998
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6162328675387296e-06,
+      "loss": 0.9223,
+      "step": 21999
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.607303897495424e-06,
+      "loss": 1.0113,
+      "step": 22000
+    },
+    {
+      "epoch": 2.95,
+      "eval_loss": 1.024954080581665,
+      "eval_runtime": 3076.4844,
+      "eval_samples_per_second": 6.113,
+      "eval_steps_per_second": 0.764,
+      "step": 22000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 22404,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 2000,
+  "total_flos": 2.0266812421662206e+20,
+  "train_batch_size": 12,
+  "trial_name": null,
+  "trial_params": null
+}