diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,90027 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.881821603312006,
+  "eval_steps": 10000,
+  "global_step": 15000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999875730626e-05,
+      "loss": 2.7188,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999502922508e-05,
+      "loss": 2.75,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999998881575648e-05,
+      "loss": 2.7969,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999998011690067e-05,
+      "loss": 3.0156,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999999689326579e-05,
+      "loss": 1.4297,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999995526302832e-05,
+      "loss": 1.5078,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999993910801242e-05,
+      "loss": 2.6719,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999992046761053e-05,
+      "loss": 1.4922,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999989934182315e-05,
+      "loss": 2.2812,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999987573065076e-05,
+      "loss": 2.0469,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999984963409394e-05,
+      "loss": 3.4531,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999998210521534e-05,
+      "loss": 1.5391,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997899848298e-05,
+      "loss": 2.2656,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999975643212388e-05,
+      "loss": 1.3984,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999972039403656e-05,
+      "loss": 1.8203,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999996818705687e-05,
+      "loss": 2.0156,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999996408617212e-05,
+      "loss": 1.4688,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999959736749518e-05,
+      "loss": 1.6406,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999955138789166e-05,
+      "loss": 2.0625,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999995029229118e-05,
+      "loss": 1.8984,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999945197255682e-05,
+      "loss": 1.3047,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999939853682794e-05,
+      "loss": 1.5703,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999934261572652e-05,
+      "loss": 2.6094,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999992842092539e-05,
+      "loss": 3.1719,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999922331741164e-05,
+      "loss": 1.9766,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999915994020115e-05,
+      "loss": 1.375,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999909407762406e-05,
+      "loss": 1.9531,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99999025729682e-05,
+      "loss": 3.0469,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999895489637666e-05,
+      "loss": 2.8281,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999888157770978e-05,
+      "loss": 1.8125,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999988057736832e-05,
+      "loss": 2.5781,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999987274842988e-05,
+      "loss": 2.1094,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999864670955854e-05,
+      "loss": 2.1719,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999856344946444e-05,
+      "loss": 2.1719,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999847770401853e-05,
+      "loss": 2.375,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999838947322297e-05,
+      "loss": 2.5312,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999982987570799e-05,
+      "loss": 1.9844,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999982055555917e-05,
+      "loss": 2.5,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999810986876052e-05,
+      "loss": 2.25,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999980116965889e-05,
+      "loss": 1.9219,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999791103907917e-05,
+      "loss": 2.3594,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999780789623385e-05,
+      "loss": 2.4375,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999770226805553e-05,
+      "loss": 2.6875,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999975941545468e-05,
+      "loss": 2.4219,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999748355571037e-05,
+      "loss": 2.4062,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999737047154902e-05,
+      "loss": 2.3906,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999972549020655e-05,
+      "loss": 2.3906,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999713684726273e-05,
+      "loss": 2.3594,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999701630714363e-05,
+      "loss": 2.375,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999689328171115e-05,
+      "loss": 2.6094,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999676777096847e-05,
+      "loss": 2.0938,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999663977491857e-05,
+      "loss": 1.9844,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999965092935647e-05,
+      "loss": 1.2031,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999637632691008e-05,
+      "loss": 1.3594,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999624087495806e-05,
+      "loss": 1.7734,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999610293771194e-05,
+      "loss": 1.5938,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999959625151752e-05,
+      "loss": 1.2344,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999581960735132e-05,
+      "loss": 1.25,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999956742142438e-05,
+      "loss": 1.6016,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999552633585635e-05,
+      "loss": 0.9258,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999537597219257e-05,
+      "loss": 1.5547,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999522312325622e-05,
+      "loss": 1.0703,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999506778905112e-05,
+      "loss": 2.0469,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999490996958107e-05,
+      "loss": 0.9297,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999474966485006e-05,
+      "loss": 1.1484,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999458687486203e-05,
+      "loss": 1.9531,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999442159962106e-05,
+      "loss": 1.0078,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999942538391312e-05,
+      "loss": 0.8945,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999408359339673e-05,
+      "loss": 2.1406,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999391086242176e-05,
+      "loss": 1.8203,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999373564621066e-05,
+      "loss": 2.2656,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999355794476773e-05,
+      "loss": 1.2266,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999337775809743e-05,
+      "loss": 1.5156,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999319508620424e-05,
+      "loss": 2.0,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999300992909266e-05,
+      "loss": 1.8438,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999928222867673e-05,
+      "loss": 1.5469,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999263215923286e-05,
+      "loss": 1.0,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999243954649406e-05,
+      "loss": 2.1719,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999224444855563e-05,
+      "loss": 1.3516,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999920468654225e-05,
+      "loss": 1.3672,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999184679709953e-05,
+      "loss": 1.4219,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999916442435917e-05,
+      "loss": 1.5781,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999143920490406e-05,
+      "loss": 1.2422,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999123168104167e-05,
+      "loss": 2.0,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999102167200975e-05,
+      "loss": 1.4219,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999080917781345e-05,
+      "loss": 1.2188,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999059419845806e-05,
+      "loss": 1.3672,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99990376733949e-05,
+      "loss": 2.25,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999901567842916e-05,
+      "loss": 1.5469,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998993434949132e-05,
+      "loss": 1.6484,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998970942955377e-05,
+      "loss": 1.9297,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998948202448444e-05,
+      "loss": 2.1406,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998925213428905e-05,
+      "loss": 1.5,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998901975897327e-05,
+      "loss": 1.9297,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999887848985429e-05,
+      "loss": 1.7266,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999885475530038e-05,
+      "loss": 1.8438,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999883077223618e-05,
+      "loss": 1.3359,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998806540662294e-05,
+      "loss": 1.5234,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998782060579322e-05,
+      "loss": 1.3594,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998757331987867e-05,
+      "loss": 1.2266,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999873235488855e-05,
+      "loss": 2.2188,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998707129281986e-05,
+      "loss": 1.5781,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999868165516881e-05,
+      "loss": 1.1719,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998655932549647e-05,
+      "loss": 0.9648,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999862996142514e-05,
+      "loss": 1.3984,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998603741795938e-05,
+      "loss": 1.5703,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998577273662683e-05,
+      "loss": 0.7891,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998550557026042e-05,
+      "loss": 0.832,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999852359188668e-05,
+      "loss": 0.6289,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998496378245257e-05,
+      "loss": 2.1875,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998468916102457e-05,
+      "loss": 1.4531,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999844120545896e-05,
+      "loss": 0.7227,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998413246315457e-05,
+      "loss": 1.2578,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999838503867264e-05,
+      "loss": 1.9609,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998356582531217e-05,
+      "loss": 1.0234,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998327877891884e-05,
+      "loss": 1.6172,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998298924755358e-05,
+      "loss": 0.7656,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998269723122367e-05,
+      "loss": 0.3281,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998240272993627e-05,
+      "loss": 1.4297,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998210574369875e-05,
+      "loss": 0.8164,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998180627251848e-05,
+      "loss": 1.7266,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999815043164029e-05,
+      "loss": 1.6641,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999811998753595e-05,
+      "loss": 1.9219,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998089294939587e-05,
+      "loss": 0.9766,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999805835385196e-05,
+      "loss": 2.0156,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998027164273845e-05,
+      "loss": 1.4844,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997995726206012e-05,
+      "loss": 1.7734,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997964039649245e-05,
+      "loss": 1.7969,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999793210460433e-05,
+      "loss": 1.7891,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997899921072058e-05,
+      "loss": 1.4062,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997867489053232e-05,
+      "loss": 1.6641,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997834808548656e-05,
+      "loss": 1.7734,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997801879559148e-05,
+      "loss": 1.1719,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999776870208552e-05,
+      "loss": 2.0469,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99977352761286e-05,
+      "loss": 0.9922,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997701601689216e-05,
+      "loss": 0.9453,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997667678768205e-05,
+      "loss": 1.0,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999763350736642e-05,
+      "loss": 0.7656,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997599087484694e-05,
+      "loss": 0.9531,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999756441912389e-05,
+      "loss": 1.2578,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997529502284875e-05,
+      "loss": 2.0312,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999749433696851e-05,
+      "loss": 2.2031,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999745892317567e-05,
+      "loss": 1.0078,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997423260907238e-05,
+      "loss": 1.2109,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99973873501641e-05,
+      "loss": 1.4766,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999735119094714e-05,
+      "loss": 1.2422,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999731478325727e-05,
+      "loss": 1.8438,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999727812709539e-05,
+      "loss": 1.1641,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997241222462403e-05,
+      "loss": 1.75,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999720406935924e-05,
+      "loss": 0.8711,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999716666778681e-05,
+      "loss": 1.4453,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997129017746053e-05,
+      "loss": 0.9492,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997091119237903e-05,
+      "loss": 0.7227,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997052972263296e-05,
+      "loss": 1.4219,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999701457682319e-05,
+      "loss": 0.8594,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999697593291853e-05,
+      "loss": 1.0078,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996937040550282e-05,
+      "loss": 1.3516,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999689789971941e-05,
+      "loss": 0.9102,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999685851042689e-05,
+      "loss": 1.4766,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99968188726737e-05,
+      "loss": 1.7266,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999677898646082e-05,
+      "loss": 0.8555,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996738851789242e-05,
+      "loss": 0.9297,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999669846865997e-05,
+      "loss": 0.9727,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999665783707401e-05,
+      "loss": 0.9336,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996616957032357e-05,
+      "loss": 2.0938,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999657582853604e-05,
+      "loss": 1.125,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996534451586082e-05,
+      "loss": 1.1875,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99964928261835e-05,
+      "loss": 0.5938,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996450952329338e-05,
+      "loss": 0.6484,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996408830024637e-05,
+      "loss": 1.6328,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996366459270437e-05,
+      "loss": 1.1172,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996323840067795e-05,
+      "loss": 2.4531,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996280972417773e-05,
+      "loss": 1.0312,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999623785632143e-05,
+      "loss": 1.1719,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996194491779845e-05,
+      "loss": 0.9844,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999615087879409e-05,
+      "loss": 1.2812,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996107017365255e-05,
+      "loss": 1.6562,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996062907494423e-05,
+      "loss": 1.2422,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996018549182694e-05,
+      "loss": 1.3828,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995973942431167e-05,
+      "loss": 1.6719,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995929087240957e-05,
+      "loss": 1.7266,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995883983613174e-05,
+      "loss": 1.9375,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995838631548944e-05,
+      "loss": 1.2422,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995793031049385e-05,
+      "loss": 0.7891,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995747182115642e-05,
+      "loss": 1.1406,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995701084748843e-05,
+      "loss": 1.4688,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999565473895014e-05,
+      "loss": 1.8125,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999560814472069e-05,
+      "loss": 0.9531,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995561302061638e-05,
+      "loss": 1.6406,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999551421097416e-05,
+      "loss": 1.7031,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995466871459418e-05,
+      "loss": 1.6562,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995419283518594e-05,
+      "loss": 1.5469,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995371447152873e-05,
+      "loss": 1.0312,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995323362363434e-05,
+      "loss": 1.3906,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999527502915148e-05,
+      "loss": 1.7656,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995226447518212e-05,
+      "loss": 1.2188,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995177617464835e-05,
+      "loss": 1.2656,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999512853899256e-05,
+      "loss": 1.3125,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995079212102614e-05,
+      "loss": 1.4062,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9995029636796216e-05,
+      "loss": 1.1562,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994979813074607e-05,
+      "loss": 1.2578,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994929740939014e-05,
+      "loss": 1.4531,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994879420390688e-05,
+      "loss": 1.3047,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999482885143088e-05,
+      "loss": 0.8672,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994778034060845e-05,
+      "loss": 1.9609,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999472696828185e-05,
+      "loss": 1.625,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994675654095154e-05,
+      "loss": 0.8594,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994624091502044e-05,
+      "loss": 0.7422,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994572280503797e-05,
+      "loss": 1.3672,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99945202211017e-05,
+      "loss": 0.7148,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994467913297046e-05,
+      "loss": 1.0938,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994415357091137e-05,
+      "loss": 1.375,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994362552485275e-05,
+      "loss": 0.5586,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999430949948078e-05,
+      "loss": 1.2969,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999425619807897e-05,
+      "loss": 1.1484,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994202648281157e-05,
+      "loss": 0.9414,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994148850088687e-05,
+      "loss": 1.3281,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999409480350289e-05,
+      "loss": 1.3125,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9994040508525108e-05,
+      "loss": 1.3359,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993985965156697e-05,
+      "loss": 1.1875,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993931173399008e-05,
+      "loss": 1.5703,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993876133253403e-05,
+      "loss": 1.3594,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999382084472125e-05,
+      "loss": 1.75,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993765307803922e-05,
+      "loss": 1.7656,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993709522502803e-05,
+      "loss": 1.4922,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993653488819274e-05,
+      "loss": 2.5781,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993597206754735e-05,
+      "loss": 1.7344,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999354067631058e-05,
+      "loss": 1.6172,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993483897488214e-05,
+      "loss": 0.8789,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999342687028905e-05,
+      "loss": 1.3984,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99933695947145e-05,
+      "loss": 1.0312,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993312070765994e-05,
+      "loss": 1.0312,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999325429844496e-05,
+      "loss": 0.5,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993196277752833e-05,
+      "loss": 0.9922,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993138008691057e-05,
+      "loss": 0.8594,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993079491261077e-05,
+      "loss": 1.3203,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999302072546435e-05,
+      "loss": 1.1875,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992961711302334e-05,
+      "loss": 1.8516,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99929024487765e-05,
+      "loss": 1.7734,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992842937888314e-05,
+      "loss": 1.2266,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992783178639265e-05,
+      "loss": 1.5469,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992723171030825e-05,
+      "loss": 1.5,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992662915064498e-05,
+      "loss": 0.9453,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992602410741775e-05,
+      "loss": 1.5938,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992541658064163e-05,
+      "loss": 1.3359,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992480657033168e-05,
+      "loss": 1.375,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999241940765031e-05,
+      "loss": 1.4141,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992357909917104e-05,
+      "loss": 1.2578,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992296163835092e-05,
+      "loss": 1.7031,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992234169405798e-05,
+      "loss": 1.4297,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992171926630767e-05,
+      "loss": 1.4297,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999210943551154e-05,
+      "loss": 1.0078,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9992046696049677e-05,
+      "loss": 0.9531,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991983708246736e-05,
+      "loss": 0.6445,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999192047210428e-05,
+      "loss": 1.4844,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991856987623883e-05,
+      "loss": 1.5078,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991793254807123e-05,
+      "loss": 1.5469,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999172927365558e-05,
+      "loss": 0.7188,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999166504417085e-05,
+      "loss": 0.7422,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991600566354523e-05,
+      "loss": 0.5664,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991535840208205e-05,
+      "loss": 0.9453,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991470865733508e-05,
+      "loss": 1.1406,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999140564293204e-05,
+      "loss": 0.7266,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991340171805426e-05,
+      "loss": 1.7734,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991274452355294e-05,
+      "loss": 0.793,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991208484583278e-05,
+      "loss": 1.6094,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991142268491013e-05,
+      "loss": 1.1406,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999107580408015e-05,
+      "loss": 1.2422,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991009091352337e-05,
+      "loss": 1.3438,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990942130309236e-05,
+      "loss": 1.8047,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990874920952507e-05,
+      "loss": 1.1172,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990807463283818e-05,
+      "loss": 1.0938,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990739757304855e-05,
+      "loss": 1.2344,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990671803017296e-05,
+      "loss": 1.3281,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990603600422824e-05,
+      "loss": 1.2344,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990535149523142e-05,
+      "loss": 1.7031,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999046645031995e-05,
+      "loss": 2.1406,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990397502814952e-05,
+      "loss": 1.7734,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990328307009864e-05,
+      "loss": 1.3047,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990258862906408e-05,
+      "loss": 1.9531,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990189170506303e-05,
+      "loss": 1.0938,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999011922981129e-05,
+      "loss": 0.9609,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99900490408231e-05,
+      "loss": 0.7422,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989978603543484e-05,
+      "loss": 1.3203,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989907917974185e-05,
+      "loss": 0.7305,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989836984116968e-05,
+      "loss": 0.7891,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989765801973587e-05,
+      "loss": 0.5742,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989694371545818e-05,
+      "loss": 1.1016,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989622692835437e-05,
+      "loss": 0.9727,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998955076584422e-05,
+      "loss": 1.6562,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989478590573958e-05,
+      "loss": 1.0625,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989406167026442e-05,
+      "loss": 1.3438,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998933349520348e-05,
+      "loss": 1.625,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998926057510687e-05,
+      "loss": 1.3828,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989187406738424e-05,
+      "loss": 1.3438,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989113990099963e-05,
+      "loss": 1.4609,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989040325193317e-05,
+      "loss": 1.2734,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998896641202031e-05,
+      "loss": 1.0,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988892250582777e-05,
+      "loss": 1.0,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998881784088257e-05,
+      "loss": 0.9844,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998874318292153e-05,
+      "loss": 1.4688,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988668276701523e-05,
+      "loss": 1.4219,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988593122224396e-05,
+      "loss": 1.125,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988517719492025e-05,
+      "loss": 0.8281,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988442068506288e-05,
+      "loss": 1.2031,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998836616926906e-05,
+      "loss": 1.7812,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988290021782224e-05,
+      "loss": 1.2031,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998821362604768e-05,
+      "loss": 1.0078,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988136982067324e-05,
+      "loss": 1.2734,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998806008984306e-05,
+      "loss": 0.582,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99879829493768e-05,
+      "loss": 1.6484,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998790556067046e-05,
+      "loss": 1.1797,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987827923725966e-05,
+      "loss": 0.4551,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987750038545242e-05,
+      "loss": 0.6133,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998767190513023e-05,
+      "loss": 0.6641,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987593523482867e-05,
+      "loss": 1.0938,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998751489360511e-05,
+      "loss": 1.5234,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99874360154989e-05,
+      "loss": 0.9648,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987356889166207e-05,
+      "loss": 1.2188,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987277514608993e-05,
+      "loss": 1.3125,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9987197891829236e-05,
+      "loss": 1.0781,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998711802082891e-05,
+      "loss": 0.5586,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998703790161e-05,
+      "loss": 1.5469,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99869575341745e-05,
+      "loss": 1.1797,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986876918524406e-05,
+      "loss": 1.7812,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986796054661722e-05,
+      "loss": 1.4922,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998671494258846e-05,
+      "loss": 1.1562,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998663358230663e-05,
+      "loss": 1.2188,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998655197381826e-05,
+      "loss": 1.1797,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986470117125375e-05,
+      "loss": 1.1562,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998638801223001e-05,
+      "loss": 0.5977,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986305659134208e-05,
+      "loss": 1.5078,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986223057840012e-05,
+      "loss": 0.793,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998614020834948e-05,
+      "loss": 0.9102,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9986057110664662e-05,
+      "loss": 1.0703,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985973764787635e-05,
+      "loss": 1.3906,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985890170720463e-05,
+      "loss": 1.5938,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985806328465225e-05,
+      "loss": 1.6797,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985722238024004e-05,
+      "loss": 1.1641,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985637899398894e-05,
+      "loss": 1.8438,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998555331259199e-05,
+      "loss": 0.793,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985468477605388e-05,
+      "loss": 1.7266,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985383394441204e-05,
+      "loss": 1.1641,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985298063101547e-05,
+      "loss": 1.2656,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985212483588544e-05,
+      "loss": 1.2266,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985126655904317e-05,
+      "loss": 1.25,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9985040580051e-05,
+      "loss": 1.1719,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984954256030735e-05,
+      "loss": 0.8828,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984867683845665e-05,
+      "loss": 1.0547,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998478086349794e-05,
+      "loss": 1.0547,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984693794989725e-05,
+      "loss": 1.4297,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984606478323173e-05,
+      "loss": 0.8633,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984518913500464e-05,
+      "loss": 1.1719,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998443110052377e-05,
+      "loss": 1.0547,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984343039395274e-05,
+      "loss": 0.9805,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984254730117164e-05,
+      "loss": 0.9844,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984166172691632e-05,
+      "loss": 1.1953,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984077367120886e-05,
+      "loss": 1.3047,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998398831340713e-05,
+      "loss": 1.1328,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983899011552578e-05,
+      "loss": 1.4453,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983809461559445e-05,
+      "loss": 0.7969,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983719663429962e-05,
+      "loss": 1.0625,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998362961716636e-05,
+      "loss": 1.6328,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983539322770876e-05,
+      "loss": 1.2266,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998344878024575e-05,
+      "loss": 0.7695,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983357989593242e-05,
+      "loss": 1.3359,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99832669508156e-05,
+      "loss": 0.9258,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998317566391509e-05,
+      "loss": 1.5234,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998308412889398e-05,
+      "loss": 1.3281,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998299234575455e-05,
+      "loss": 1.1172,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982900314499072e-05,
+      "loss": 1.0547,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982808035129837e-05,
+      "loss": 1.9922,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998271550764914e-05,
+      "loss": 1.1562,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998262273205928e-05,
+      "loss": 1.2266,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982529708362568e-05,
+      "loss": 1.5859,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982436436561305e-05,
+      "loss": 0.7734,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982342916657817e-05,
+      "loss": 1.3359,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982249148654425e-05,
+      "loss": 1.4375,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982155132553463e-05,
+      "loss": 0.7656,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982060868357266e-05,
+      "loss": 1.5781,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981966356068175e-05,
+      "loss": 1.2734,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981871595688544e-05,
+      "loss": 1.0,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998177658722072e-05,
+      "loss": 1.2344,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998168133066707e-05,
+      "loss": 0.5547,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981585826029963e-05,
+      "loss": 1.1406,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981490073311766e-05,
+      "loss": 1.25,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981394072514865e-05,
+      "loss": 0.8086,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981297823641644e-05,
+      "loss": 1.5703,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981201326694496e-05,
+      "loss": 1.125,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998110458167582e-05,
+      "loss": 1.375,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981007588588013e-05,
+      "loss": 1.5078,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980910347433495e-05,
+      "loss": 1.4375,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998081285821468e-05,
+      "loss": 1.3359,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980715120933988e-05,
+      "loss": 1.0234,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980617135593854e-05,
+      "loss": 1.2734,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980518902196705e-05,
+      "loss": 1.1406,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998042042074499e-05,
+      "loss": 1.2109,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980321691241155e-05,
+      "loss": 1.2188,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980222713687653e-05,
+      "loss": 1.3281,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980123488086942e-05,
+      "loss": 1.2969,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998002401444149e-05,
+      "loss": 1.5,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979924292753767e-05,
+      "loss": 1.5547,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979824323026257e-05,
+      "loss": 0.5977,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997972410526144e-05,
+      "loss": 1.125,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979623639461808e-05,
+      "loss": 1.0078,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979522925629857e-05,
+      "loss": 0.6055,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979421963768092e-05,
+      "loss": 0.9922,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979320753879023e-05,
+      "loss": 0.7891,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979219295965162e-05,
+      "loss": 1.4844,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979117590029036e-05,
+      "loss": 0.6797,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997901563607317e-05,
+      "loss": 1.2578,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978913434100088e-05,
+      "loss": 0.6992,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978810984112348e-05,
+      "loss": 0.9336,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978708286112484e-05,
+      "loss": 0.8516,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978605340103053e-05,
+      "loss": 1.4609,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978502146086614e-05,
+      "loss": 1.4375,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997839870406573e-05,
+      "loss": 1.5312,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978295014042967e-05,
+      "loss": 0.7305,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978191076020916e-05,
+      "loss": 0.9219,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978086890002145e-05,
+      "loss": 1.2031,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977982455989252e-05,
+      "loss": 1.5859,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977877773984834e-05,
+      "loss": 1.2422,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977772843991486e-05,
+      "loss": 0.7148,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997766766601182e-05,
+      "loss": 2.1406,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977562240048452e-05,
+      "loss": 1.2109,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977456566103997e-05,
+      "loss": 1.0156,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977350644181085e-05,
+      "loss": 1.375,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997724447428235e-05,
+      "loss": 0.9375,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977138056410425e-05,
+      "loss": 1.5234,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997703139056796e-05,
+      "loss": 1.2266,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976924476757604e-05,
+      "loss": 1.1172,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976817314982017e-05,
+      "loss": 0.8359,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997670990524386e-05,
+      "loss": 1.1953,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99766022475458e-05,
+      "loss": 1.4141,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976494341890515e-05,
+      "loss": 0.9062,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976386188280688e-05,
+      "loss": 0.5664,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997627778671901e-05,
+      "loss": 1.0156,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976169137208162e-05,
+      "loss": 0.8789,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9976060239750864e-05,
+      "loss": 1.0625,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975951094349806e-05,
+      "loss": 0.4648,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997584170100771e-05,
+      "loss": 1.5391,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997573205972729e-05,
+      "loss": 1.0,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975622170511272e-05,
+      "loss": 1.5156,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975512033362393e-05,
+      "loss": 1.0703,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975401648283383e-05,
+      "loss": 1.625,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975291015276985e-05,
+      "loss": 1.1875,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975180134345954e-05,
+      "loss": 0.8633,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975069005493042e-05,
+      "loss": 1.2344,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974957628721013e-05,
+      "loss": 1.4375,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974846004032636e-05,
+      "loss": 1.1641,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997473413143068e-05,
+      "loss": 1.625,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974622010917933e-05,
+      "loss": 1.8359,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997450964249718e-05,
+      "loss": 0.6406,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974397026171205e-05,
+      "loss": 1.0234,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974284161942815e-05,
+      "loss": 0.7734,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974171049814817e-05,
+      "loss": 0.7422,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997405768979002e-05,
+      "loss": 1.6875,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973944081871236e-05,
+      "loss": 0.8164,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973830226061293e-05,
+      "loss": 0.9141,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973716122363025e-05,
+      "loss": 0.3027,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973601770779264e-05,
+      "loss": 0.4688,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973487171312846e-05,
+      "loss": 0.6445,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997337232396663e-05,
+      "loss": 0.9219,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973257228743463e-05,
+      "loss": 0.4629,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973141885646207e-05,
+      "loss": 1.8984,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973026294677732e-05,
+      "loss": 1.4766,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997291045584091e-05,
+      "loss": 0.5586,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972794369138615e-05,
+      "loss": 0.8945,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997267803457374e-05,
+      "loss": 1.4922,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972561452149168e-05,
+      "loss": 0.6211,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972444621867804e-05,
+      "loss": 0.9336,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972327543732545e-05,
+      "loss": 1.4219,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972210217746308e-05,
+      "loss": 0.377,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972092643912003e-05,
+      "loss": 1.5078,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971974822232554e-05,
+      "loss": 1.6719,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971856752710893e-05,
+      "loss": 0.8672,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997173843534995e-05,
+      "loss": 1.375,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971619870152663e-05,
+      "loss": 1.0547,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971501057121987e-05,
+      "loss": 1.0547,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997138199626087e-05,
+      "loss": 1.2188,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971262687572272e-05,
+      "loss": 1.4609,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971143131059155e-05,
+      "loss": 1.2891,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971023326724496e-05,
+      "loss": 1.0938,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997090327457127e-05,
+      "loss": 1.9922,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997078297460246e-05,
+      "loss": 0.6602,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997066242682106e-05,
+      "loss": 1.4609,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9970541631230057e-05,
+      "loss": 0.9453,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9970420587832465e-05,
+      "loss": 0.9141,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9970299296631285e-05,
+      "loss": 1.1016,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997017775762953e-05,
+      "loss": 1.7812,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9970055970830225e-05,
+      "loss": 1.7578,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969933936236396e-05,
+      "loss": 0.9922,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969811653851078e-05,
+      "loss": 1.1094,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969689123677306e-05,
+      "loss": 1.1094,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969566345718127e-05,
+      "loss": 1.125,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969443319976592e-05,
+      "loss": 1.3047,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969320046455763e-05,
+      "loss": 1.1406,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969196525158697e-05,
+      "loss": 1.0625,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969072756088468e-05,
+      "loss": 0.8477,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996894873924815e-05,
+      "loss": 1.1562,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968824474640828e-05,
+      "loss": 1.6484,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996869996226959e-05,
+      "loss": 0.8477,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968575202137527e-05,
+      "loss": 1.1641,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968450194247745e-05,
+      "loss": 1.6641,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968324938603347e-05,
+      "loss": 1.2266,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968199435207446e-05,
+      "loss": 1.2031,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968073684063167e-05,
+      "loss": 1.4922,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996794768517363e-05,
+      "loss": 1.1016,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967821438541967e-05,
+      "loss": 1.1016,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967694944171318e-05,
+      "loss": 1.7266,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967568202064823e-05,
+      "loss": 1.0469,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967441212225633e-05,
+      "loss": 0.6328,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996731397465691e-05,
+      "loss": 1.2109,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996718648936181e-05,
+      "loss": 0.793,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99670587563435e-05,
+      "loss": 0.5625,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996693077560516e-05,
+      "loss": 1.4844,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966802547149974e-05,
+      "loss": 1.3516,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996667407098112e-05,
+      "loss": 1.2812,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9966545347101792e-05,
+      "loss": 1.875,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9966416375515194e-05,
+      "loss": 1.0078,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996628715622453e-05,
+      "loss": 1.5859,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996615768923301e-05,
+      "loss": 1.0391,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996602797454386e-05,
+      "loss": 1.4375,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996589801216029e-05,
+      "loss": 0.918,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9965767802085538e-05,
+      "loss": 0.707,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9965637344322842e-05,
+      "loss": 1.6875,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9965506638875438e-05,
+      "loss": 1.6406,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9965375685746583e-05,
+      "loss": 0.5469,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9965244484939522e-05,
+      "loss": 1.5547,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9965113036457523e-05,
+      "loss": 0.6953,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996498134030385e-05,
+      "loss": 1.2578,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9964849396481777e-05,
+      "loss": 1.0078,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9964717204994582e-05,
+      "loss": 0.5195,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9964584765845556e-05,
+      "loss": 0.7148,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9964452079037983e-05,
+      "loss": 1.3203,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9964319144575165e-05,
+      "loss": 0.8672,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9964185962460404e-05,
+      "loss": 0.8633,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996405253269701e-05,
+      "loss": 1.125,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9963918855288306e-05,
+      "loss": 1.5938,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9963784930237603e-05,
+      "loss": 0.6094,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9963650757548236e-05,
+      "loss": 1.5391,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9963516337223542e-05,
+      "loss": 0.8203,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9963381669266858e-05,
+      "loss": 0.8789,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996324675368153e-05,
+      "loss": 1.1719,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996311159047091e-05,
+      "loss": 1.1641,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962976179638367e-05,
+      "loss": 1.7578,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962840521187257e-05,
+      "loss": 1.2266,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962704615120956e-05,
+      "loss": 1.4688,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962568461442838e-05,
+      "loss": 0.9219,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996243206015629e-05,
+      "loss": 1.1719,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962295411264704e-05,
+      "loss": 2.0625,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962158514771472e-05,
+      "loss": 1.5469,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962021370679995e-05,
+      "loss": 1.4766,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9961883978993687e-05,
+      "loss": 1.6562,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996174633971596e-05,
+      "loss": 1.1328,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996160845285024e-05,
+      "loss": 0.7656,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9961470318399946e-05,
+      "loss": 1.7812,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9961331936368512e-05,
+      "loss": 1.8125,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996119330675938e-05,
+      "loss": 0.668,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9961054429575997e-05,
+      "loss": 0.668,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996091530482181e-05,
+      "loss": 1.4922,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996077593250028e-05,
+      "loss": 0.5508,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9960636312614875e-05,
+      "loss": 0.8086,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996049644516906e-05,
+      "loss": 0.707,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.996035633016631e-05,
+      "loss": 0.7188,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9960215967610107e-05,
+      "loss": 1.2422,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9960075357503944e-05,
+      "loss": 0.8672,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9959934499851315e-05,
+      "loss": 1.3672,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995979339465572e-05,
+      "loss": 1.0859,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995965204192066e-05,
+      "loss": 0.4473,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995951044164966e-05,
+      "loss": 0.8867,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995936859384623e-05,
+      "loss": 1.1641,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9959226498513897e-05,
+      "loss": 1.3672,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9959084155656195e-05,
+      "loss": 2.0312,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995894156527666e-05,
+      "loss": 1.5625,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9958798727378838e-05,
+      "loss": 0.6484,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995865564196628e-05,
+      "loss": 0.9414,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9958512309042537e-05,
+      "loss": 1.2422,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9958368728611176e-05,
+      "loss": 1.4141,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9958224900675764e-05,
+      "loss": 1.1562,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9958080825239874e-05,
+      "loss": 1.4219,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9957936502307088e-05,
+      "loss": 0.7773,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9957791931880996e-05,
+      "loss": 1.3516,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9957647113965185e-05,
+      "loss": 0.7734,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995750204856326e-05,
+      "loss": 0.668,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9957356735678824e-05,
+      "loss": 0.9102,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995721117531549e-05,
+      "loss": 0.6172,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9957065367476875e-05,
+      "loss": 1.4297,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99569193121666e-05,
+      "loss": 1.1016,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99567730093883e-05,
+      "loss": 0.9727,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9956626459145606e-05,
+      "loss": 1.5,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9956479661442164e-05,
+      "loss": 0.707,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9956332616281624e-05,
+      "loss": 1.5156,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9956185323667636e-05,
+      "loss": 1.5781,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9956037783603862e-05,
+      "loss": 1.1484,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9955889996093974e-05,
+      "loss": 1.375,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9955741961141636e-05,
+      "loss": 0.9688,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9955593678750533e-05,
+      "loss": 1.1641,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9955445148924352e-05,
+      "loss": 0.9414,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9955296371666782e-05,
+      "loss": 1.125,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995514734698152e-05,
+      "loss": 1.0625,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995499807487227e-05,
+      "loss": 1.125,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9954848555342744e-05,
+      "loss": 1.6328,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995469878839666e-05,
+      "loss": 1.0703,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9954548774037733e-05,
+      "loss": 0.4492,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9954398512269695e-05,
+      "loss": 1.3828,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9954248003096282e-05,
+      "loss": 0.582,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9954097246521232e-05,
+      "loss": 0.8711,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99539462425483e-05,
+      "loss": 0.8867,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9953794991181227e-05,
+      "loss": 1.5625,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9953643492423782e-05,
+      "loss": 1.0547,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9953491746279723e-05,
+      "loss": 0.7578,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9953339752752824e-05,
+      "loss": 0.375,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9953187511846867e-05,
+      "loss": 1.375,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995303502356563e-05,
+      "loss": 1.4219,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995288228791291e-05,
+      "loss": 0.4043,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9952729304892493e-05,
+      "loss": 1.6484,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9952576074508186e-05,
+      "loss": 1.3984,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99524225967638e-05,
+      "loss": 1.5078,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995226887166315e-05,
+      "loss": 1.1016,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9952114899210048e-05,
+      "loss": 1.125,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995196067940833e-05,
+      "loss": 1.8281,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9951806212261826e-05,
+      "loss": 1.0,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9951651497774375e-05,
+      "loss": 1.2109,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995149653594982e-05,
+      "loss": 0.8984,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9951341326792018e-05,
+      "loss": 1.1641,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995118587030482e-05,
+      "loss": 0.7305,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9951030166492093e-05,
+      "loss": 0.4258,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9950874215357708e-05,
+      "loss": 1.3516,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.995071801690554e-05,
+      "loss": 1.0312,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9950561571139468e-05,
+      "loss": 1.2422,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9950404878063385e-05,
+      "loss": 1.1641,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9950247937681183e-05,
+      "loss": 1.3047,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9950090749996765e-05,
+      "loss": 0.8594,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9949933315014034e-05,
+      "loss": 0.9805,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9949775632736906e-05,
+      "loss": 0.8555,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9949617703169296e-05,
+      "loss": 0.418,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9949459526315134e-05,
+      "loss": 1.3125,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9949301102178345e-05,
+      "loss": 1.3438,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9949142430762878e-05,
+      "loss": 1.4453,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9948983512072662e-05,
+      "loss": 0.9336,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994882434611166e-05,
+      "loss": 1.1719,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9948664932883818e-05,
+      "loss": 1.0391,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99485052723931e-05,
+      "loss": 2.2656,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9948345364643482e-05,
+      "loss": 1.2578,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994818520963893e-05,
+      "loss": 1.0938,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9948024807383425e-05,
+      "loss": 1.2734,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9947864157880955e-05,
+      "loss": 1.0391,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9947703261135515e-05,
+      "loss": 0.7422,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99475421171511e-05,
+      "loss": 0.7891,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994738072593172e-05,
+      "loss": 0.7148,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9947219087481383e-05,
+      "loss": 1.6406,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9947057201804105e-05,
+      "loss": 1.1875,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9946895068903914e-05,
+      "loss": 1.7812,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9946732688784832e-05,
+      "loss": 0.6055,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9946570061450906e-05,
+      "loss": 1.1484,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9946407186906163e-05,
+      "loss": 1.6641,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9946244065154668e-05,
+      "loss": 0.8125,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9946080696200463e-05,
+      "loss": 0.8438,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9945917080047613e-05,
+      "loss": 1.1562,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9945753216700182e-05,
+      "loss": 0.6523,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9945589106162245e-05,
+      "loss": 0.6914,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994542474843788e-05,
+      "loss": 0.7461,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9945260143531174e-05,
+      "loss": 0.8516,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9945095291446215e-05,
+      "loss": 0.4395,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9944930192187098e-05,
+      "loss": 1.9844,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9944764845757936e-05,
+      "loss": 1.5391,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994459925216283e-05,
+      "loss": 0.5742,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9944433411405893e-05,
+      "loss": 0.9922,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9944267323491255e-05,
+      "loss": 1.3906,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9944100988423042e-05,
+      "loss": 0.7969,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9943934406205384e-05,
+      "loss": 1.1094,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9943767576842424e-05,
+      "loss": 1.4219,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994360050033831e-05,
+      "loss": 0.8164,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9943433176697188e-05,
+      "loss": 1.2109,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9943265605923227e-05,
+      "loss": 0.8359,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9943097788020586e-05,
+      "loss": 1.0781,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.994292972299343e-05,
+      "loss": 1.3047,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9942761410845946e-05,
+      "loss": 0.8203,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9942592851582313e-05,
+      "loss": 0.8828,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.994242404520672e-05,
+      "loss": 1.1797,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9942254991723364e-05,
+      "loss": 1.9922,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9942085691136444e-05,
+      "loss": 1.1641,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.994191614345017e-05,
+      "loss": 1.3047,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9941746348668758e-05,
+      "loss": 1.6406,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9941576306796424e-05,
+      "loss": 1.3203,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9941406017837394e-05,
+      "loss": 0.6055,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9941235481795905e-05,
+      "loss": 0.3652,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.994106469867619e-05,
+      "loss": 0.7773,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9940893668482496e-05,
+      "loss": 1.1406,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9940722391219077e-05,
+      "loss": 0.7031,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.994055086689018e-05,
+      "loss": 1.4297,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9940379095500084e-05,
+      "loss": 0.7109,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.994020707705304e-05,
+      "loss": 1.4375,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.994003481155334e-05,
+      "loss": 1.4531,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9939862299005254e-05,
+      "loss": 1.1094,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9939689539413076e-05,
+      "loss": 1.0859,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9939516532781093e-05,
+      "loss": 0.9453,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993934327911361e-05,
+      "loss": 1.3594,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9939169778414936e-05,
+      "loss": 0.9961,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9938996030689375e-05,
+      "loss": 1.8516,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993882203594125e-05,
+      "loss": 2.1406,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9938647794174887e-05,
+      "loss": 1.1641,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9938473305394614e-05,
+      "loss": 0.6328,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9938298569604768e-05,
+      "loss": 1.4453,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9938123586809692e-05,
+      "loss": 1.6797,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9937948357013734e-05,
+      "loss": 0.8633,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993777288022125e-05,
+      "loss": 0.7578,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9937597156436606e-05,
+      "loss": 1.3984,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993742118566416e-05,
+      "loss": 1.2578,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9937244967908293e-05,
+      "loss": 0.8203,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993706850317338e-05,
+      "loss": 1.75,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9936891791463812e-05,
+      "loss": 0.7031,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993671483278398e-05,
+      "loss": 0.6562,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9936537627138274e-05,
+      "loss": 0.9336,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9936360174531107e-05,
+      "loss": 0.6992,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9936182474966886e-05,
+      "loss": 1.0859,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993600452845003e-05,
+      "loss": 0.5781,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993582633498496e-05,
+      "loss": 0.75,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99356478945761e-05,
+      "loss": 1.5703,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99354692072279e-05,
+      "loss": 0.8203,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993529027294478e-05,
+      "loss": 0.9648,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9935111091731203e-05,
+      "loss": 0.7734,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934931663591617e-05,
+      "loss": 1.7734,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934751988530482e-05,
+      "loss": 1.8672,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934572066552264e-05,
+      "loss": 1.1484,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934391897661433e-05,
+      "loss": 1.2812,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934211481862468e-05,
+      "loss": 1.6172,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9934030819159852e-05,
+      "loss": 1.1328,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993384990955808e-05,
+      "loss": 1.0078,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9933668753061643e-05,
+      "loss": 1.9609,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993348734967504e-05,
+      "loss": 1.1797,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993330569940279e-05,
+      "loss": 0.9062,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99331238022494e-05,
+      "loss": 1.3281,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9932941658219395e-05,
+      "loss": 1.3438,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99327592673173e-05,
+      "loss": 0.2754,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993257662954765e-05,
+      "loss": 1.3672,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993239374491498e-05,
+      "loss": 1.8906,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9932210613423838e-05,
+      "loss": 0.9102,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9932027235078774e-05,
+      "loss": 0.9922,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9931843609884348e-05,
+      "loss": 1.0547,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9931659737845124e-05,
+      "loss": 1.1797,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993147561896567e-05,
+      "loss": 1.1641,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9931291253250565e-05,
+      "loss": 0.9688,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9931106640704384e-05,
+      "loss": 1.5156,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9930921781331725e-05,
+      "loss": 1.5,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9930736675137176e-05,
+      "loss": 1.1172,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9930551322125344e-05,
+      "loss": 1.1172,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9930365722300826e-05,
+      "loss": 1.2656,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993017987566824e-05,
+      "loss": 0.9023,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9929993782232208e-05,
+      "loss": 1.3125,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992980744199735e-05,
+      "loss": 0.6992,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9929620854968302e-05,
+      "loss": 0.9062,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9929434021149697e-05,
+      "loss": 0.5,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9929246940546182e-05,
+      "loss": 0.7734,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9929059613162404e-05,
+      "loss": 1.0391,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992887203900302e-05,
+      "loss": 1.3672,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9928684218072692e-05,
+      "loss": 1.6094,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992849615037609e-05,
+      "loss": 0.7031,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9928307835917882e-05,
+      "loss": 1.2969,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9928119274702755e-05,
+      "loss": 0.75,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9927930466735396e-05,
+      "loss": 0.7383,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9927741412020493e-05,
+      "loss": 1.4688,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9927552110562744e-05,
+      "loss": 1.1406,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9927362562366855e-05,
+      "loss": 1.1562,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992717276743754e-05,
+      "loss": 1.2734,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992698272577952e-05,
+      "loss": 0.8555,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9926792437397506e-05,
+      "loss": 0.625,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9926601902296236e-05,
+      "loss": 0.7617,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992641112048044e-05,
+      "loss": 1.6953,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992622009195487e-05,
+      "loss": 1.1562,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9926028816724264e-05,
+      "loss": 2.375,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992583729479338e-05,
+      "loss": 1.2422,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9925645526166978e-05,
+      "loss": 1.0547,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9925453510849818e-05,
+      "loss": 0.6953,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9925261248846683e-05,
+      "loss": 1.3672,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9925068740162342e-05,
+      "loss": 1.3281,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9924875984801587e-05,
+      "loss": 1.0469,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9924682982769204e-05,
+      "loss": 0.7305,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9924489734069993e-05,
+      "loss": 0.7266,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9924296238708756e-05,
+      "loss": 0.8594,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9924102496690295e-05,
+      "loss": 2.25,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9923908508019438e-05,
+      "loss": 1.1797,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9923714272701e-05,
+      "loss": 0.5312,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9923519790739804e-05,
+      "loss": 0.7227,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992332506214069e-05,
+      "loss": 1.0938,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99231300869085e-05,
+      "loss": 1.4062,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992293486504807e-05,
+      "loss": 1.6641,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9922739396564256e-05,
+      "loss": 0.6641,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9922543681461925e-05,
+      "loss": 1.6094,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9922347719745928e-05,
+      "loss": 1.3594,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9922151511421143e-05,
+      "loss": 1.3906,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9921955056492446e-05,
+      "loss": 0.8867,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992175835496472e-05,
+      "loss": 1.1172,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9921561406842853e-05,
+      "loss": 1.2422,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9921364212131738e-05,
+      "loss": 1.3594,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9921166770836278e-05,
+      "loss": 1.2734,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9920969082961378e-05,
+      "loss": 1.1328,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9920771148511955e-05,
+      "loss": 1.0391,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9920572967492925e-05,
+      "loss": 0.8203,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9920374539909214e-05,
+      "loss": 0.7305,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9920175865765756e-05,
+      "loss": 1.0547,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991997694506749e-05,
+      "loss": 0.5938,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9919777777819357e-05,
+      "loss": 0.5586,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9919578364026305e-05,
+      "loss": 1.1406,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9919378703693293e-05,
+      "loss": 0.6328,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9919178796825283e-05,
+      "loss": 0.6641,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9918978643427243e-05,
+      "loss": 1.9531,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9918778243504154e-05,
+      "loss": 0.7305,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9918577597060985e-05,
+      "loss": 0.7852,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991837670410273e-05,
+      "loss": 1.1172,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9918175564634382e-05,
+      "loss": 1.1953,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991797417866094e-05,
+      "loss": 1.7109,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9917772546187405e-05,
+      "loss": 1.1797,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991757066721879e-05,
+      "loss": 1.3047,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9917368541760117e-05,
+      "loss": 1.1172,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9917166169816404e-05,
+      "loss": 0.9805,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9916963551392687e-05,
+      "loss": 1.9844,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9916760686493996e-05,
+      "loss": 1.8438,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9916557575125374e-05,
+      "loss": 1.1172,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991635421729187e-05,
+      "loss": 1.3359,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991615061299854e-05,
+      "loss": 0.8945,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991594676225044e-05,
+      "loss": 1.0078,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9915742665052643e-05,
+      "loss": 1.7422,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9915538321410216e-05,
+      "loss": 0.7031,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991533373132824e-05,
+      "loss": 0.6797,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9915128894811797e-05,
+      "loss": 1.2266,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991492381186598e-05,
+      "loss": 0.6523,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9914718482495893e-05,
+      "loss": 1.8281,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9914512906706625e-05,
+      "loss": 1.8516,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9914307084503296e-05,
+      "loss": 1.0156,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9914101015891018e-05,
+      "loss": 0.8828,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9913894700874915e-05,
+      "loss": 0.5,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991368813946011e-05,
+      "loss": 0.9062,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9913481331651742e-05,
+      "loss": 1.0469,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991327427745495e-05,
+      "loss": 1.0156,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9913066976874876e-05,
+      "loss": 0.9258,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9912859429916676e-05,
+      "loss": 0.8242,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9912651636585507e-05,
+      "loss": 1.5234,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9912443596886534e-05,
+      "loss": 1.6172,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9912235310824925e-05,
+      "loss": 1.2031,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9912026778405863e-05,
+      "loss": 1.3047,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9911817999634526e-05,
+      "loss": 1.1172,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9911608974516105e-05,
+      "loss": 1.4141,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.991139970305579e-05,
+      "loss": 1.3438,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9911190185258787e-05,
+      "loss": 1.0859,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9910980421130305e-05,
+      "loss": 0.8594,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9910770410675556e-05,
+      "loss": 1.0781,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9910560153899753e-05,
+      "loss": 1.125,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.991034965080813e-05,
+      "loss": 0.8164,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9910138901405918e-05,
+      "loss": 1.3594,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9909927905698352e-05,
+      "loss": 1.2656,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9909716663690678e-05,
+      "loss": 1.3438,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9909505175388143e-05,
+      "loss": 1.6406,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9909293440796007e-05,
+      "loss": 1.0781,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990908145991953e-05,
+      "loss": 1.8438,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9908869232763983e-05,
+      "loss": 1.4062,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990865675933464e-05,
+      "loss": 1.1953,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990844403963678e-05,
+      "loss": 1.8203,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9908231073675686e-05,
+      "loss": 0.7305,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9908017861456664e-05,
+      "loss": 0.7031,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9907804402985002e-05,
+      "loss": 1.8359,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9907590698266004e-05,
+      "loss": 1.5312,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990737674730499e-05,
+      "loss": 0.7266,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9907162550107273e-05,
+      "loss": 1.5391,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9906948106678178e-05,
+      "loss": 1.4375,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990673341702303e-05,
+      "loss": 0.8203,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990651848114717e-05,
+      "loss": 0.4922,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990630329905594e-05,
+      "loss": 2.2344,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9906087870754688e-05,
+      "loss": 1.0078,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9905872196248764e-05,
+      "loss": 3.0625,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9905656275543532e-05,
+      "loss": 1.7812,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9905440108644356e-05,
+      "loss": 0.9297,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9905223695556612e-05,
+      "loss": 0.9844,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9905007036285674e-05,
+      "loss": 0.9297,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9904790130836936e-05,
+      "loss": 1.3281,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990457297921578e-05,
+      "loss": 1.4062,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9904355581427602e-05,
+      "loss": 1.4375,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9904137937477813e-05,
+      "loss": 0.9727,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9903920047371814e-05,
+      "loss": 0.8125,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9903701911115028e-05,
+      "loss": 1.5156,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9903483528712874e-05,
+      "loss": 0.7539,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9903264900170778e-05,
+      "loss": 0.8164,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9903046025494172e-05,
+      "loss": 0.7422,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99028269046885e-05,
+      "loss": 0.4648,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9902607537759206e-05,
+      "loss": 0.4785,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9902387924711746e-05,
+      "loss": 0.5156,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990216806555157e-05,
+      "loss": 1.1953,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990194796028415e-05,
+      "loss": 0.4004,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9901727608914954e-05,
+      "loss": 0.6602,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9901507011449458e-05,
+      "loss": 0.832,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9901286167893142e-05,
+      "loss": 0.9922,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99010650782515e-05,
+      "loss": 0.8867,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9900843742530026e-05,
+      "loss": 1.4219,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.990062216073422e-05,
+      "loss": 1.3906,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9900400332869586e-05,
+      "loss": 1.0234,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9900178258941644e-05,
+      "loss": 1.2656,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9899955938955906e-05,
+      "loss": 0.9727,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9899733372917902e-05,
+      "loss": 0.7656,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989951056083316e-05,
+      "loss": 0.8359,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9899287502707226e-05,
+      "loss": 1.1172,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9899064198545633e-05,
+      "loss": 1.7734,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9898840648353942e-05,
+      "loss": 2.125,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9898616852137696e-05,
+      "loss": 1.6641,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9898392809902473e-05,
+      "loss": 0.7188,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9898168521653827e-05,
+      "loss": 1.2422,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989794398739734e-05,
+      "loss": 1.2031,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9897719207138593e-05,
+      "loss": 1.2031,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989749418088317e-05,
+      "loss": 0.4902,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989726890863666e-05,
+      "loss": 1.0781,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989704339040467e-05,
+      "loss": 0.832,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9896817626192798e-05,
+      "loss": 1.2031,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9896591616006663e-05,
+      "loss": 1.0625,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9896365359851874e-05,
+      "loss": 1.4922,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9896138857734058e-05,
+      "loss": 0.7227,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9895912109658847e-05,
+      "loss": 0.7969,
+      "step": 916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989568511563187e-05,
+      "loss": 0.5234,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9895457875658777e-05,
+      "loss": 0.8906,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9895230389745207e-05,
+      "loss": 2.4062,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9895002657896823e-05,
+      "loss": 1.5156,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9894774680119278e-05,
+      "loss": 0.9961,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989454645641824e-05,
+      "loss": 1.125,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989431798679938e-05,
+      "loss": 1.5938,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9894089271268383e-05,
+      "loss": 1.3125,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9893860309830923e-05,
+      "loss": 0.8516,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9893631102492698e-05,
+      "loss": 1.1875,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9893401649259405e-05,
+      "loss": 1.9453,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9893171950136744e-05,
+      "loss": 1.5469,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9892942005130423e-05,
+      "loss": 1.0625,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989271181424616e-05,
+      "loss": 1.5703,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9892481377489673e-05,
+      "loss": 0.6875,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9892250694866693e-05,
+      "loss": 0.5938,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989201976638295e-05,
+      "loss": 1.0469,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9891788592044184e-05,
+      "loss": 0.75,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989155717185614e-05,
+      "loss": 0.8555,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989132550582457e-05,
+      "loss": 0.6484,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9891093593955238e-05,
+      "loss": 1.5469,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98908614362539e-05,
+      "loss": 1.0859,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.989062903272633e-05,
+      "loss": 1.75,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9890396383378304e-05,
+      "loss": 1.5469,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9890163488215598e-05,
+      "loss": 1.4375,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9889930347244012e-05,
+      "loss": 1.0391,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9889696960469327e-05,
+      "loss": 1.3828,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9889463327897352e-05,
+      "loss": 1.0703,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9889229449533896e-05,
+      "loss": 1.5234,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9888995325384764e-05,
+      "loss": 1.3359,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9888760955455782e-05,
+      "loss": 0.9336,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988852633975277e-05,
+      "loss": 1.3359,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9888291478281564e-05,
+      "loss": 0.8633,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9888056371047993e-05,
+      "loss": 1.2344,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9887821018057913e-05,
+      "loss": 0.7109,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988758541931716e-05,
+      "loss": 0.7031,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9887349574831597e-05,
+      "loss": 1.4141,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988711348460708e-05,
+      "loss": 0.6445,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9886877148649483e-05,
+      "loss": 1.7422,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988664056696468e-05,
+      "loss": 0.6758,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988640373955855e-05,
+      "loss": 0.5156,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988616666643697e-05,
+      "loss": 0.7188,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9885929347605847e-05,
+      "loss": 0.7461,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9885691783071074e-05,
+      "loss": 0.6562,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988545397283855e-05,
+      "loss": 1.0078,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9885215916914187e-05,
+      "loss": 1.2891,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9884977615303907e-05,
+      "loss": 1.0781,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988473906801363e-05,
+      "loss": 0.4355,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988450027504928e-05,
+      "loss": 1.0234,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98842612364168e-05,
+      "loss": 1.0547,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9884021952122125e-05,
+      "loss": 1.5938,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9883782422171207e-05,
+      "loss": 1.0469,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988354264657e-05,
+      "loss": 0.7305,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988330262532445e-05,
+      "loss": 1.1953,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988306235844054e-05,
+      "loss": 1.6719,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988282184592423e-05,
+      "loss": 1.2344,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9882581087781504e-05,
+      "loss": 1.8984,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9882340084018344e-05,
+      "loss": 0.7695,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988209883464074e-05,
+      "loss": 1.8672,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9881857339654685e-05,
+      "loss": 0.9492,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9881615599066184e-05,
+      "loss": 0.7031,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9881373612881245e-05,
+      "loss": 1.1172,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988113138110588e-05,
+      "loss": 0.6562,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9880888903746115e-05,
+      "loss": 1.7656,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988064618080797e-05,
+      "loss": 1.6484,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.988040321229748e-05,
+      "loss": 1.3125,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9880159998220686e-05,
+      "loss": 1.2188,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.987991653858363e-05,
+      "loss": 1.2578,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9879672833392365e-05,
+      "loss": 0.7031,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9879428882652944e-05,
+      "loss": 1.4141,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9879184686371435e-05,
+      "loss": 0.5625,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9878940244553906e-05,
+      "loss": 0.5781,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9878695557206428e-05,
+      "loss": 1.125,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.987845062433509e-05,
+      "loss": 0.8984,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9878205445945973e-05,
+      "loss": 1.0859,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9877960022045177e-05,
+      "loss": 1.5234,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9877714352638792e-05,
+      "loss": 1.4375,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9877468437732936e-05,
+      "loss": 0.8242,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.987722227733371e-05,
+      "loss": 0.9375,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9876975871447237e-05,
+      "loss": 0.8828,
+      "step": 996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9876729220079643e-05,
+      "loss": 0.8633,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9876482323237055e-05,
+      "loss": 1.2109,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.987623518092561e-05,
+      "loss": 1.1562,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.987598779315145e-05,
+      "loss": 1.2344,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9875740159920728e-05,
+      "loss": 1.375,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9875492281239592e-05,
+      "loss": 0.6016,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9875244157114203e-05,
+      "loss": 0.9453,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9874995787550734e-05,
+      "loss": 0.7852,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9874747172555356e-05,
+      "loss": 1.0859,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9874498312134245e-05,
+      "loss": 1.3672,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9874249206293588e-05,
+      "loss": 1.375,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9873999855039577e-05,
+      "loss": 0.7539,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9873750258378404e-05,
+      "loss": 0.5664,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9873500416316282e-05,
+      "loss": 0.75,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.987325032885941e-05,
+      "loss": 0.7969,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9872999996014016e-05,
+      "loss": 0.5703,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.987274941778631e-05,
+      "loss": 1.3359,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9872498594182524e-05,
+      "loss": 1.25,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9872247525208896e-05,
+      "loss": 0.8164,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.987199621087166e-05,
+      "loss": 1.1562,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9871744651177065e-05,
+      "loss": 1.5234,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9871492846131365e-05,
+      "loss": 1.0234,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9871240795740813e-05,
+      "loss": 1.3906,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9870988500011677e-05,
+      "loss": 1.3125,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.987073595895023e-05,
+      "loss": 1.0391,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9870483172562745e-05,
+      "loss": 1.0547,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9870230140855505e-05,
+      "loss": 1.3281,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9869976863834797e-05,
+      "loss": 1.0234,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986972334150692e-05,
+      "loss": 1.5391,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9869469573878173e-05,
+      "loss": 1.0703,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9869215560954864e-05,
+      "loss": 0.957,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9868961302743306e-05,
+      "loss": 1.4297,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9868706799249818e-05,
+      "loss": 0.8125,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9868452050480724e-05,
+      "loss": 0.7969,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9868197056442358e-05,
+      "loss": 1.2422,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9867941817141055e-05,
+      "loss": 1.0469,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986768633258316e-05,
+      "loss": 0.7695,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9867430602775023e-05,
+      "loss": 1.5547,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9867174627723002e-05,
+      "loss": 0.3926,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9866918407433456e-05,
+      "loss": 0.4121,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9866661941912754e-05,
+      "loss": 1.0781,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986640523116727e-05,
+      "loss": 1.5703,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986614827520338e-05,
+      "loss": 1.4609,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986589107402748e-05,
+      "loss": 0.9766,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9865633627645954e-05,
+      "loss": 0.9688,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9865375936065203e-05,
+      "loss": 1.4531,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9865117999291633e-05,
+      "loss": 0.9648,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9864859817331657e-05,
+      "loss": 1.7812,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9864601390191686e-05,
+      "loss": 1.0547,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9864342717878143e-05,
+      "loss": 0.7305,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9864083800397463e-05,
+      "loss": 1.3359,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9863824637756077e-05,
+      "loss": 0.9531,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9863565229960428e-05,
+      "loss": 1.0234,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9863305577016957e-05,
+      "loss": 0.9219,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986304567893213e-05,
+      "loss": 2.0156,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9862785535712396e-05,
+      "loss": 0.7109,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986252514736422e-05,
+      "loss": 1.0547,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9862264513894082e-05,
+      "loss": 1.7891,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9862003635308455e-05,
+      "loss": 0.3906,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986174251161382e-05,
+      "loss": 1.1484,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9861481142816674e-05,
+      "loss": 1.2578,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9861219528923504e-05,
+      "loss": 1.8672,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986095766994082e-05,
+      "loss": 0.5352,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9860695565875127e-05,
+      "loss": 0.793,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.986043321673294e-05,
+      "loss": 0.5234,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9860170622520777e-05,
+      "loss": 1.0312,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9859907783245168e-05,
+      "loss": 0.8984,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.985964469891264e-05,
+      "loss": 1.1016,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.985938136952974e-05,
+      "loss": 0.9805,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9859117795103008e-05,
+      "loss": 1.4766,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9858853975638995e-05,
+      "loss": 1.4062,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9858589911144258e-05,
+      "loss": 0.9922,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.985832560162536e-05,
+      "loss": 1.9297,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9858061047088868e-05,
+      "loss": 1.5234,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.985779624754136e-05,
+      "loss": 0.7773,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.985753120298942e-05,
+      "loss": 1.4844,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.985726591343963e-05,
+      "loss": 1.1797,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9857000378898586e-05,
+      "loss": 1.625,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9856734599372886e-05,
+      "loss": 0.8672,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.985646857486914e-05,
+      "loss": 0.8438,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985620230539395e-05,
+      "loss": 0.9141,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985593579095395e-05,
+      "loss": 0.8633,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9855669031555747e-05,
+      "loss": 1.0859,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9855402027205982e-05,
+      "loss": 1.6094,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9855134777911286e-05,
+      "loss": 1.5938,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9854867283678302e-05,
+      "loss": 0.5977,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985459954451368e-05,
+      "loss": 0.8242,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9854331560424075e-05,
+      "loss": 0.6953,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985406333141614e-05,
+      "loss": 1.2188,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9853794857496552e-05,
+      "loss": 0.9102,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985352613867198e-05,
+      "loss": 0.5,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.98532571749491e-05,
+      "loss": 0.8555,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9852987966334596e-05,
+      "loss": 1.3125,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9852718512835163e-05,
+      "loss": 0.9727,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9852448814457498e-05,
+      "loss": 0.9492,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.98521788712083e-05,
+      "loss": 1.1406,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9851908683094278e-05,
+      "loss": 1.2109,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985163825012215e-05,
+      "loss": 1.2812,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985136757229864e-05,
+      "loss": 1.1016,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985109664963047e-05,
+      "loss": 1.2656,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985082548212438e-05,
+      "loss": 0.8164,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.98505540697871e-05,
+      "loss": 0.7148,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9850282412625384e-05,
+      "loss": 0.7656,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.985001051064598e-05,
+      "loss": 0.7422,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9849738363855645e-05,
+      "loss": 0.8984,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984946597226115e-05,
+      "loss": 1.2812,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9849193335869255e-05,
+      "loss": 1.3438,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9848920454686738e-05,
+      "loss": 1.2109,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984864732872039e-05,
+      "loss": 1.0391,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9848373957976985e-05,
+      "loss": 0.793,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984810034246333e-05,
+      "loss": 1.125,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9847826482186222e-05,
+      "loss": 0.7891,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9847552377152463e-05,
+      "loss": 0.5547,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984727802736887e-05,
+      "loss": 1.4766,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984700343284226e-05,
+      "loss": 0.7852,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9846728593579457e-05,
+      "loss": 0.9805,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9846453509587295e-05,
+      "loss": 1.6016,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984617818087261e-05,
+      "loss": 0.75,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984590260744224e-05,
+      "loss": 1.3672,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984562678930304e-05,
+      "loss": 0.6758,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9845350726461864e-05,
+      "loss": 2.4531,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9845074418925568e-05,
+      "loss": 0.6172,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9844797866701027e-05,
+      "loss": 1.1328,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984452106979511e-05,
+      "loss": 1.2812,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9844244028214697e-05,
+      "loss": 0.7266,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9843966741966674e-05,
+      "loss": 1.1875,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9843689211057935e-05,
+      "loss": 1.5078,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984341143549537e-05,
+      "loss": 0.6484,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9843133415285892e-05,
+      "loss": 0.9375,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9842855150436405e-05,
+      "loss": 0.9844,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9842576640953832e-05,
+      "loss": 1.1484,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9842297886845085e-05,
+      "loss": 2.1094,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9842018888117098e-05,
+      "loss": 1.3281,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9841739644776804e-05,
+      "loss": 0.7344,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9841460156831146e-05,
+      "loss": 1.3828,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9841180424287065e-05,
+      "loss": 1.0938,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984090044715152e-05,
+      "loss": 1.5625,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9840620225431462e-05,
+      "loss": 1.6641,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.984033975913386e-05,
+      "loss": 0.6016,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9840059048265687e-05,
+      "loss": 1.2969,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9839778092833916e-05,
+      "loss": 1.2891,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.983949689284553e-05,
+      "loss": 0.4883,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9839215448307518e-05,
+      "loss": 1.3594,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9838933759226874e-05,
+      "loss": 1.5625,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9838651825610603e-05,
+      "loss": 1.8984,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9838369647465708e-05,
+      "loss": 0.9844,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9838087224799207e-05,
+      "loss": 1.2891,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9837804557618116e-05,
+      "loss": 1.2422,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9837521645929462e-05,
+      "loss": 1.2344,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9837238489740272e-05,
+      "loss": 1.6562,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9836955089057586e-05,
+      "loss": 1.1094,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.983667144388845e-05,
+      "loss": 0.6406,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9836387554239912e-05,
+      "loss": 0.8906,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.983610342011903e-05,
+      "loss": 1.1172,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.983581904153286e-05,
+      "loss": 2.0625,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9835534418488476e-05,
+      "loss": 1.5938,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.983524955099295e-05,
+      "loss": 1.2344,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.983496443905336e-05,
+      "loss": 0.7695,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.983467908267679e-05,
+      "loss": 0.6445,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.983439348187034e-05,
+      "loss": 0.4375,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9834107636641103e-05,
+      "loss": 0.8906,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.983382154699619e-05,
+      "loss": 1.3359,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98335352129427e-05,
+      "loss": 0.6953,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.983324863448776e-05,
+      "loss": 0.5898,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.983296181163848e-05,
+      "loss": 0.9961,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9832674744402003e-05,
+      "loss": 1.2188,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9832387432785455e-05,
+      "loss": 1.4531,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9832099876795984e-05,
+      "loss": 1.3516,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9831812076440727e-05,
+      "loss": 0.5234,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9831524031726843e-05,
+      "loss": 1.9375,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9831235742661493e-05,
+      "loss": 0.6602,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9830947209251838e-05,
+      "loss": 0.5117,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.983065843150505e-05,
+      "loss": 1.2422,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9830369409428304e-05,
+      "loss": 1.2734,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9830080143028788e-05,
+      "loss": 1.5547,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982979063231369e-05,
+      "loss": 0.9648,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9829500877290205e-05,
+      "loss": 0.9805,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9829210877965534e-05,
+      "loss": 1.0078,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9828920634346885e-05,
+      "loss": 1.2812,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9828630146441474e-05,
+      "loss": 0.5547,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9828339414256514e-05,
+      "loss": 0.875,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9828048437799237e-05,
+      "loss": 1.2891,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9827757217076874e-05,
+      "loss": 1.5,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982746575209666e-05,
+      "loss": 0.625,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9827174042865845e-05,
+      "loss": 1.3047,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9826882089391672e-05,
+      "loss": 1.2422,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9826589891681403e-05,
+      "loss": 1.0859,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9826297449742293e-05,
+      "loss": 0.7969,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982600476358162e-05,
+      "loss": 0.8594,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982571183320665e-05,
+      "loss": 1.4453,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982541865862467e-05,
+      "loss": 0.6523,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982512523984296e-05,
+      "loss": 1.5469,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9824831576868816e-05,
+      "loss": 1.2812,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982453766970954e-05,
+      "loss": 1.1406,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982424351837243e-05,
+      "loss": 0.7148,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98239491228648e-05,
+      "loss": 1.1562,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9823654483193972e-05,
+      "loss": 1.5,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9823359599367258e-05,
+      "loss": 1.2969,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9823064471391994e-05,
+      "loss": 1.2266,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9822769099275515e-05,
+      "loss": 0.9961,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9822473483025165e-05,
+      "loss": 0.8633,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9822177622648284e-05,
+      "loss": 0.9688,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9821881518152225e-05,
+      "loss": 0.7773,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9821585169544357e-05,
+      "loss": 0.8867,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9821288576832036e-05,
+      "loss": 1.4922,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9820991740022638e-05,
+      "loss": 0.9922,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9820694659123538e-05,
+      "loss": 1.1719,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982039733414212e-05,
+      "loss": 0.9766,
+      "step": 1204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.982009976508578e-05,
+      "loss": 1.2188,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9819801951961904e-05,
+      "loss": 1.3125,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98195038947779e-05,
+      "loss": 0.9531,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9819205593541172e-05,
+      "loss": 0.4629,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9818907048259136e-05,
+      "loss": 1.7734,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9818608258939214e-05,
+      "loss": 0.707,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9818309225588832e-05,
+      "loss": 1.3203,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9818009948215417e-05,
+      "loss": 1.1953,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9817710426826413e-05,
+      "loss": 0.8906,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.981741066142926e-05,
+      "loss": 1.0859,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9817110652031412e-05,
+      "loss": 1.5859,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.981681039864032e-05,
+      "loss": 0.6523,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9816509901263453e-05,
+      "loss": 0.9453,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9816209159908273e-05,
+      "loss": 1.0391,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9815908174582262e-05,
+      "loss": 1.2656,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9815606945292896e-05,
+      "loss": 1.1016,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.981530547204766e-05,
+      "loss": 1.2109,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9815003754854052e-05,
+      "loss": 1.5312,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9814701793719567e-05,
+      "loss": 0.9375,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9814399588651708e-05,
+      "loss": 1.625,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9814097139657992e-05,
+      "loss": 1.3125,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9813794446745936e-05,
+      "loss": 1.1797,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9813491509923058e-05,
+      "loss": 1.7812,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9813188329196884e-05,
+      "loss": 1.1875,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.981288490457496e-05,
+      "loss": 1.1328,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9812581236064822e-05,
+      "loss": 1.5938,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.981227732367401e-05,
+      "loss": 1.4844,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.981197316741009e-05,
+      "loss": 0.9375,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.981166876728062e-05,
+      "loss": 1.1719,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9811364123293157e-05,
+      "loss": 2.0781,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9811059235455273e-05,
+      "loss": 1.5391,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9810754103774554e-05,
+      "loss": 0.8516,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.981044872825858e-05,
+      "loss": 0.6055,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9810143108914938e-05,
+      "loss": 0.9414,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9809837245751227e-05,
+      "loss": 0.8711,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9809531138775048e-05,
+      "loss": 1.2188,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9809224787994008e-05,
+      "loss": 1.0859,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9808918193415723e-05,
+      "loss": 1.2578,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9808611355047808e-05,
+      "loss": 1.2266,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9808304272897897e-05,
+      "loss": 1.375,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980799694697362e-05,
+      "loss": 0.8359,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980768937728261e-05,
+      "loss": 0.8477,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9807381563832514e-05,
+      "loss": 1.0234,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9807073506630984e-05,
+      "loss": 0.875,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9806765205685674e-05,
+      "loss": 0.9609,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980645666100425e-05,
+      "loss": 1.2812,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9806147872594377e-05,
+      "loss": 0.5547,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9805838840463732e-05,
+      "loss": 1.3281,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9805529564619997e-05,
+      "loss": 0.8359,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9805220045070853e-05,
+      "loss": 1.1328,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9804910281824e-05,
+      "loss": 1.5234,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9804600274887128e-05,
+      "loss": 0.7734,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980429002426795e-05,
+      "loss": 1.5469,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980397952997417e-05,
+      "loss": 0.3262,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9803668792013513e-05,
+      "loss": 0.6758,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9803357810393697e-05,
+      "loss": 1.8672,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980304658512245e-05,
+      "loss": 1.6484,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980273511620751e-05,
+      "loss": 1.1719,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980242340365662e-05,
+      "loss": 1.3984,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9802111447477523e-05,
+      "loss": 0.7305,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.980179924767797e-05,
+      "loss": 0.793,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9801486804265727e-05,
+      "loss": 1.5,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9801174117248555e-05,
+      "loss": 0.9023,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9800861186634232e-05,
+      "loss": 1.3203,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9800548012430526e-05,
+      "loss": 0.6055,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9800234594645227e-05,
+      "loss": 1.1328,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9799920933286124e-05,
+      "loss": 1.375,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979960702836101e-05,
+      "loss": 1.3906,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9799292879877687e-05,
+      "loss": 0.7617,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979897848784397e-05,
+      "loss": 1.3828,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979866385226766e-05,
+      "loss": 1.6016,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9798348973156588e-05,
+      "loss": 1.7031,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9798033850518572e-05,
+      "loss": 1.25,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9797718484361453e-05,
+      "loss": 1.1328,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9797402874693062e-05,
+      "loss": 1.0938,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9797087021521245e-05,
+      "loss": 0.75,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979677092485385e-05,
+      "loss": 0.7539,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979645458469874e-05,
+      "loss": 0.8203,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979613800106377e-05,
+      "loss": 1.2188,
+      "step": 1283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9795821173956812e-05,
+      "loss": 1.6172,
+      "step": 1284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9795504103385736e-05,
+      "loss": 1.1875,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9795186789358428e-05,
+      "loss": 1.125,
+      "step": 1286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9794869231882775e-05,
+      "loss": 0.8242,
+      "step": 1287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9794551430966665e-05,
+      "loss": 1.0,
+      "step": 1288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9794233386618002e-05,
+      "loss": 1.1719,
+      "step": 1289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9793915098844684e-05,
+      "loss": 0.8477,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9793596567654622e-05,
+      "loss": 0.9453,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979327779305574e-05,
+      "loss": 1.7812,
+      "step": 1292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9792958775055953e-05,
+      "loss": 1.2969,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9792639513663196e-05,
+      "loss": 1.1641,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.97923200088854e-05,
+      "loss": 1.4453,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979200026073051e-05,
+      "loss": 0.8242,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979168026920647e-05,
+      "loss": 0.9922,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979136003432123e-05,
+      "loss": 1.375,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9791039556082748e-05,
+      "loss": 1.125,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9790718834499e-05,
+      "loss": 1.4297,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979039786957795e-05,
+      "loss": 0.3887,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.979007666132757e-05,
+      "loss": 1.4219,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9789755209755853e-05,
+      "loss": 0.8789,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9789433514870784e-05,
+      "loss": 1.4062,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.978911157668036e-05,
+      "loss": 0.9922,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.978878939519258e-05,
+      "loss": 0.5586,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9788466970415448e-05,
+      "loss": 1.6406,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9788144302356987e-05,
+      "loss": 0.7617,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9787821391025213e-05,
+      "loss": 0.6211,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9787498236428147e-05,
+      "loss": 0.582,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9787174838573824e-05,
+      "loss": 1.2812,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9786851197470283e-05,
+      "loss": 1.6875,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9786527313125565e-05,
+      "loss": 0.7344,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9786203185547722e-05,
+      "loss": 0.6953,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9785878814744806e-05,
+      "loss": 1.25,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9785554200724884e-05,
+      "loss": 0.625,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9785229343496022e-05,
+      "loss": 1.2422,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.978490424306629e-05,
+      "loss": 0.8516,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9784578899443775e-05,
+      "loss": 2.1719,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.978425331263656e-05,
+      "loss": 1.0625,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9783927482652735e-05,
+      "loss": 1.0703,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.97836014095004e-05,
+      "loss": 1.5312,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.978327509318766e-05,
+      "loss": 1.1328,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9782948533722624e-05,
+      "loss": 1.125,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.978262173111341e-05,
+      "loss": 0.957,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9782294685368133e-05,
+      "loss": 1.2344,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.978196739649493e-05,
+      "loss": 0.8477,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9781639864501935e-05,
+      "loss": 0.9766,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9781312089397286e-05,
+      "loss": 0.7266,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9780984071189127e-05,
+      "loss": 0.707,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9780655809885615e-05,
+      "loss": 1.2891,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9780327305494906e-05,
+      "loss": 1.5703,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9779998558025165e-05,
+      "loss": 0.5039,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977966956748456e-05,
+      "loss": 1.5312,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9779340333881274e-05,
+      "loss": 0.8008,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9779010857223488e-05,
+      "loss": 0.8125,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9778681137519386e-05,
+      "loss": 0.7461,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9778351174777164e-05,
+      "loss": 0.7383,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977802096900503e-05,
+      "loss": 1.1797,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977769052021118e-05,
+      "loss": 0.8008,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9777359828403836e-05,
+      "loss": 1.2656,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977702889359121e-05,
+      "loss": 1.3203,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9776697715781538e-05,
+      "loss": 0.9219,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9776366294983035e-05,
+      "loss": 0.9766,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977603463120395e-05,
+      "loss": 1.2578,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9775702724452524e-05,
+      "loss": 1.1094,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9775370574737007e-05,
+      "loss": 1.0,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9775038182065646e-05,
+      "loss": 1.1328,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977470554644671e-05,
+      "loss": 0.6758,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977437266788847e-05,
+      "loss": 0.9219,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977403954639919e-05,
+      "loss": 0.6367,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977370618198715e-05,
+      "loss": 1.9375,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9773372574660646e-05,
+      "loss": 1.3203,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9773038724427956e-05,
+      "loss": 0.7148,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9772704631297385e-05,
+      "loss": 0.5273,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977237029527724e-05,
+      "loss": 0.793,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977203571637582e-05,
+      "loss": 1.1562,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977170089460145e-05,
+      "loss": 1.4219,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9771365829962444e-05,
+      "loss": 1.1641,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977103052246714e-05,
+      "loss": 1.4922,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9770694972123858e-05,
+      "loss": 0.7891,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.977035917894095e-05,
+      "loss": 0.3457,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9770023142926752e-05,
+      "loss": 0.8281,
+      "step": 1363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9769686864089626e-05,
+      "loss": 0.8555,
+      "step": 1364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.976935034243792e-05,
+      "loss": 0.9648,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9769013577980005e-05,
+      "loss": 0.8125,
+      "step": 1366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9768676570724248e-05,
+      "loss": 1.0547,
+      "step": 1367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9768339320679025e-05,
+      "loss": 0.5156,
+      "step": 1368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9768001827852717e-05,
+      "loss": 0.707,
+      "step": 1369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.976766409225371e-05,
+      "loss": 1.3516,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9767326113890408e-05,
+      "loss": 0.582,
+      "step": 1371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9766987892771202e-05,
+      "loss": 0.6797,
+      "step": 1372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.97666494289045e-05,
+      "loss": 1.2656,
+      "step": 1373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.976631072229871e-05,
+      "loss": 1.0938,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9765971772962258e-05,
+      "loss": 1.3828,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9765632580903564e-05,
+      "loss": 1.1797,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9765293146131062e-05,
+      "loss": 1.4922,
+      "step": 1377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.976495346865318e-05,
+      "loss": 0.9727,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.976461354847837e-05,
+      "loss": 1.7344,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9764273385615076e-05,
+      "loss": 1.7969,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.976393298007175e-05,
+      "loss": 0.9648,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9763592331856856e-05,
+      "loss": 0.5469,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9763251440978858e-05,
+      "loss": 0.4297,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9762910307446232e-05,
+      "loss": 1.1172,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9762568931267452e-05,
+      "loss": 0.4004,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9762227312451005e-05,
+      "loss": 0.5703,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9761885451005383e-05,
+      "loss": 0.9258,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9761543346939076e-05,
+      "loss": 0.7344,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9761201000260594e-05,
+      "loss": 0.9141,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9760858410978446e-05,
+      "loss": 1.1094,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.976051557910114e-05,
+      "loss": 0.8672,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9760172504637203e-05,
+      "loss": 1.3906,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9759829187595154e-05,
+      "loss": 1.4453,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9759485627983537e-05,
+      "loss": 0.707,
+      "step": 1394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975914182581088e-05,
+      "loss": 0.9648,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9758797781085733e-05,
+      "loss": 1.0,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975845349381665e-05,
+      "loss": 0.7852,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975810896401218e-05,
+      "loss": 0.8867,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9757764191680894e-05,
+      "loss": 1.1484,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9757419176831355e-05,
+      "loss": 0.875,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9757073919472137e-05,
+      "loss": 1.2656,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9756728419611828e-05,
+      "loss": 0.5625,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975638267725901e-05,
+      "loss": 0.6289,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9756036692422276e-05,
+      "loss": 0.5586,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975569046511023e-05,
+      "loss": 0.5234,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9755343995331467e-05,
+      "loss": 1.3359,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9754997283094604e-05,
+      "loss": 0.9648,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9754650328408266e-05,
+      "loss": 0.707,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9754303131281064e-05,
+      "loss": 0.6523,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975395569172163e-05,
+      "loss": 0.6992,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9753608009738604e-05,
+      "loss": 0.4121,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9753260085340623e-05,
+      "loss": 0.8789,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9752911918536333e-05,
+      "loss": 0.6562,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9752563509334393e-05,
+      "loss": 0.9648,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975221485774346e-05,
+      "loss": 0.7461,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9751865963772198e-05,
+      "loss": 1.4844,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9751516827429277e-05,
+      "loss": 0.7227,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9751167448723376e-05,
+      "loss": 0.5508,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9750817827663178e-05,
+      "loss": 1.6328,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9750467964257374e-05,
+      "loss": 0.8516,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.975011785851466e-05,
+      "loss": 1.2344,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9749767510443733e-05,
+      "loss": 0.7812,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9749416920053308e-05,
+      "loss": 1.4375,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.974906608735209e-05,
+      "loss": 1.0312,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9748715012348806e-05,
+      "loss": 1.9688,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9748363695052174e-05,
+      "loss": 0.5977,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9748012135470934e-05,
+      "loss": 1.2891,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9747660333613818e-05,
+      "loss": 0.4062,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9747308289489568e-05,
+      "loss": 1.3281,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.974695600310694e-05,
+      "loss": 1.1719,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9746603474474684e-05,
+      "loss": 1.0234,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9746250703601566e-05,
+      "loss": 1.125,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.974589769049635e-05,
+      "loss": 1.0234,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9745544435167815e-05,
+      "loss": 1.1094,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9745190937624735e-05,
+      "loss": 0.6797,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9744837197875898e-05,
+      "loss": 1.7734,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9744483215930095e-05,
+      "loss": 0.4258,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.974412899179613e-05,
+      "loss": 2.0781,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9743774525482796e-05,
+      "loss": 1.1094,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9743419816998907e-05,
+      "loss": 0.9883,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9743064866353287e-05,
+      "loss": 2.0156,
+      "step": 1441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9742709673554743e-05,
+      "loss": 1.0234,
+      "step": 1442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.974235423861212e-05,
+      "loss": 1.1406,
+      "step": 1443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9741998561534234e-05,
+      "loss": 1.3594,
+      "step": 1444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9741642642329938e-05,
+      "loss": 1.4375,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9741286481008075e-05,
+      "loss": 1.2422,
+      "step": 1446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9740930077577496e-05,
+      "loss": 1.1719,
+      "step": 1447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.974057343204706e-05,
+      "loss": 1.2344,
+      "step": 1448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9740216544425624e-05,
+      "loss": 0.6133,
+      "step": 1449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9739859414722068e-05,
+      "loss": 0.9336,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9739502042945267e-05,
+      "loss": 1.3125,
+      "step": 1451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9739144429104095e-05,
+      "loss": 0.9961,
+      "step": 1452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9738786573207453e-05,
+      "loss": 1.2578,
+      "step": 1453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973842847526422e-05,
+      "loss": 0.8789,
+      "step": 1454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9738070135283306e-05,
+      "loss": 0.7344,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9737711553273613e-05,
+      "loss": 1.3516,
+      "step": 1456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9737352729244055e-05,
+      "loss": 1.5938,
+      "step": 1457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973699366320355e-05,
+      "loss": 1.5469,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9736634355161023e-05,
+      "loss": 1.6953,
+      "step": 1459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9736274805125408e-05,
+      "loss": 0.5078,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973591501310563e-05,
+      "loss": 0.4844,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9735554979110638e-05,
+      "loss": 0.6016,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9735194703149382e-05,
+      "loss": 1.2891,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9734834185230814e-05,
+      "loss": 1.4766,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9734473425363895e-05,
+      "loss": 0.5195,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9734112423557586e-05,
+      "loss": 1.8438,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973375117982087e-05,
+      "loss": 0.7188,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9733389694162717e-05,
+      "loss": 1.2344,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9733027966592115e-05,
+      "loss": 2.0156,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9732665997118054e-05,
+      "loss": 1.3984,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9732303785749526e-05,
+      "loss": 1.1406,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973194133249554e-05,
+      "loss": 0.8242,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9731578637365095e-05,
+      "loss": 1.5078,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973121570036722e-05,
+      "loss": 1.4375,
+      "step": 1474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.973085252151092e-05,
+      "loss": 0.9688,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9730489100805232e-05,
+      "loss": 0.6875,
+      "step": 1476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9730125438259185e-05,
+      "loss": 2.0,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9729761533881816e-05,
+      "loss": 1.6797,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972939738768217e-05,
+      "loss": 1.7969,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97290329996693e-05,
+      "loss": 1.4297,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9728668369852256e-05,
+      "loss": 1.1719,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9728303498240108e-05,
+      "loss": 1.5469,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9727938384841924e-05,
+      "loss": 0.9297,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972757302966677e-05,
+      "loss": 1.5938,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9727207432723738e-05,
+      "loss": 1.3359,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9726841594021903e-05,
+      "loss": 1.0469,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972647551357037e-05,
+      "loss": 1.5859,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9726109191378223e-05,
+      "loss": 1.5,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9725742627454582e-05,
+      "loss": 1.0234,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9725375821808547e-05,
+      "loss": 0.8672,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9725008774449237e-05,
+      "loss": 1.2031,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9724641485385776e-05,
+      "loss": 1.0938,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972427395462729e-05,
+      "loss": 0.7031,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972390618218292e-05,
+      "loss": 1.2422,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972353816806179e-05,
+      "loss": 0.6992,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972316991227307e-05,
+      "loss": 0.832,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97228014148259e-05,
+      "loss": 1.2266,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972243267572943e-05,
+      "loss": 0.9492,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9722063694992845e-05,
+      "loss": 1.0391,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9721694472625297e-05,
+      "loss": 0.9883,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9721325008635977e-05,
+      "loss": 1.2266,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9720955303034057e-05,
+      "loss": 0.8359,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972058535582873e-05,
+      "loss": 0.6172,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.972021516702919e-05,
+      "loss": 1.4297,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9719844736644636e-05,
+      "loss": 0.8789,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971947406468428e-05,
+      "loss": 0.9375,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971910315115733e-05,
+      "loss": 0.2539,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9718731996073005e-05,
+      "loss": 0.7695,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971836059944053e-05,
+      "loss": 0.9141,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9717988961269134e-05,
+      "loss": 0.8047,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971761708156806e-05,
+      "loss": 0.9844,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9717244960346544e-05,
+      "loss": 0.8203,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9716872597613838e-05,
+      "loss": 1.4688,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971649999337919e-05,
+      "loss": 0.7656,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9716127147651873e-05,
+      "loss": 1.1094,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9715754060441147e-05,
+      "loss": 1.3672,
+      "step": 1516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971538073175628e-05,
+      "loss": 1.9062,
+      "step": 1517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9715007161606557e-05,
+      "loss": 0.9531,
+      "step": 1518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971463335000126e-05,
+      "loss": 1.4766,
+      "step": 1519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9714259296949683e-05,
+      "loss": 1.0625,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971388500246112e-05,
+      "loss": 1.6016,
+      "step": 1521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971351046654487e-05,
+      "loss": 0.8398,
+      "step": 1522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971313568921025e-05,
+      "loss": 0.5938,
+      "step": 1523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9712760670466567e-05,
+      "loss": 1.5078,
+      "step": 1524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9712385410323146e-05,
+      "loss": 1.2344,
+      "step": 1525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9712009908789315e-05,
+      "loss": 0.6016,
+      "step": 1526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9711634165874404e-05,
+      "loss": 1.0156,
+      "step": 1527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971125818158775e-05,
+      "loss": 0.8984,
+      "step": 1528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9710881955938697e-05,
+      "loss": 1.1562,
+      "step": 1529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9710505488936603e-05,
+      "loss": 0.9102,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9710128780590815e-05,
+      "loss": 1.4609,
+      "step": 1531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9709751830910707e-05,
+      "loss": 0.9219,
+      "step": 1532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9709374639905636e-05,
+      "loss": 0.875,
+      "step": 1533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9708997207584982e-05,
+      "loss": 1.4375,
+      "step": 1534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.970861953395813e-05,
+      "loss": 0.7891,
+      "step": 1535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9708241619034457e-05,
+      "loss": 1.1797,
+      "step": 1536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9707863462823363e-05,
+      "loss": 1.3359,
+      "step": 1537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9707485065334243e-05,
+      "loss": 0.4609,
+      "step": 1538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9707106426576505e-05,
+      "loss": 1.9453,
+      "step": 1539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9706727546559555e-05,
+      "loss": 1.3516,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9706348425292816e-05,
+      "loss": 0.7852,
+      "step": 1541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9705969062785707e-05,
+      "loss": 1.4531,
+      "step": 1542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9705589459047652e-05,
+      "loss": 1.2031,
+      "step": 1543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9705209614088096e-05,
+      "loss": 1.2266,
+      "step": 1544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.970482952791647e-05,
+      "loss": 0.7773,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9704449200542228e-05,
+      "loss": 0.9727,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9704068631974818e-05,
+      "loss": 1.3516,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9703687822223698e-05,
+      "loss": 1.1875,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9703306771298333e-05,
+      "loss": 0.9922,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9702925479208198e-05,
+      "loss": 0.9062,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9702543945962764e-05,
+      "loss": 1.6484,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.970216217157152e-05,
+      "loss": 1.5547,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9701780156043947e-05,
+      "loss": 1.1016,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9701397899389544e-05,
+      "loss": 1.5078,
+      "step": 1554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9701015401617813e-05,
+      "loss": 1.2812,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9700632662738256e-05,
+      "loss": 1.25,
+      "step": 1556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.970024968276039e-05,
+      "loss": 0.8242,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9699866461693727e-05,
+      "loss": 1.7188,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.96994829995478e-05,
+      "loss": 1.0234,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9699099296332135e-05,
+      "loss": 0.5586,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9698715352056267e-05,
+      "loss": 0.668,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969833116672974e-05,
+      "loss": 0.3359,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9697946740362106e-05,
+      "loss": 0.7422,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9697562072962915e-05,
+      "loss": 0.6641,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969717716454173e-05,
+      "loss": 0.7852,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9696792015108116e-05,
+      "loss": 1.2344,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9696406624671643e-05,
+      "loss": 1.1484,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9696020993241897e-05,
+      "loss": 0.6914,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969563512082845e-05,
+      "loss": 0.6914,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9695249007440908e-05,
+      "loss": 0.6523,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9694862653088856e-05,
+      "loss": 0.6367,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.96944760577819e-05,
+      "loss": 1.7422,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969408922152965e-05,
+      "loss": 0.8906,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969370214434172e-05,
+      "loss": 1.2422,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9693314826227726e-05,
+      "loss": 1.3203,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.96929272671973e-05,
+      "loss": 1.1641,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969253946726007e-05,
+      "loss": 1.25,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9692151426425673e-05,
+      "loss": 1.3672,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969176314470376e-05,
+      "loss": 0.6914,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969137462210398e-05,
+      "loss": 1.1641,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9690985858635985e-05,
+      "loss": 1.2891,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9690596854309438e-05,
+      "loss": 1.0078,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9690207609134012e-05,
+      "loss": 0.8242,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9689818123119375e-05,
+      "loss": 1.4609,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968942839627521e-05,
+      "loss": 1.25,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9689038428611202e-05,
+      "loss": 1.4062,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9688648220137047e-05,
+      "loss": 0.4043,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9688257770862444e-05,
+      "loss": 1.375,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968786708079709e-05,
+      "loss": 1.1875,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.96874761499507e-05,
+      "loss": 1.2344,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9687084978332984e-05,
+      "loss": 1.3984,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9686693565953673e-05,
+      "loss": 1.2656,
+      "step": 1592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9686301912822495e-05,
+      "loss": 1.0469,
+      "step": 1593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9685910018949175e-05,
+      "loss": 1.6875,
+      "step": 1594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968551788434346e-05,
+      "loss": 1.3594,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9685125509015096e-05,
+      "loss": 1.3828,
+      "step": 1596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9684732892973835e-05,
+      "loss": 0.9102,
+      "step": 1597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9684340036229432e-05,
+      "loss": 1.0312,
+      "step": 1598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9683946938791653e-05,
+      "loss": 0.7461,
+      "step": 1599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9683553600670267e-05,
+      "loss": 0.7695,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968316002187505e-05,
+      "loss": 0.3398,
+      "step": 1601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968276620241579e-05,
+      "loss": 0.4648,
+      "step": 1602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9682372142302263e-05,
+      "loss": 0.8242,
+      "step": 1603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9681977841544275e-05,
+      "loss": 0.4395,
+      "step": 1604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968158330015162e-05,
+      "loss": 1.5078,
+      "step": 1605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.96811885181341e-05,
+      "loss": 1.5312,
+      "step": 1606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9680793495501532e-05,
+      "loss": 1.3594,
+      "step": 1607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9680398232263736e-05,
+      "loss": 0.5312,
+      "step": 1608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968000272843053e-05,
+      "loss": 1.1328,
+      "step": 1609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.967960698401175e-05,
+      "loss": 0.9102,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9679210999017228e-05,
+      "loss": 0.7578,
+      "step": 1611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9678814773456803e-05,
+      "loss": 0.7266,
+      "step": 1612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.967841830734033e-05,
+      "loss": 1.1172,
+      "step": 1613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9678021600677654e-05,
+      "loss": 0.6484,
+      "step": 1614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9677624653478643e-05,
+      "loss": 1.6406,
+      "step": 1615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9677227465753157e-05,
+      "loss": 0.9766,
+      "step": 1616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.967683003751107e-05,
+      "loss": 0.6602,
+      "step": 1617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9676432368762263e-05,
+      "loss": 0.8477,
+      "step": 1618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.967603445951661e-05,
+      "loss": 0.5977,
+      "step": 1619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9675636309784008e-05,
+      "loss": 1.4688,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9675237919574355e-05,
+      "loss": 1.0312,
+      "step": 1621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9674839288897548e-05,
+      "loss": 1.4766,
+      "step": 1622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9674440417763492e-05,
+      "loss": 1.4922,
+      "step": 1623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9674041306182102e-05,
+      "loss": 1.1641,
+      "step": 1624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9673641954163302e-05,
+      "loss": 0.9609,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9673242361717016e-05,
+      "loss": 0.8516,
+      "step": 1626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9672842528853173e-05,
+      "loss": 0.8828,
+      "step": 1627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9672442455581712e-05,
+      "loss": 1.0781,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9672042141912577e-05,
+      "loss": 1.1172,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.967164158785571e-05,
+      "loss": 1.3281,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.967124079342108e-05,
+      "loss": 0.6172,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9670839758618633e-05,
+      "loss": 1.5703,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.967043848345835e-05,
+      "loss": 1.3203,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9670036967950195e-05,
+      "loss": 0.9531,
+      "step": 1634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9669635212104153e-05,
+      "loss": 0.9062,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9669233215930204e-05,
+      "loss": 0.6406,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966883097943834e-05,
+      "loss": 0.957,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9668428502638566e-05,
+      "loss": 1.2891,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9668025785540877e-05,
+      "loss": 1.4297,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966762282815528e-05,
+      "loss": 0.918,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.96672196304918e-05,
+      "loss": 1.0156,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966681619256045e-05,
+      "loss": 1.0547,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9666412514371255e-05,
+      "loss": 0.832,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9666008595934258e-05,
+      "loss": 1.2578,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966560443725949e-05,
+      "loss": 1.0547,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9665200038357e-05,
+      "loss": 0.6172,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9664795399236835e-05,
+      "loss": 0.9336,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966439051990905e-05,
+      "loss": 0.7852,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9663985400383715e-05,
+      "loss": 0.9688,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9663580040670895e-05,
+      "loss": 0.9844,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966317444078067e-05,
+      "loss": 1.9844,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966276860072311e-05,
+      "loss": 1.3828,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9662362520508308e-05,
+      "loss": 1.3906,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9661956200146356e-05,
+      "loss": 1.3438,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9661549639647354e-05,
+      "loss": 1.2891,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9661142839021407e-05,
+      "loss": 0.7695,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966073579827862e-05,
+      "loss": 0.9453,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9660328517429116e-05,
+      "loss": 0.7227,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9659920996483016e-05,
+      "loss": 1.0078,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9659513235450446e-05,
+      "loss": 0.6641,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9659105234341545e-05,
+      "loss": 0.5,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9658696993166447e-05,
+      "loss": 0.8242,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.96582885119353e-05,
+      "loss": 0.4414,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9657879790658262e-05,
+      "loss": 0.9805,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9657470829345487e-05,
+      "loss": 1.3906,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965706162800714e-05,
+      "loss": 1.1953,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965665218665339e-05,
+      "loss": 0.9688,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9656242505294413e-05,
+      "loss": 0.5586,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9655832583940394e-05,
+      "loss": 1.0625,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9655422422601517e-05,
+      "loss": 0.6211,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965501202128798e-05,
+      "loss": 1.0469,
+      "step": 1671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965460138000998e-05,
+      "loss": 1.3359,
+      "step": 1672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965419049877773e-05,
+      "loss": 0.9609,
+      "step": 1673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965377937760143e-05,
+      "loss": 1.0547,
+      "step": 1674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965336801649131e-05,
+      "loss": 1.8438,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9652956415457585e-05,
+      "loss": 0.4082,
+      "step": 1676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965254457451049e-05,
+      "loss": 0.6094,
+      "step": 1677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9652132493660257e-05,
+      "loss": 0.7461,
+      "step": 1678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9651720172917133e-05,
+      "loss": 0.9727,
+      "step": 1679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.965130761229136e-05,
+      "loss": 0.9453,
+      "step": 1680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9650894811793194e-05,
+      "loss": 0.9023,
+      "step": 1681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9650481771432898e-05,
+      "loss": 1.3359,
+      "step": 1682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9650068491220735e-05,
+      "loss": 0.6719,
+      "step": 1683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9649654971166978e-05,
+      "loss": 1.0547,
+      "step": 1684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9649241211281896e-05,
+      "loss": 1.0078,
+      "step": 1685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9648827211575784e-05,
+      "loss": 0.5547,
+      "step": 1686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9648412972058926e-05,
+      "loss": 1.1328,
+      "step": 1687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9647998492741624e-05,
+      "loss": 0.9297,
+      "step": 1688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964758377363417e-05,
+      "loss": 1.1719,
+      "step": 1689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9647168814746873e-05,
+      "loss": 0.9766,
+      "step": 1690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964675361609005e-05,
+      "loss": 0.9492,
+      "step": 1691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964633817767402e-05,
+      "loss": 0.9883,
+      "step": 1692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9645922499509103e-05,
+      "loss": 0.4785,
+      "step": 1693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964550658160564e-05,
+      "loss": 0.5742,
+      "step": 1694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964509042397396e-05,
+      "loss": 0.6758,
+      "step": 1695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964467402662441e-05,
+      "loss": 0.8672,
+      "step": 1696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964425738956734e-05,
+      "loss": 0.9141,
+      "step": 1697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.96438405128131e-05,
+      "loss": 0.7031,
+      "step": 1698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964342339637205e-05,
+      "loss": 0.8594,
+      "step": 1699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9643006040254568e-05,
+      "loss": 0.8672,
+      "step": 1700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9642588444471016e-05,
+      "loss": 1.8047,
+      "step": 1701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9642170609031777e-05,
+      "loss": 0.4727,
+      "step": 1702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9641752533947236e-05,
+      "loss": 0.5742,
+      "step": 1703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964133421922778e-05,
+      "loss": 1.75,
+      "step": 1704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9640915664883816e-05,
+      "loss": 0.8398,
+      "step": 1705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9640496870925733e-05,
+      "loss": 0.9336,
+      "step": 1706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.964007783736395e-05,
+      "loss": 0.7305,
+      "step": 1707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9639658564208876e-05,
+      "loss": 1.2344,
+      "step": 1708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9639239051470933e-05,
+      "loss": 1.5781,
+      "step": 1709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.963881929916055e-05,
+      "loss": 1.2109,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9638399307288156e-05,
+      "loss": 0.7695,
+      "step": 1711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.963797907586419e-05,
+      "loss": 0.6406,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.96375586048991e-05,
+      "loss": 0.5977,
+      "step": 1713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.963713789440333e-05,
+      "loss": 0.7734,
+      "step": 1714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9636716944387345e-05,
+      "loss": 1.2266,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9636295754861597e-05,
+      "loss": 1.6172,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9635874325836565e-05,
+      "loss": 0.9102,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9635452657322714e-05,
+      "loss": 0.7617,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9635030749330525e-05,
+      "loss": 0.5742,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9634608601870494e-05,
+      "loss": 1.0234,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96341862149531e-05,
+      "loss": 1.5078,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9633763588588847e-05,
+      "loss": 0.9102,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9633340722788237e-05,
+      "loss": 1.3203,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9632917617561784e-05,
+      "loss": 1.375,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9632494272919998e-05,
+      "loss": 1.6641,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.963207068887341e-05,
+      "loss": 0.7852,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.963164686543254e-05,
+      "loss": 0.5664,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.963122280260792e-05,
+      "loss": 1.7109,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9630798500410094e-05,
+      "loss": 0.6992,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9630373958849604e-05,
+      "loss": 1.2734,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9629949177937008e-05,
+      "loss": 1.1562,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962952415768286e-05,
+      "loss": 1.1719,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9629098898097717e-05,
+      "loss": 1.4453,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962867339919216e-05,
+      "loss": 1.3984,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9628247660976754e-05,
+      "loss": 1.1094,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962782168346209e-05,
+      "loss": 1.4688,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9627395466658746e-05,
+      "loss": 0.8555,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962696901057732e-05,
+      "loss": 1.125,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9626542315228408e-05,
+      "loss": 1.0469,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962611538062262e-05,
+      "loss": 0.6289,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9625688206770563e-05,
+      "loss": 1.4922,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9625260793682855e-05,
+      "loss": 0.3398,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962483314137012e-05,
+      "loss": 1.25,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962440524984298e-05,
+      "loss": 1.6172,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962397711911208e-05,
+      "loss": 1.5625,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9623548749188056e-05,
+      "loss": 1.1094,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9623120140081557e-05,
+      "loss": 1.0703,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962269129180323e-05,
+      "loss": 0.5781,
+      "step": 1748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962226220436374e-05,
+      "loss": 1.4766,
+      "step": 1749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9621832877773745e-05,
+      "loss": 0.8555,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9621403312043922e-05,
+      "loss": 1.4375,
+      "step": 1751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962097350718494e-05,
+      "loss": 1.0859,
+      "step": 1752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.962054346320749e-05,
+      "loss": 2.0469,
+      "step": 1753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9620113180122255e-05,
+      "loss": 0.7266,
+      "step": 1754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961968265793993e-05,
+      "loss": 1.0,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9619251896671212e-05,
+      "loss": 1.2891,
+      "step": 1756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9618820896326813e-05,
+      "loss": 0.6758,
+      "step": 1757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9618389656917444e-05,
+      "loss": 1.0078,
+      "step": 1758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961795817845382e-05,
+      "loss": 1.3359,
+      "step": 1759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961752646094666e-05,
+      "loss": 0.9023,
+      "step": 1760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9617094504406706e-05,
+      "loss": 0.6289,
+      "step": 1761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9616662308844686e-05,
+      "loss": 0.8828,
+      "step": 1762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9616229874271348e-05,
+      "loss": 1.1016,
+      "step": 1763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961579720069743e-05,
+      "loss": 1.7266,
+      "step": 1764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9615364288133693e-05,
+      "loss": 0.7383,
+      "step": 1765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9614931136590892e-05,
+      "loss": 0.5625,
+      "step": 1766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96144977460798e-05,
+      "loss": 0.6445,
+      "step": 1767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961406411661118e-05,
+      "loss": 0.9766,
+      "step": 1768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9613630248195812e-05,
+      "loss": 1.1719,
+      "step": 1769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9613196140844482e-05,
+      "loss": 0.8867,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9612761794567977e-05,
+      "loss": 1.4844,
+      "step": 1771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961232720937709e-05,
+      "loss": 1.0703,
+      "step": 1772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961189238528263e-05,
+      "loss": 0.8438,
+      "step": 1773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9611457322295392e-05,
+      "loss": 1.5156,
+      "step": 1774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96110220204262e-05,
+      "loss": 1.375,
+      "step": 1775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9610586479685864e-05,
+      "loss": 0.8594,
+      "step": 1776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.961015070008522e-05,
+      "loss": 1.8594,
+      "step": 1777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9609714681635088e-05,
+      "loss": 1.25,
+      "step": 1778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960927842434631e-05,
+      "loss": 0.9922,
+      "step": 1779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9608841928229725e-05,
+      "loss": 1.3047,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960840519329619e-05,
+      "loss": 1.3594,
+      "step": 1781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960796821955655e-05,
+      "loss": 0.957,
+      "step": 1782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9607531007021666e-05,
+      "loss": 1.1016,
+      "step": 1783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9607093555702414e-05,
+      "loss": 1.4297,
+      "step": 1784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9606655865609657e-05,
+      "loss": 0.9844,
+      "step": 1785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960621793675428e-05,
+      "loss": 1.3359,
+      "step": 1786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960577976914716e-05,
+      "loss": 1.2656,
+      "step": 1787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960534136279919e-05,
+      "loss": 1.2109,
+      "step": 1788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9604902717721267e-05,
+      "loss": 1.375,
+      "step": 1789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9604463833924296e-05,
+      "loss": 0.9922,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9604024711419184e-05,
+      "loss": 1.8984,
+      "step": 1791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960358535021684e-05,
+      "loss": 1.0469,
+      "step": 1792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9603145750328183e-05,
+      "loss": 1.2891,
+      "step": 1793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9602705911764145e-05,
+      "loss": 1.5,
+      "step": 1794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9602265834535655e-05,
+      "loss": 1.25,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9601825518653653e-05,
+      "loss": 0.8516,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9601384964129077e-05,
+      "loss": 1.1094,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9600944170972883e-05,
+      "loss": 0.9766,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9600503139196022e-05,
+      "loss": 0.793,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9600061868809455e-05,
+      "loss": 1.1562,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.959962035982415e-05,
+      "loss": 1.8594,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9599178612251085e-05,
+      "loss": 0.918,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9598736626101233e-05,
+      "loss": 1.0469,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.959829440138558e-05,
+      "loss": 0.5039,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.959785193811512e-05,
+      "loss": 1.25,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9597409236300845e-05,
+      "loss": 0.9258,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9596966295953765e-05,
+      "loss": 1.0156,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9596523117084884e-05,
+      "loss": 0.3438,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9596079699705216e-05,
+      "loss": 0.7266,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9595636043825784e-05,
+      "loss": 0.9492,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9595192149457615e-05,
+      "loss": 1.75,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9594748016611738e-05,
+      "loss": 0.8359,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9594303645299194e-05,
+      "loss": 0.3652,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.959385903553103e-05,
+      "loss": 1.0703,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.959341418731829e-05,
+      "loss": 1.4297,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9592969100672033e-05,
+      "loss": 1.1875,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.959252377560332e-05,
+      "loss": 0.9531,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9592078212123225e-05,
+      "loss": 0.75,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9591632410242818e-05,
+      "loss": 0.6328,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9591186369973175e-05,
+      "loss": 0.9922,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9590740091325386e-05,
+      "loss": 0.7422,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9590293574310543e-05,
+      "loss": 1.1562,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9589846818939743e-05,
+      "loss": 1.1953,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9589399825224088e-05,
+      "loss": 1.5547,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.958895259317469e-05,
+      "loss": 1.5078,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9588505122802665e-05,
+      "loss": 1.0625,
+      "step": 1826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9588057414119133e-05,
+      "loss": 1.6641,
+      "step": 1827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9587609467135218e-05,
+      "loss": 1.0781,
+      "step": 1828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9587161281862056e-05,
+      "loss": 0.9062,
+      "step": 1829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9586712858310787e-05,
+      "loss": 0.8945,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9586264196492556e-05,
+      "loss": 0.9453,
+      "step": 1831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9585815296418515e-05,
+      "loss": 1.1406,
+      "step": 1832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.958536615809982e-05,
+      "loss": 1.6875,
+      "step": 1833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9584916781547627e-05,
+      "loss": 1.2969,
+      "step": 1834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9584467166773118e-05,
+      "loss": 0.8242,
+      "step": 1835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9584017313787456e-05,
+      "loss": 1.2109,
+      "step": 1836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9583567222601827e-05,
+      "loss": 1.7031,
+      "step": 1837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.958311689322742e-05,
+      "loss": 1.4141,
+      "step": 1838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9582666325675422e-05,
+      "loss": 1.2578,
+      "step": 1839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.958221551995703e-05,
+      "loss": 1.0234,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9581764476083455e-05,
+      "loss": 1.6641,
+      "step": 1841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9581313194065904e-05,
+      "loss": 1.3672,
+      "step": 1842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9580861673915592e-05,
+      "loss": 1.3125,
+      "step": 1843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9580409915643742e-05,
+      "loss": 1.2188,
+      "step": 1844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9579957919261586e-05,
+      "loss": 0.8047,
+      "step": 1845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.957950568478035e-05,
+      "loss": 1.1094,
+      "step": 1846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9579053212211276e-05,
+      "loss": 1.2344,
+      "step": 1847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9578600501565616e-05,
+      "loss": 0.7031,
+      "step": 1848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9578147552854613e-05,
+      "loss": 1.1797,
+      "step": 1849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.957769436608953e-05,
+      "loss": 1.4531,
+      "step": 1850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9577240941281628e-05,
+      "loss": 1.0156,
+      "step": 1851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9576787278442176e-05,
+      "loss": 0.75,
+      "step": 1852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9576333377582454e-05,
+      "loss": 1.3672,
+      "step": 1853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9575879238713738e-05,
+      "loss": 0.9805,
+      "step": 1854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9575424861847318e-05,
+      "loss": 1.1875,
+      "step": 1855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9574970246994485e-05,
+      "loss": 1.1328,
+      "step": 1856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.957451539416654e-05,
+      "loss": 1.2891,
+      "step": 1857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9574060303374788e-05,
+      "loss": 0.4668,
+      "step": 1858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9573604974630532e-05,
+      "loss": 0.543,
+      "step": 1859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.95731494079451e-05,
+      "loss": 1.1328,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9572693603329812e-05,
+      "loss": 1.25,
+      "step": 1861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.957223756079599e-05,
+      "loss": 0.7148,
+      "step": 1862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9571781280354972e-05,
+      "loss": 0.668,
+      "step": 1863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9571324762018103e-05,
+      "loss": 0.6875,
+      "step": 1864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9570868005796723e-05,
+      "loss": 1.2891,
+      "step": 1865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9570411011702185e-05,
+      "loss": 1.7812,
+      "step": 1866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.956995377974585e-05,
+      "loss": 0.9141,
+      "step": 1867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.956949630993908e-05,
+      "loss": 1.2422,
+      "step": 1868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.956903860229325e-05,
+      "loss": 2.0469,
+      "step": 1869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9568580656819725e-05,
+      "loss": 1.1172,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9568122473529897e-05,
+      "loss": 1.0,
+      "step": 1871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9567664052435145e-05,
+      "loss": 0.7344,
+      "step": 1872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.956720539354687e-05,
+      "loss": 1.1406,
+      "step": 1873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956674649687647e-05,
+      "loss": 1.5,
+      "step": 1874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9566287362435346e-05,
+      "loss": 1.2656,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9565827990234914e-05,
+      "loss": 1.1719,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956536838028659e-05,
+      "loss": 0.5664,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9564908532601793e-05,
+      "loss": 0.5664,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9564448447191956e-05,
+      "loss": 1.4766,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9563988124068516e-05,
+      "loss": 0.8789,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956352756324291e-05,
+      "loss": 0.875,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9563066764726586e-05,
+      "loss": 0.6445,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9562605728530997e-05,
+      "loss": 1.1953,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95621444546676e-05,
+      "loss": 0.5469,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956168294314786e-05,
+      "loss": 1.5078,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956122119398325e-05,
+      "loss": 0.5781,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9560759207185243e-05,
+      "loss": 1.4375,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956029698276532e-05,
+      "loss": 1.5469,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9559834520734973e-05,
+      "loss": 1.3516,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9559371821105694e-05,
+      "loss": 1.0781,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9558908883888986e-05,
+      "loss": 1.0391,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955844570909635e-05,
+      "loss": 2.0625,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95579822967393e-05,
+      "loss": 1.1016,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955751864682935e-05,
+      "loss": 0.9297,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955705475937803e-05,
+      "loss": 1.2188,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9556590634396867e-05,
+      "loss": 1.1875,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955612627189739e-05,
+      "loss": 0.8086,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9555661671891152e-05,
+      "loss": 0.7344,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9555196834389694e-05,
+      "loss": 1.6484,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9554731759404562e-05,
+      "loss": 0.8164,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9554266446947326e-05,
+      "loss": 0.5977,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955380089702955e-05,
+      "loss": 1.2344,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9553335109662798e-05,
+      "loss": 1.9609,
+      "step": 1903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9552869084858646e-05,
+      "loss": 1.2578,
+      "step": 1904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9552402822628683e-05,
+      "loss": 0.5781,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9551936322984498e-05,
+      "loss": 0.582,
+      "step": 1906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9551469585937677e-05,
+      "loss": 0.8906,
+      "step": 1907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955100261149983e-05,
+      "loss": 0.5391,
+      "step": 1908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9550535399682557e-05,
+      "loss": 0.9375,
+      "step": 1909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955006795049747e-05,
+      "loss": 0.6758,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9549600263956188e-05,
+      "loss": 1.3125,
+      "step": 1911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9549132340070335e-05,
+      "loss": 0.8555,
+      "step": 1912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9548664178851545e-05,
+      "loss": 1.4609,
+      "step": 1913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9548195780311444e-05,
+      "loss": 0.5039,
+      "step": 1914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9547727144461682e-05,
+      "loss": 0.9609,
+      "step": 1915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9547258271313906e-05,
+      "loss": 0.7188,
+      "step": 1916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9546789160879763e-05,
+      "loss": 1.4688,
+      "step": 1917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9546319813170916e-05,
+      "loss": 1.125,
+      "step": 1918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9545850228199033e-05,
+      "loss": 1.4844,
+      "step": 1919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.954538040597578e-05,
+      "loss": 0.9336,
+      "step": 1920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9544910346512838e-05,
+      "loss": 0.6523,
+      "step": 1921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9544440049821887e-05,
+      "loss": 2.0156,
+      "step": 1922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9543969515914615e-05,
+      "loss": 0.377,
+      "step": 1923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.954349874480272e-05,
+      "loss": 0.8867,
+      "step": 1924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95430277364979e-05,
+      "loss": 0.9609,
+      "step": 1925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9542556491011858e-05,
+      "loss": 0.6562,
+      "step": 1926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9542085008356315e-05,
+      "loss": 1.2812,
+      "step": 1927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9541613288542985e-05,
+      "loss": 1.1172,
+      "step": 1928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9541141331583587e-05,
+      "loss": 0.3457,
+      "step": 1929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9540669137489862e-05,
+      "loss": 1.3984,
+      "step": 1930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9540196706273536e-05,
+      "loss": 0.8555,
+      "step": 1931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9539724037946352e-05,
+      "loss": 1.4141,
+      "step": 1932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.953925113252006e-05,
+      "loss": 1.2188,
+      "step": 1933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9538777990006417e-05,
+      "loss": 1.4375,
+      "step": 1934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9538304610417177e-05,
+      "loss": 1.8359,
+      "step": 1935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9537830993764105e-05,
+      "loss": 0.9883,
+      "step": 1936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9537357140058978e-05,
+      "loss": 0.4434,
+      "step": 1937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9536883049313563e-05,
+      "loss": 1.2422,
+      "step": 1938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9536408721539656e-05,
+      "loss": 0.248,
+      "step": 1939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9535934156749034e-05,
+      "loss": 0.6172,
+      "step": 1940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9535459354953498e-05,
+      "loss": 0.9688,
+      "step": 1941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.953498431616485e-05,
+      "loss": 0.7852,
+      "step": 1942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9534509040394895e-05,
+      "loss": 0.8906,
+      "step": 1943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9534033527655444e-05,
+      "loss": 1.4141,
+      "step": 1944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9533557777958313e-05,
+      "loss": 0.7422,
+      "step": 1945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9533081791315333e-05,
+      "loss": 0.9961,
+      "step": 1946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9532605567738324e-05,
+      "loss": 0.918,
+      "step": 1947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9532129107239133e-05,
+      "loss": 0.8438,
+      "step": 1948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9531652409829597e-05,
+      "loss": 0.9336,
+      "step": 1949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9531175475521564e-05,
+      "loss": 1.2266,
+      "step": 1950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9530698304326887e-05,
+      "loss": 1.1797,
+      "step": 1951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9530220896257423e-05,
+      "loss": 1.6719,
+      "step": 1952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9529743251325048e-05,
+      "loss": 1.7734,
+      "step": 1953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952926536954162e-05,
+      "loss": 1.2266,
+      "step": 1954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9528787250919023e-05,
+      "loss": 0.8047,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952830889546914e-05,
+      "loss": 0.5156,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952783030320386e-05,
+      "loss": 0.8086,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527351474135076e-05,
+      "loss": 1.4141,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952687240827469e-05,
+      "loss": 1.5781,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9526393105634608e-05,
+      "loss": 1.0781,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525913566226747e-05,
+      "loss": 0.7344,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525433790063017e-05,
+      "loss": 0.8672,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952495377715535e-05,
+      "loss": 1.3047,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952447352751567e-05,
+      "loss": 1.2812,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9523993041155917e-05,
+      "loss": 1.4297,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952351231808803e-05,
+      "loss": 1.1719,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952303135832396e-05,
+      "loss": 1.5625,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9522550161875662e-05,
+      "loss": 0.9062,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952206872875509e-05,
+      "loss": 0.9609,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521587058974216e-05,
+      "loss": 1.2578,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521105152545005e-05,
+      "loss": 1.0703,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520623009479437e-05,
+      "loss": 1.4375,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520140629789494e-05,
+      "loss": 0.8047,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9519658013487167e-05,
+      "loss": 0.9023,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9519175160584453e-05,
+      "loss": 0.8047,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951869207109335e-05,
+      "loss": 1.0156,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951820874502586e-05,
+      "loss": 1.3516,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9517725182394006e-05,
+      "loss": 1.1406,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9517241383209796e-05,
+      "loss": 1.8984,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9516757347485263e-05,
+      "loss": 1.2031,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9516273075232427e-05,
+      "loss": 0.5156,
+      "step": 1981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515788566463337e-05,
+      "loss": 1.5078,
+      "step": 1982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515303821190024e-05,
+      "loss": 1.0469,
+      "step": 1983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951481883942454e-05,
+      "loss": 1.1953,
+      "step": 1984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9514333621178938e-05,
+      "loss": 0.9023,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951384816646528e-05,
+      "loss": 0.7852,
+      "step": 1986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9513362475295626e-05,
+      "loss": 0.5195,
+      "step": 1987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512876547682053e-05,
+      "loss": 0.832,
+      "step": 1988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512390383636633e-05,
+      "loss": 1.3594,
+      "step": 1989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9511903983171457e-05,
+      "loss": 1.1016,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9511417346298603e-05,
+      "loss": 0.957,
+      "step": 1991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951093047303018e-05,
+      "loss": 1.1094,
+      "step": 1992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9510443363378272e-05,
+      "loss": 1.4375,
+      "step": 1993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509956017354998e-05,
+      "loss": 1.2891,
+      "step": 1994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509468434972465e-05,
+      "loss": 1.2109,
+      "step": 1995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508980616242793e-05,
+      "loss": 0.9375,
+      "step": 1996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508492561178104e-05,
+      "loss": 1.4922,
+      "step": 1997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508004269790535e-05,
+      "loss": 0.9922,
+      "step": 1998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9507515742092213e-05,
+      "loss": 0.7812,
+      "step": 1999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9507026978095286e-05,
+      "loss": 0.8203,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95065379778119e-05,
+      "loss": 0.7734,
+      "step": 2001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9506048741254203e-05,
+      "loss": 1.0859,
+      "step": 2002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950555926843436e-05,
+      "loss": 0.9805,
+      "step": 2003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950506955936454e-05,
+      "loss": 1.2969,
+      "step": 2004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9504579614056907e-05,
+      "loss": 1.6719,
+      "step": 2005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950408943252364e-05,
+      "loss": 0.7734,
+      "step": 2006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9503599014776925e-05,
+      "loss": 0.4336,
+      "step": 2007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9503108360828946e-05,
+      "loss": 1.0703,
+      "step": 2008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95026174706919e-05,
+      "loss": 0.3691,
+      "step": 2009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950212634437799e-05,
+      "loss": 1.0469,
+      "step": 2010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9501634981899423e-05,
+      "loss": 0.3223,
+      "step": 2011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9501143383268405e-05,
+      "loss": 0.7227,
+      "step": 2012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9500651548497154e-05,
+      "loss": 0.5234,
+      "step": 2013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9500159477597906e-05,
+      "loss": 1.6875,
+      "step": 2014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9499667170582873e-05,
+      "loss": 1.1328,
+      "step": 2015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9499174627464304e-05,
+      "loss": 0.457,
+      "step": 2016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9498681848254438e-05,
+      "loss": 0.8477,
+      "step": 2017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9498188832965522e-05,
+      "loss": 1.2109,
+      "step": 2018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9497695581609804e-05,
+      "loss": 0.8008,
+      "step": 2019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949720209419955e-05,
+      "loss": 1.3594,
+      "step": 2020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9496708370747023e-05,
+      "loss": 0.8242,
+      "step": 2021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9496214411264494e-05,
+      "loss": 0.5156,
+      "step": 2022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949572021576424e-05,
+      "loss": 0.9922,
+      "step": 2023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9495225784258544e-05,
+      "loss": 0.4395,
+      "step": 2024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949473111675969e-05,
+      "loss": 0.8984,
+      "step": 2025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949423621327998e-05,
+      "loss": 0.8711,
+      "step": 2026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9493741073831706e-05,
+      "loss": 1.2656,
+      "step": 2027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9493245698427184e-05,
+      "loss": 1.1484,
+      "step": 2028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9492750087078715e-05,
+      "loss": 1.1953,
+      "step": 2029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9492254239798626e-05,
+      "loss": 0.5391,
+      "step": 2030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9491758156599235e-05,
+      "loss": 1.0,
+      "step": 2031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9491261837492877e-05,
+      "loss": 1.1016,
+      "step": 2032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9490765282491882e-05,
+      "loss": 1.2812,
+      "step": 2033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9490268491608596e-05,
+      "loss": 1.8047,
+      "step": 2034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948977146485536e-05,
+      "loss": 1.5,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948927420224453e-05,
+      "loss": 0.9609,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948877670378847e-05,
+      "loss": 1.1641,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948827896949954e-05,
+      "loss": 0.5391,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948778099939011e-05,
+      "loss": 0.4668,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9487282793472557e-05,
+      "loss": 1.1953,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9486784351759262e-05,
+      "loss": 1.3203,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9486285674262616e-05,
+      "loss": 1.5703,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485786760995014e-05,
+      "loss": 1.4219,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485287611968852e-05,
+      "loss": 0.7227,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9484788227196534e-05,
+      "loss": 1.8984,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9484288606690485e-05,
+      "loss": 1.2734,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483788750463106e-05,
+      "loss": 0.793,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948328865852683e-05,
+      "loss": 1.0781,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9482788330894085e-05,
+      "loss": 1.2734,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9482287767577302e-05,
+      "loss": 0.6914,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9481786968588926e-05,
+      "loss": 1.9453,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94812859339414e-05,
+      "loss": 1.5547,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9480784663647178e-05,
+      "loss": 1.8281,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9480283157718725e-05,
+      "loss": 0.8164,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9479781416168494e-05,
+      "loss": 1.1875,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9479279439008968e-05,
+      "loss": 0.4883,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947877722625261e-05,
+      "loss": 1.8438,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9478274777911913e-05,
+      "loss": 0.5625,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9477772093999358e-05,
+      "loss": 1.7266,
+      "step": 2059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9477269174527438e-05,
+      "loss": 0.6406,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9476766019508656e-05,
+      "loss": 0.6016,
+      "step": 2061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9476262628955517e-05,
+      "loss": 0.7812,
+      "step": 2062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9475759002880534e-05,
+      "loss": 1.3828,
+      "step": 2063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947525514129622e-05,
+      "loss": 1.1094,
+      "step": 2064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94747510442151e-05,
+      "loss": 0.7891,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9474246711649703e-05,
+      "loss": 1.0,
+      "step": 2066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947374214361256e-05,
+      "loss": 0.8281,
+      "step": 2067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9473237340116217e-05,
+      "loss": 0.9297,
+      "step": 2068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9472732301173215e-05,
+      "loss": 1.0625,
+      "step": 2069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947222702679611e-05,
+      "loss": 0.6094,
+      "step": 2070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9471721516997465e-05,
+      "loss": 0.8281,
+      "step": 2071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947121577178983e-05,
+      "loss": 1.2578,
+      "step": 2072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9470709791185786e-05,
+      "loss": 1.1797,
+      "step": 2073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9470203575197906e-05,
+      "loss": 1.3828,
+      "step": 2074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9469697123838768e-05,
+      "loss": 0.4902,
+      "step": 2075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946919043712096e-05,
+      "loss": 1.5,
+      "step": 2076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946868351505708e-05,
+      "loss": 1.6172,
+      "step": 2077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946817635765973e-05,
+      "loss": 1.1328,
+      "step": 2078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9467668964941498e-05,
+      "loss": 1.1875,
+      "step": 2079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9467161336915014e-05,
+      "loss": 1.75,
+      "step": 2080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946665347359288e-05,
+      "loss": 0.6406,
+      "step": 2081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9466145374987726e-05,
+      "loss": 1.1719,
+      "step": 2082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946563704111218e-05,
+      "loss": 0.8398,
+      "step": 2083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9465128471978875e-05,
+      "loss": 0.8008,
+      "step": 2084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9464619667600448e-05,
+      "loss": 1.25,
+      "step": 2085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946411062798955e-05,
+      "loss": 1.2109,
+      "step": 2086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946360135315883e-05,
+      "loss": 0.4688,
+      "step": 2087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9463091843120945e-05,
+      "loss": 0.8711,
+      "step": 2088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9462582097888555e-05,
+      "loss": 1.4141,
+      "step": 2089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9462072117474337e-05,
+      "loss": 0.9805,
+      "step": 2090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9461561901890958e-05,
+      "loss": 1.0703,
+      "step": 2091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9461051451151108e-05,
+      "loss": 1.3672,
+      "step": 2092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9460540765267465e-05,
+      "loss": 1.3906,
+      "step": 2093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9460029844252727e-05,
+      "loss": 0.9727,
+      "step": 2094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945951868811959e-05,
+      "loss": 0.9805,
+      "step": 2095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459007296880756e-05,
+      "loss": 1.2578,
+      "step": 2096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9458495670548937e-05,
+      "loss": 1.4609,
+      "step": 2097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945798380913685e-05,
+      "loss": 1.2188,
+      "step": 2098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945747171265722e-05,
+      "loss": 1.2266,
+      "step": 2099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9456959381122767e-05,
+      "loss": 1.0156,
+      "step": 2100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9456446814546227e-05,
+      "loss": 1.7266,
+      "step": 2101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9455934012940343e-05,
+      "loss": 1.1562,
+      "step": 2102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945542097631786e-05,
+      "loss": 0.5234,
+      "step": 2103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9454907704691522e-05,
+      "loss": 0.5977,
+      "step": 2104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9454394198074092e-05,
+      "loss": 0.5273,
+      "step": 2105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9453880456478333e-05,
+      "loss": 1.3672,
+      "step": 2106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9453366479917006e-05,
+      "loss": 0.6016,
+      "step": 2107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9452852268402894e-05,
+      "loss": 0.707,
+      "step": 2108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9452337821948776e-05,
+      "loss": 2.0156,
+      "step": 2109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945182314056743e-05,
+      "loss": 0.9141,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9451308224271658e-05,
+      "loss": 0.4316,
+      "step": 2111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9450793073074252e-05,
+      "loss": 1.3516,
+      "step": 2112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9450277686988016e-05,
+      "loss": 0.5898,
+      "step": 2113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944976206602576e-05,
+      "loss": 1.4688,
+      "step": 2114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9449246210200302e-05,
+      "loss": 0.6055,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9448730119524458e-05,
+      "loss": 0.6719,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9448213794011057e-05,
+      "loss": 0.5234,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9447697233672933e-05,
+      "loss": 0.8516,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9447180438522924e-05,
+      "loss": 1.6172,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944666340857387e-05,
+      "loss": 0.9961,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944614614383863e-05,
+      "loss": 1.0703,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9445628644330054e-05,
+      "loss": 0.8633,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9445110910061003e-05,
+      "loss": 1.4219,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9444592941044346e-05,
+      "loss": 1.125,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944407473729296e-05,
+      "loss": 1.2422,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944355629881972e-05,
+      "loss": 1.4141,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944303762563751e-05,
+      "loss": 1.7188,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9442518717759228e-05,
+      "loss": 0.5859,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441999575197762e-05,
+      "loss": 1.2891,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441480197966022e-05,
+      "loss": 0.9844,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9440960586076916e-05,
+      "loss": 1.1719,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9440440739543352e-05,
+      "loss": 1.3984,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9439920658378257e-05,
+      "loss": 0.9688,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9439400342594553e-05,
+      "loss": 1.1875,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438879792205176e-05,
+      "loss": 0.957,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438359007223058e-05,
+      "loss": 0.7148,
+      "step": 2136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9437837987661145e-05,
+      "loss": 1.0938,
+      "step": 2137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943731673353239e-05,
+      "loss": 1.1016,
+      "step": 2138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9436795244849744e-05,
+      "loss": 0.4316,
+      "step": 2139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9436273521626167e-05,
+      "loss": 1.1172,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943575156387463e-05,
+      "loss": 1.1094,
+      "step": 2141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9435229371608105e-05,
+      "loss": 1.0312,
+      "step": 2142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9434706944839566e-05,
+      "loss": 1.4844,
+      "step": 2143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9434184283582e-05,
+      "loss": 0.9844,
+      "step": 2144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94336613878484e-05,
+      "loss": 1.1328,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943313825765176e-05,
+      "loss": 1.2812,
+      "step": 2146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943261489300508e-05,
+      "loss": 1.1484,
+      "step": 2147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9432091293921366e-05,
+      "loss": 0.75,
+      "step": 2148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943156746041364e-05,
+      "loss": 1.0156,
+      "step": 2149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9431043392494914e-05,
+      "loss": 0.8477,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9430519090178212e-05,
+      "loss": 1.5156,
+      "step": 2151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942999455347657e-05,
+      "loss": 0.8711,
+      "step": 2152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942946978240302e-05,
+      "loss": 1.0156,
+      "step": 2153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9428944776970614e-05,
+      "loss": 0.6172,
+      "step": 2154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942841953719239e-05,
+      "loss": 0.9258,
+      "step": 2155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9427894063081404e-05,
+      "loss": 1.6094,
+      "step": 2156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942736835465072e-05,
+      "loss": 0.9961,
+      "step": 2157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94268424119134e-05,
+      "loss": 1.5312,
+      "step": 2158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942631623488252e-05,
+      "loss": 1.3906,
+      "step": 2159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942578982357115e-05,
+      "loss": 1.2656,
+      "step": 2160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942526317799238e-05,
+      "loss": 0.9688,
+      "step": 2161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9424736298159303e-05,
+      "loss": 1.125,
+      "step": 2162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9424209184085004e-05,
+      "loss": 0.7266,
+      "step": 2163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942368183578259e-05,
+      "loss": 0.5352,
+      "step": 2164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9423154253265167e-05,
+      "loss": 0.9453,
+      "step": 2165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9422626436545843e-05,
+      "loss": 1.1562,
+      "step": 2166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9422098385637742e-05,
+      "loss": 0.9883,
+      "step": 2167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9421570100553987e-05,
+      "loss": 0.8711,
+      "step": 2168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9421041581307705e-05,
+      "loss": 1.2188,
+      "step": 2169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9420512827912033e-05,
+      "loss": 1.5625,
+      "step": 2170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9419983840380117e-05,
+      "loss": 1.3828,
+      "step": 2171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9419454618725096e-05,
+      "loss": 1.4609,
+      "step": 2172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9418925162960133e-05,
+      "loss": 1.5391,
+      "step": 2173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941839547309838e-05,
+      "loss": 1.1719,
+      "step": 2174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9417865549153e-05,
+      "loss": 1.3516,
+      "step": 2175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9417335391137172e-05,
+      "loss": 0.4668,
+      "step": 2176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9416804999064066e-05,
+      "loss": 1.3984,
+      "step": 2177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9416274372946863e-05,
+      "loss": 0.9766,
+      "step": 2178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9415743512798756e-05,
+      "loss": 0.8008,
+      "step": 2179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9415212418632942e-05,
+      "loss": 1.4297,
+      "step": 2180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941468109046261e-05,
+      "loss": 1.6094,
+      "step": 2181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9414149528300976e-05,
+      "loss": 1.0938,
+      "step": 2182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9413617732161242e-05,
+      "loss": 1.4219,
+      "step": 2183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9413085702056636e-05,
+      "loss": 0.8516,
+      "step": 2184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941255343800037e-05,
+      "loss": 0.5234,
+      "step": 2185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941202094000568e-05,
+      "loss": 1.2891,
+      "step": 2186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94114882080858e-05,
+      "loss": 1.0234,
+      "step": 2187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9410955242253965e-05,
+      "loss": 1.0078,
+      "step": 2188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9410422042523426e-05,
+      "loss": 0.6055,
+      "step": 2189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9409888608907438e-05,
+      "loss": 1.0,
+      "step": 2190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9409354941419253e-05,
+      "loss": 0.7188,
+      "step": 2191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408821040072136e-05,
+      "loss": 1.5938,
+      "step": 2192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9408286904879362e-05,
+      "loss": 0.8672,
+      "step": 2193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9407752535854194e-05,
+      "loss": 1.1641,
+      "step": 2194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9407217933009925e-05,
+      "loss": 1.3516,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.940668309635984e-05,
+      "loss": 0.793,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9406148025917228e-05,
+      "loss": 1.125,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9405612721695387e-05,
+      "loss": 0.8203,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9405077183707628e-05,
+      "loss": 1.2422,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9404541411967253e-05,
+      "loss": 0.8906,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9404005406487584e-05,
+      "loss": 1.0391,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.940346916728194e-05,
+      "loss": 1.5391,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.940293269436365e-05,
+      "loss": 1.125,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9402395987746044e-05,
+      "loss": 1.1875,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9401859047442464e-05,
+      "loss": 0.498,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9401321873466258e-05,
+      "loss": 0.6523,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9400784465830772e-05,
+      "loss": 0.8672,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9400246824549366e-05,
+      "loss": 0.4453,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9399708949635395e-05,
+      "loss": 0.7031,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939917084110224e-05,
+      "loss": 0.5234,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9398632498963265e-05,
+      "loss": 0.7578,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9398093923231856e-05,
+      "loss": 0.7773,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9397555113921394e-05,
+      "loss": 0.5312,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9397016071045273e-05,
+      "loss": 1.2891,
+      "step": 2214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939647679461689e-05,
+      "loss": 1.2344,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9395937284649643e-05,
+      "loss": 1.125,
+      "step": 2216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9395397541156952e-05,
+      "loss": 0.8711,
+      "step": 2217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9394857564152222e-05,
+      "loss": 0.9609,
+      "step": 2218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9394317353648878e-05,
+      "loss": 0.8398,
+      "step": 2219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9393776909660346e-05,
+      "loss": 1.1172,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9393236232200053e-05,
+      "loss": 0.21,
+      "step": 2221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9392695321281448e-05,
+      "loss": 1.1172,
+      "step": 2222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9392154176917964e-05,
+      "loss": 1.1484,
+      "step": 2223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391612799123055e-05,
+      "loss": 2.125,
+      "step": 2224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391071187910175e-05,
+      "loss": 0.9844,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9390529343292787e-05,
+      "loss": 0.5742,
+      "step": 2226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938998726528436e-05,
+      "loss": 1.0547,
+      "step": 2227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9389444953898362e-05,
+      "loss": 1.0938,
+      "step": 2228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9388902409148273e-05,
+      "loss": 1.25,
+      "step": 2229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9388359631047576e-05,
+      "loss": 1.3281,
+      "step": 2230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9387816619609762e-05,
+      "loss": 1.3125,
+      "step": 2231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938727337484833e-05,
+      "loss": 0.8828,
+      "step": 2232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9386729896776778e-05,
+      "loss": 0.5508,
+      "step": 2233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9386186185408616e-05,
+      "loss": 1.5859,
+      "step": 2234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938564224075736e-05,
+      "loss": 1.7734,
+      "step": 2235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938509806283652e-05,
+      "loss": 0.7578,
+      "step": 2236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9384553651659626e-05,
+      "loss": 0.8438,
+      "step": 2237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938400900724021e-05,
+      "loss": 1.1875,
+      "step": 2238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938346412959181e-05,
+      "loss": 1.0,
+      "step": 2239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9382919018727965e-05,
+      "loss": 0.3379,
+      "step": 2240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9382373674662226e-05,
+      "loss": 1.2578,
+      "step": 2241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938182809740814e-05,
+      "loss": 1.2344,
+      "step": 2242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9381282286979273e-05,
+      "loss": 2.2656,
+      "step": 2243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9380736243389192e-05,
+      "loss": 1.3125,
+      "step": 2244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9380189966651463e-05,
+      "loss": 1.5156,
+      "step": 2245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9379643456779665e-05,
+      "loss": 0.9258,
+      "step": 2246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9379096713787385e-05,
+      "loss": 1.0703,
+      "step": 2247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378549737688205e-05,
+      "loss": 1.0547,
+      "step": 2248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378002528495723e-05,
+      "loss": 1.0625,
+      "step": 2249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937745508622354e-05,
+      "loss": 0.7969,
+      "step": 2250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937690741088526e-05,
+      "loss": 1.0156,
+      "step": 2251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9376359502494497e-05,
+      "loss": 1.8438,
+      "step": 2252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937581136106487e-05,
+      "loss": 1.3281,
+      "step": 2253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9375262986609994e-05,
+      "loss": 0.4824,
+      "step": 2254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9374714379143505e-05,
+      "loss": 1.2734,
+      "step": 2255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937416553867904e-05,
+      "loss": 0.9102,
+      "step": 2256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9373616465230235e-05,
+      "loss": 1.3438,
+      "step": 2257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937306715881074e-05,
+      "loss": 0.6211,
+      "step": 2258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9372517619434206e-05,
+      "loss": 0.9219,
+      "step": 2259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937196784711429e-05,
+      "loss": 2.1875,
+      "step": 2260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9371417841864656e-05,
+      "loss": 1.4453,
+      "step": 2261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9370867603698976e-05,
+      "loss": 1.1875,
+      "step": 2262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9370317132630926e-05,
+      "loss": 0.4863,
+      "step": 2263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9369766428674185e-05,
+      "loss": 0.7891,
+      "step": 2264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936921549184244e-05,
+      "loss": 0.8438,
+      "step": 2265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9368664322149386e-05,
+      "loss": 1.7578,
+      "step": 2266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936811291960872e-05,
+      "loss": 0.9531,
+      "step": 2267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9367561284234147e-05,
+      "loss": 0.8594,
+      "step": 2268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9367009416039376e-05,
+      "loss": 1.7266,
+      "step": 2269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9366457315038124e-05,
+      "loss": 1.0156,
+      "step": 2270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9365904981244116e-05,
+      "loss": 1.2344,
+      "step": 2271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9365352414671076e-05,
+      "loss": 1.4219,
+      "step": 2272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9364799615332738e-05,
+      "loss": 1.0703,
+      "step": 2273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9364246583242838e-05,
+      "loss": 1.0156,
+      "step": 2274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9363693318415128e-05,
+      "loss": 0.8047,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9363139820863354e-05,
+      "loss": 0.9141,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9362586090601275e-05,
+      "loss": 1.7812,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.936203212764265e-05,
+      "loss": 2.0625,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9361477932001253e-05,
+      "loss": 1.6484,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.936092350369085e-05,
+      "loss": 1.1328,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9360368842725227e-05,
+      "loss": 1.1641,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9359813949118167e-05,
+      "loss": 0.5586,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9359258822883463e-05,
+      "loss": 0.9258,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9358703464034908e-05,
+      "loss": 1.4219,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9358147872586307e-05,
+      "loss": 0.7617,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9357592048551472e-05,
+      "loss": 0.9688,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.935703599194421e-05,
+      "loss": 0.6445,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.935647970277835e-05,
+      "loss": 0.7734,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9355923181067714e-05,
+      "loss": 0.4023,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.935536642682613e-05,
+      "loss": 0.4648,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9354809440067437e-05,
+      "loss": 0.7852,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9354252220805485e-05,
+      "loss": 0.8359,
+      "step": 2292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9353694769054116e-05,
+      "loss": 1.5312,
+      "step": 2293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9353137084827186e-05,
+      "loss": 0.9609,
+      "step": 2294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.935257916813856e-05,
+      "loss": 1.3281,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9352021019002098e-05,
+      "loss": 0.8555,
+      "step": 2296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9351462637431672e-05,
+      "loss": 0.9219,
+      "step": 2297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9350904023441165e-05,
+      "loss": 0.9453,
+      "step": 2298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.935034517704446e-05,
+      "loss": 0.6953,
+      "step": 2299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9349786098255446e-05,
+      "loss": 0.8047,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9349226787088018e-05,
+      "loss": 0.7422,
+      "step": 2301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9348667243556077e-05,
+      "loss": 1.3672,
+      "step": 2302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9348107467673525e-05,
+      "loss": 0.4316,
+      "step": 2303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9347547459454282e-05,
+      "loss": 1.2891,
+      "step": 2304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9346987218912264e-05,
+      "loss": 0.5039,
+      "step": 2305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934642674606139e-05,
+      "loss": 0.957,
+      "step": 2306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9345866040915602e-05,
+      "loss": 1.2188,
+      "step": 2307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934530510348882e-05,
+      "loss": 0.6641,
+      "step": 2308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9344743933795002e-05,
+      "loss": 1.0078,
+      "step": 2309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9344182531848082e-05,
+      "loss": 0.6602,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9343620897662018e-05,
+      "loss": 1.0469,
+      "step": 2311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9343059031250768e-05,
+      "loss": 0.8906,
+      "step": 2312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93424969326283e-05,
+      "loss": 1.0938,
+      "step": 2313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934193460180858e-05,
+      "loss": 1.0547,
+      "step": 2314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9341372038805582e-05,
+      "loss": 1.0938,
+      "step": 2315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9340809243633297e-05,
+      "loss": 1.6484,
+      "step": 2316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9340246216305703e-05,
+      "loss": 1.875,
+      "step": 2317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93396829568368e-05,
+      "loss": 1.3906,
+      "step": 2318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9339119465240586e-05,
+      "loss": 0.7695,
+      "step": 2319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9338555741531062e-05,
+      "loss": 1.4297,
+      "step": 2320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933799178572224e-05,
+      "loss": 1.0547,
+      "step": 2321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9337427597828142e-05,
+      "loss": 1.5156,
+      "step": 2322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9336863177862787e-05,
+      "loss": 0.9141,
+      "step": 2323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93362985258402e-05,
+      "loss": 2.4844,
+      "step": 2324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9335733641774413e-05,
+      "loss": 1.0312,
+      "step": 2325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9335168525679472e-05,
+      "loss": 0.8633,
+      "step": 2326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9334603177569423e-05,
+      "loss": 1.0469,
+      "step": 2327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933403759745831e-05,
+      "loss": 0.9453,
+      "step": 2328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9333471785360195e-05,
+      "loss": 0.9453,
+      "step": 2329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933290574128914e-05,
+      "loss": 1.1875,
+      "step": 2330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9332339465259214e-05,
+      "loss": 1.4609,
+      "step": 2331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331772957284487e-05,
+      "loss": 1.9062,
+      "step": 2332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331206217379044e-05,
+      "loss": 0.6836,
+      "step": 2333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9330639245556967e-05,
+      "loss": 1.2188,
+      "step": 2334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933007204183235e-05,
+      "loss": 0.6875,
+      "step": 2335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932950460621929e-05,
+      "loss": 0.9805,
+      "step": 2336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932893693873189e-05,
+      "loss": 0.6523,
+      "step": 2337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9328369039384255e-05,
+      "loss": 0.75,
+      "step": 2338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9327800908190503e-05,
+      "loss": 0.707,
+      "step": 2339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9327232545164755e-05,
+      "loss": 0.8828,
+      "step": 2340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9326663950321135e-05,
+      "loss": 1.5938,
+      "step": 2341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9326095123673777e-05,
+      "loss": 1.5156,
+      "step": 2342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9325526065236815e-05,
+      "loss": 0.9258,
+      "step": 2343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9324956775024395e-05,
+      "loss": 0.7227,
+      "step": 2344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932438725305066e-05,
+      "loss": 1.3359,
+      "step": 2345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932381749932978e-05,
+      "loss": 1.5391,
+      "step": 2346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93232475138759e-05,
+      "loss": 0.7656,
+      "step": 2347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9322677296703194e-05,
+      "loss": 1.7812,
+      "step": 2348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9322106847825833e-05,
+      "loss": 0.6406,
+      "step": 2349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9321536167257992e-05,
+      "loss": 0.9531,
+      "step": 2350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9320965255013856e-05,
+      "loss": 1.8047,
+      "step": 2351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9320394111107618e-05,
+      "loss": 1.2734,
+      "step": 2352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9319822735553472e-05,
+      "loss": 1.1406,
+      "step": 2353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9319251128365613e-05,
+      "loss": 1.0234,
+      "step": 2354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9318679289558255e-05,
+      "loss": 1.4766,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9318107219145604e-05,
+      "loss": 1.7578,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9317534917141884e-05,
+      "loss": 0.832,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.931696238356131e-05,
+      "loss": 0.7734,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9316389618418125e-05,
+      "loss": 0.7305,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9315816621726557e-05,
+      "loss": 0.9023,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9315243393500847e-05,
+      "loss": 1.1953,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9314669933755238e-05,
+      "loss": 0.9883,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9314096242503992e-05,
+      "loss": 1.0469,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9313522319761362e-05,
+      "loss": 1.4766,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9312948165541613e-05,
+      "loss": 0.7461,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9312373779859014e-05,
+      "loss": 1.3203,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.931179916272784e-05,
+      "loss": 0.9453,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9311224314162377e-05,
+      "loss": 1.3359,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.931064923417691e-05,
+      "loss": 0.7773,
+      "step": 2369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.931007392278573e-05,
+      "loss": 1.1094,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9309498380003137e-05,
+      "loss": 0.8672,
+      "step": 2371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9308922605843434e-05,
+      "loss": 1.3125,
+      "step": 2372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9308346600320936e-05,
+      "loss": 0.8398,
+      "step": 2373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9307770363449952e-05,
+      "loss": 1.3125,
+      "step": 2374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9307193895244807e-05,
+      "loss": 1.3125,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9306617195719833e-05,
+      "loss": 0.7969,
+      "step": 2376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9306040264889355e-05,
+      "loss": 1.3203,
+      "step": 2377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9305463102767718e-05,
+      "loss": 1.3438,
+      "step": 2378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9304885709369264e-05,
+      "loss": 0.9688,
+      "step": 2379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9304308084708342e-05,
+      "loss": 0.4004,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9303730228799315e-05,
+      "loss": 1.0391,
+      "step": 2381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9303152141656537e-05,
+      "loss": 1.2891,
+      "step": 2382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.930257382329438e-05,
+      "loss": 0.5117,
+      "step": 2383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9301995273727217e-05,
+      "loss": 1.0781,
+      "step": 2384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9301416492969426e-05,
+      "loss": 0.5625,
+      "step": 2385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9300837481035392e-05,
+      "loss": 0.7539,
+      "step": 2386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9300258237939507e-05,
+      "loss": 0.8906,
+      "step": 2387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9299678763696168e-05,
+      "loss": 0.7656,
+      "step": 2388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9299099058319774e-05,
+      "loss": 1.0703,
+      "step": 2389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9298519121824734e-05,
+      "loss": 0.7812,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9297938954225466e-05,
+      "loss": 0.5703,
+      "step": 2391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9297358555536382e-05,
+      "loss": 0.9141,
+      "step": 2392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9296777925771915e-05,
+      "loss": 0.5312,
+      "step": 2393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.929619706494649e-05,
+      "loss": 1.1797,
+      "step": 2394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9295615973074546e-05,
+      "loss": 1.5,
+      "step": 2395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9295034650170524e-05,
+      "loss": 0.9961,
+      "step": 2396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9294453096248877e-05,
+      "loss": 0.8789,
+      "step": 2397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9293871311324052e-05,
+      "loss": 1.1484,
+      "step": 2398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.929328929541051e-05,
+      "loss": 1.2734,
+      "step": 2399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.929270704852272e-05,
+      "loss": 0.8008,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9292124570675153e-05,
+      "loss": 1.5547,
+      "step": 2401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.929154186188228e-05,
+      "loss": 1.0234,
+      "step": 2402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.929095892215859e-05,
+      "loss": 0.9492,
+      "step": 2403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9290375751518567e-05,
+      "loss": 0.9805,
+      "step": 2404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9289792349976708e-05,
+      "loss": 1.1875,
+      "step": 2405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928920871754751e-05,
+      "loss": 0.5078,
+      "step": 2406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9288624854245483e-05,
+      "loss": 1.2031,
+      "step": 2407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9288040760085132e-05,
+      "loss": 0.5234,
+      "step": 2408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928745643508098e-05,
+      "loss": 0.9961,
+      "step": 2409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9286871879247545e-05,
+      "loss": 2.0938,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928628709259936e-05,
+      "loss": 1.3047,
+      "step": 2411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9285702075150955e-05,
+      "loss": 1.2031,
+      "step": 2412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9285116826916877e-05,
+      "loss": 0.7656,
+      "step": 2413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928453134791166e-05,
+      "loss": 1.2188,
+      "step": 2414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9283945638149862e-05,
+      "loss": 0.6875,
+      "step": 2415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9283359697646046e-05,
+      "loss": 0.7305,
+      "step": 2416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9282773526414765e-05,
+      "loss": 1.1094,
+      "step": 2417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928218712447059e-05,
+      "loss": 1.1172,
+      "step": 2418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.92816004918281e-05,
+      "loss": 1.2031,
+      "step": 2419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9281013628501872e-05,
+      "loss": 0.5352,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9280426534506488e-05,
+      "loss": 1.6094,
+      "step": 2421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9279839209856548e-05,
+      "loss": 1.125,
+      "step": 2422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927925165456664e-05,
+      "loss": 1.5703,
+      "step": 2423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9278663868651377e-05,
+      "loss": 1.5312,
+      "step": 2424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927807585212536e-05,
+      "loss": 0.7812,
+      "step": 2425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927748760500321e-05,
+      "loss": 0.8789,
+      "step": 2426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9276899127299537e-05,
+      "loss": 0.9453,
+      "step": 2427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9276310419028976e-05,
+      "loss": 1.0859,
+      "step": 2428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9275721480206156e-05,
+      "loss": 0.9531,
+      "step": 2429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927513231084571e-05,
+      "loss": 0.6523,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927454291096229e-05,
+      "loss": 0.7461,
+      "step": 2431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.927395328057054e-05,
+      "loss": 1.0703,
+      "step": 2432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9273363419685113e-05,
+      "loss": 1.125,
+      "step": 2433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9272773328320676e-05,
+      "loss": 0.7734,
+      "step": 2434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9272183006491886e-05,
+      "loss": 1.2891,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.927159245421342e-05,
+      "loss": 1.1484,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9271001671499953e-05,
+      "loss": 0.875,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9270410658366172e-05,
+      "loss": 0.3281,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.926981941482676e-05,
+      "loss": 0.9688,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.926922794089642e-05,
+      "loss": 1.0156,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9268636236589847e-05,
+      "loss": 0.918,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9268044301921746e-05,
+      "loss": 0.7578,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9267452136906833e-05,
+      "loss": 1.0391,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9266859741559823e-05,
+      "loss": 1.1953,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.926626711589544e-05,
+      "loss": 0.9883,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.926567425992841e-05,
+      "loss": 0.9336,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9265081173673475e-05,
+      "loss": 0.8477,
+      "step": 2447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9264487857145368e-05,
+      "loss": 0.707,
+      "step": 2448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.926389431035884e-05,
+      "loss": 0.8477,
+      "step": 2449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9263300533328643e-05,
+      "loss": 1.0312,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9262706526069528e-05,
+      "loss": 1.5781,
+      "step": 2451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9262112288596264e-05,
+      "loss": 2.0156,
+      "step": 2452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9261517820923623e-05,
+      "loss": 1.2891,
+      "step": 2453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9260923123066372e-05,
+      "loss": 1.4453,
+      "step": 2454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.92603281950393e-05,
+      "loss": 1.0547,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925973303685719e-05,
+      "loss": 0.3711,
+      "step": 2456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925913764853483e-05,
+      "loss": 1.2344,
+      "step": 2457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925854203008702e-05,
+      "loss": 0.4707,
+      "step": 2458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9257946181528563e-05,
+      "loss": 0.4785,
+      "step": 2459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9257350102874272e-05,
+      "loss": 0.9688,
+      "step": 2460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925675379413896e-05,
+      "loss": 2.0938,
+      "step": 2461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9256157255337444e-05,
+      "loss": 0.7852,
+      "step": 2462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9255560486484554e-05,
+      "loss": 0.4453,
+      "step": 2463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9254963487595122e-05,
+      "loss": 0.8125,
+      "step": 2464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9254366258683984e-05,
+      "loss": 1.4609,
+      "step": 2465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925376879976599e-05,
+      "loss": 0.9062,
+      "step": 2466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9253171110855974e-05,
+      "loss": 0.918,
+      "step": 2467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9252573191968805e-05,
+      "loss": 0.8086,
+      "step": 2468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925197504311934e-05,
+      "loss": 1.7109,
+      "step": 2469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925137666432244e-05,
+      "loss": 1.1484,
+      "step": 2470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9250778055592983e-05,
+      "loss": 0.5664,
+      "step": 2471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.925017921694585e-05,
+      "loss": 2.3281,
+      "step": 2472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9249580148395915e-05,
+      "loss": 1.4688,
+      "step": 2473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.924898084995807e-05,
+      "loss": 1.125,
+      "step": 2474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9248381321647214e-05,
+      "loss": 0.3652,
+      "step": 2475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9247781563478247e-05,
+      "loss": 1.4062,
+      "step": 2476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.924718157546607e-05,
+      "loss": 0.5938,
+      "step": 2477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.92465813576256e-05,
+      "loss": 1.6484,
+      "step": 2478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9245980909971754e-05,
+      "loss": 1.5703,
+      "step": 2479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9245380232519454e-05,
+      "loss": 1.1172,
+      "step": 2480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.924477932528363e-05,
+      "loss": 0.8633,
+      "step": 2481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9244178188279213e-05,
+      "loss": 1.0391,
+      "step": 2482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.924357682152115e-05,
+      "loss": 0.3906,
+      "step": 2483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9242975225024386e-05,
+      "loss": 0.7031,
+      "step": 2484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.924237339880387e-05,
+      "loss": 1.3281,
+      "step": 2485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9241771342874563e-05,
+      "loss": 1.1328,
+      "step": 2486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9241169057251428e-05,
+      "loss": 0.7539,
+      "step": 2487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9240566541949427e-05,
+      "loss": 1.2188,
+      "step": 2488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9239963796983547e-05,
+      "loss": 1.6641,
+      "step": 2489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.923936082236876e-05,
+      "loss": 1.6875,
+      "step": 2490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9238757618120052e-05,
+      "loss": 1.1875,
+      "step": 2491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.923815418425242e-05,
+      "loss": 0.9492,
+      "step": 2492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9237550520780857e-05,
+      "loss": 0.9023,
+      "step": 2493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.923694662772037e-05,
+      "loss": 1.0469,
+      "step": 2494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.923634250508597e-05,
+      "loss": 1.3125,
+      "step": 2495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9235738152892666e-05,
+      "loss": 1.9219,
+      "step": 2496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9235133571155477e-05,
+      "loss": 1.1641,
+      "step": 2497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9234528759889435e-05,
+      "loss": 1.1094,
+      "step": 2498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.923392371910957e-05,
+      "loss": 0.8984,
+      "step": 2499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9233318448830924e-05,
+      "loss": 0.6523,
+      "step": 2500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9232712949068534e-05,
+      "loss": 0.7461,
+      "step": 2501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9232107219837448e-05,
+      "loss": 1.3125,
+      "step": 2502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9231501261152726e-05,
+      "loss": 0.5391,
+      "step": 2503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9230895073029424e-05,
+      "loss": 0.7656,
+      "step": 2504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.923028865548261e-05,
+      "loss": 1.0078,
+      "step": 2505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.922968200852736e-05,
+      "loss": 2.0625,
+      "step": 2506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9229075132178746e-05,
+      "loss": 0.8594,
+      "step": 2507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.922846802645185e-05,
+      "loss": 0.3535,
+      "step": 2508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9227860691361763e-05,
+      "loss": 0.6367,
+      "step": 2509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9227253126923587e-05,
+      "loss": 1.1484,
+      "step": 2510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.922664533315241e-05,
+      "loss": 0.3672,
+      "step": 2511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9226037310063342e-05,
+      "loss": 1.1953,
+      "step": 2512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9225429057671502e-05,
+      "loss": 0.8789,
+      "step": 2513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9224820575991997e-05,
+      "loss": 0.9219,
+      "step": 2514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9224211865039958e-05,
+      "loss": 1.2031,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.922360292483051e-05,
+      "loss": 1.5312,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9222993755378788e-05,
+      "loss": 2.0156,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9222384356699933e-05,
+      "loss": 0.7305,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.922177472880909e-05,
+      "loss": 2.2031,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9221164871721413e-05,
+      "loss": 0.5781,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9220554785452057e-05,
+      "loss": 1.1719,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9219944470016183e-05,
+      "loss": 0.5898,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9219333925428965e-05,
+      "loss": 0.6992,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9218723151705575e-05,
+      "loss": 1.3203,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9218112148861186e-05,
+      "loss": 0.9492,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9217500916910997e-05,
+      "loss": 0.5391,
+      "step": 2526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.921688945587019e-05,
+      "loss": 1.5391,
+      "step": 2527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9216277765753964e-05,
+      "loss": 1.0078,
+      "step": 2528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9215665846577525e-05,
+      "loss": 1.2188,
+      "step": 2529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.921505369835608e-05,
+      "loss": 1.3125,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9214441321104843e-05,
+      "loss": 1.7812,
+      "step": 2531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9213828714839035e-05,
+      "loss": 0.875,
+      "step": 2532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9213215879573878e-05,
+      "loss": 0.498,
+      "step": 2533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9212602815324603e-05,
+      "loss": 1.5,
+      "step": 2534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.921198952210645e-05,
+      "loss": 1.5781,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9211375999934665e-05,
+      "loss": 0.6562,
+      "step": 2536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9210762248824492e-05,
+      "loss": 0.4785,
+      "step": 2537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9210148268791187e-05,
+      "loss": 1.1562,
+      "step": 2538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9209534059850002e-05,
+      "loss": 0.3496,
+      "step": 2539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.920891962201622e-05,
+      "loss": 1.3359,
+      "step": 2540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9208304955305094e-05,
+      "loss": 1.9141,
+      "step": 2541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9207690059731908e-05,
+      "loss": 1.0234,
+      "step": 2542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9207074935311944e-05,
+      "loss": 0.9766,
+      "step": 2543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9206459582060493e-05,
+      "loss": 1.1406,
+      "step": 2544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.920584399999285e-05,
+      "loss": 1.1797,
+      "step": 2545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9205228189124307e-05,
+      "loss": 1.4375,
+      "step": 2546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9204612149470177e-05,
+      "loss": 0.8242,
+      "step": 2547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9203995881045766e-05,
+      "loss": 1.3203,
+      "step": 2548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.920337938386639e-05,
+      "loss": 0.8789,
+      "step": 2549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9202762657947375e-05,
+      "loss": 1.0156,
+      "step": 2550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.920214570330405e-05,
+      "loss": 1.0625,
+      "step": 2551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9201528519951743e-05,
+      "loss": 1.3672,
+      "step": 2552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.92009111079058e-05,
+      "loss": 1.4688,
+      "step": 2553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.920029346718156e-05,
+      "loss": 1.6094,
+      "step": 2554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9199675597794377e-05,
+      "loss": 1.5938,
+      "step": 2555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.919905749975961e-05,
+      "loss": 1.0859,
+      "step": 2556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.919843917309262e-05,
+      "loss": 1.1562,
+      "step": 2557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9197820617808768e-05,
+      "loss": 0.4766,
+      "step": 2558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.919720183392343e-05,
+      "loss": 0.4941,
+      "step": 2559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9196582821451996e-05,
+      "loss": 0.4492,
+      "step": 2560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9195963580409834e-05,
+      "loss": 1.2578,
+      "step": 2561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.919534411081235e-05,
+      "loss": 1.0703,
+      "step": 2562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9194724412674933e-05,
+      "loss": 1.5156,
+      "step": 2563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.919410448601298e-05,
+      "loss": 0.4805,
+      "step": 2564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.919348433084191e-05,
+      "loss": 1.1328,
+      "step": 2565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9192863947177123e-05,
+      "loss": 0.7734,
+      "step": 2566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9192243335034053e-05,
+      "loss": 1.4297,
+      "step": 2567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.919162249442811e-05,
+      "loss": 0.8125,
+      "step": 2568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9191001425374734e-05,
+      "loss": 0.5469,
+      "step": 2569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9190380127889358e-05,
+      "loss": 1.6875,
+      "step": 2570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9189758601987426e-05,
+      "loss": 0.957,
+      "step": 2571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.918913684768438e-05,
+      "loss": 2.2656,
+      "step": 2572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9188514864995676e-05,
+      "loss": 0.9492,
+      "step": 2573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9187892653936772e-05,
+      "loss": 1.1016,
+      "step": 2574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9187270214523135e-05,
+      "loss": 0.6719,
+      "step": 2575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.918664754677023e-05,
+      "loss": 1.1016,
+      "step": 2576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9186024650693537e-05,
+      "loss": 0.8984,
+      "step": 2577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9185401526308537e-05,
+      "loss": 0.9375,
+      "step": 2578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9184778173630715e-05,
+      "loss": 1.4062,
+      "step": 2579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9184154592675566e-05,
+      "loss": 1.3906,
+      "step": 2580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9183530783458587e-05,
+      "loss": 0.7578,
+      "step": 2581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.918290674599528e-05,
+      "loss": 1.1641,
+      "step": 2582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9182282480301158e-05,
+      "loss": 0.8242,
+      "step": 2583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9181657986391736e-05,
+      "loss": 0.9961,
+      "step": 2584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9181033264282535e-05,
+      "loss": 0.5312,
+      "step": 2585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.918040831398908e-05,
+      "loss": 0.7188,
+      "step": 2586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9179783135526907e-05,
+      "loss": 1.2422,
+      "step": 2587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.917915772891155e-05,
+      "loss": 0.9844,
+      "step": 2588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9178532094158558e-05,
+      "loss": 0.6211,
+      "step": 2589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9177906231283476e-05,
+      "loss": 0.8516,
+      "step": 2590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.917728014030186e-05,
+      "loss": 1.4844,
+      "step": 2591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9176653821229268e-05,
+      "loss": 1.0625,
+      "step": 2592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.917602727408127e-05,
+      "loss": 0.8047,
+      "step": 2593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.917540049887344e-05,
+      "loss": 0.8242,
+      "step": 2594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9174773495621354e-05,
+      "loss": 1.0703,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.917414626434059e-05,
+      "loss": 0.5195,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.917351880504675e-05,
+      "loss": 1.0312,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.917289111775541e-05,
+      "loss": 1.3203,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9172263202482187e-05,
+      "loss": 0.8867,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9171635059242684e-05,
+      "loss": 0.8555,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9171006688052506e-05,
+      "loss": 1.2734,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9170378088927277e-05,
+      "loss": 1.7734,
+      "step": 2602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9169749261882615e-05,
+      "loss": 1.3594,
+      "step": 2603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9169120206934154e-05,
+      "loss": 1.5078,
+      "step": 2604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9168490924097527e-05,
+      "loss": 0.4062,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9167861413388372e-05,
+      "loss": 0.957,
+      "step": 2606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9167231674822336e-05,
+      "loss": 1.2422,
+      "step": 2607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.916660170841507e-05,
+      "loss": 1.4531,
+      "step": 2608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.916597151418223e-05,
+      "loss": 0.5391,
+      "step": 2609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.916534109213948e-05,
+      "loss": 1.1406,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9164710442302492e-05,
+      "loss": 1.0391,
+      "step": 2611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9164079564686933e-05,
+      "loss": 1.0469,
+      "step": 2612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.916344845930849e-05,
+      "loss": 0.6016,
+      "step": 2613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9162817126182842e-05,
+      "loss": 1.4531,
+      "step": 2614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9162185565325683e-05,
+      "loss": 0.5664,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9161553776752714e-05,
+      "loss": 0.8477,
+      "step": 2616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.916092176047963e-05,
+      "loss": 0.4746,
+      "step": 2617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9160289516522138e-05,
+      "loss": 1.6016,
+      "step": 2618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.915965704489596e-05,
+      "loss": 0.707,
+      "step": 2619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9159024345616815e-05,
+      "loss": 1.0391,
+      "step": 2620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.915839141870042e-05,
+      "loss": 0.7734,
+      "step": 2621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.915775826416251e-05,
+      "loss": 1.1094,
+      "step": 2622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9157124882018822e-05,
+      "loss": 0.7734,
+      "step": 2623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9156491272285094e-05,
+      "loss": 0.7148,
+      "step": 2624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.915585743497708e-05,
+      "loss": 1.5078,
+      "step": 2625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.915522337011053e-05,
+      "loss": 0.9883,
+      "step": 2626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9154589077701203e-05,
+      "loss": 0.7617,
+      "step": 2627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9153954557764867e-05,
+      "loss": 1.625,
+      "step": 2628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9153319810317285e-05,
+      "loss": 1.1953,
+      "step": 2629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9152684835374237e-05,
+      "loss": 1.3203,
+      "step": 2630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9152049632951505e-05,
+      "loss": 0.5625,
+      "step": 2631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9151414203064876e-05,
+      "loss": 1.5391,
+      "step": 2632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9150778545730148e-05,
+      "loss": 1.3203,
+      "step": 2633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.915014266096311e-05,
+      "loss": 1.125,
+      "step": 2634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914950654877957e-05,
+      "loss": 0.6172,
+      "step": 2635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914887020919534e-05,
+      "loss": 1.3359,
+      "step": 2636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914823364222623e-05,
+      "loss": 1.2969,
+      "step": 2637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914759684788807e-05,
+      "loss": 0.7734,
+      "step": 2638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9146959826196678e-05,
+      "loss": 0.8672,
+      "step": 2639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914632257716789e-05,
+      "loss": 1.2266,
+      "step": 2640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914568510081755e-05,
+      "loss": 0.8867,
+      "step": 2641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914504739716149e-05,
+      "loss": 1.6562,
+      "step": 2642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914440946621557e-05,
+      "loss": 1.4766,
+      "step": 2643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914377130799564e-05,
+      "loss": 1.3203,
+      "step": 2644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9143132922517556e-05,
+      "loss": 0.3164,
+      "step": 2645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9142494309797197e-05,
+      "loss": 0.8086,
+      "step": 2646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9141855469850427e-05,
+      "loss": 1.1797,
+      "step": 2647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.914121640269312e-05,
+      "loss": 1.1094,
+      "step": 2648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9140577108341167e-05,
+      "loss": 0.9766,
+      "step": 2649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9139937586810447e-05,
+      "loss": 0.875,
+      "step": 2650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.913929783811687e-05,
+      "loss": 1.0,
+      "step": 2651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9138657862276324e-05,
+      "loss": 0.8516,
+      "step": 2652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9138017659304715e-05,
+      "loss": 0.6719,
+      "step": 2653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9137377229217962e-05,
+      "loss": 0.7344,
+      "step": 2654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.913673657203198e-05,
+      "loss": 1.1562,
+      "step": 2655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9136095687762683e-05,
+      "loss": 1.3125,
+      "step": 2656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.3281,
+      "step": 2657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9134813238037892e-05,
+      "loss": 0.4512,
+      "step": 2658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9134171672614265e-05,
+      "loss": 1.3594,
+      "step": 2659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9133529880171084e-05,
+      "loss": 1.7031,
+      "step": 2660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.913288786072429e-05,
+      "loss": 1.4453,
+      "step": 2661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9132245614289838e-05,
+      "loss": 0.6367,
+      "step": 2662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.91316031408837e-05,
+      "loss": 1.2266,
+      "step": 2663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9130960440521843e-05,
+      "loss": 0.8984,
+      "step": 2664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9130317513220233e-05,
+      "loss": 1.5938,
+      "step": 2665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9129674358994853e-05,
+      "loss": 0.9453,
+      "step": 2666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9129030977861687e-05,
+      "loss": 1.125,
+      "step": 2667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.912838736983673e-05,
+      "loss": 0.9297,
+      "step": 2668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9127743534935974e-05,
+      "loss": 2.3125,
+      "step": 2669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.912709947317542e-05,
+      "loss": 1.6719,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.912645518457108e-05,
+      "loss": 1.8438,
+      "step": 2671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.912581066913896e-05,
+      "loss": 1.1562,
+      "step": 2672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9125165926895085e-05,
+      "loss": 0.707,
+      "step": 2673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9124520957855476e-05,
+      "loss": 0.9961,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9123875762036166e-05,
+      "loss": 0.9258,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.912323033945319e-05,
+      "loss": 1.5078,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9122584690122582e-05,
+      "loss": 0.9531,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9121938814060402e-05,
+      "loss": 0.7422,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.912129271128269e-05,
+      "loss": 1.3672,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9120646381805512e-05,
+      "loss": 0.4062,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.911999982564493e-05,
+      "loss": 0.7266,
+      "step": 2681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.911935304281701e-05,
+      "loss": 1.3594,
+      "step": 2682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9118706033337835e-05,
+      "loss": 0.7109,
+      "step": 2683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9118058797223477e-05,
+      "loss": 0.4141,
+      "step": 2684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9117411334490022e-05,
+      "loss": 0.4141,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9116763645153573e-05,
+      "loss": 0.7109,
+      "step": 2686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9116115729230216e-05,
+      "loss": 1.1797,
+      "step": 2687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9115467586736063e-05,
+      "loss": 1.4219,
+      "step": 2688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9114819217687214e-05,
+      "loss": 1.0625,
+      "step": 2689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.911417062209979e-05,
+      "loss": 1.2109,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.911352179998991e-05,
+      "loss": 1.2578,
+      "step": 2691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9112872751373696e-05,
+      "loss": 0.9297,
+      "step": 2692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9112223476267283e-05,
+      "loss": 0.9844,
+      "step": 2693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.911157397468681e-05,
+      "loss": 1.1953,
+      "step": 2694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9110924246648414e-05,
+      "loss": 1.0938,
+      "step": 2695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9110274292168248e-05,
+      "loss": 1.2891,
+      "step": 2696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.910962411126246e-05,
+      "loss": 1.2344,
+      "step": 2697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.910897370394722e-05,
+      "loss": 1.2266,
+      "step": 2698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.910832307023868e-05,
+      "loss": 1.1797,
+      "step": 2699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.910767221015302e-05,
+      "loss": 1.6562,
+      "step": 2700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9107021123706413e-05,
+      "loss": 0.9062,
+      "step": 2701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9106369810915045e-05,
+      "loss": 1.25,
+      "step": 2702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.91057182717951e-05,
+      "loss": 1.0547,
+      "step": 2703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9105066506362772e-05,
+      "loss": 0.6953,
+      "step": 2704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9104414514634255e-05,
+      "loss": 1.1328,
+      "step": 2705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9103762296625762e-05,
+      "loss": 0.7031,
+      "step": 2706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9103109852353498e-05,
+      "loss": 0.8008,
+      "step": 2707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9102457181833682e-05,
+      "loss": 0.7812,
+      "step": 2708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.910180428508253e-05,
+      "loss": 1.5938,
+      "step": 2709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9101151162116277e-05,
+      "loss": 1.0938,
+      "step": 2710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.910049781295115e-05,
+      "loss": 0.6953,
+      "step": 2711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9099844237603384e-05,
+      "loss": 1.125,
+      "step": 2712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.909919043608923e-05,
+      "loss": 1.0,
+      "step": 2713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9098536408424934e-05,
+      "loss": 0.5508,
+      "step": 2714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9097882154626755e-05,
+      "loss": 1.3438,
+      "step": 2715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9097227674710946e-05,
+      "loss": 1.0234,
+      "step": 2716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9096572968693783e-05,
+      "loss": 0.4473,
+      "step": 2717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.909591803659153e-05,
+      "loss": 1.2422,
+      "step": 2718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9095262878420467e-05,
+      "loss": 1.6172,
+      "step": 2719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9094607494196878e-05,
+      "loss": 1.125,
+      "step": 2720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9093951883937054e-05,
+      "loss": 0.5234,
+      "step": 2721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9093296047657287e-05,
+      "loss": 0.9453,
+      "step": 2722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9092639985373872e-05,
+      "loss": 0.8945,
+      "step": 2723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9091983697103128e-05,
+      "loss": 0.7773,
+      "step": 2724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.909132718286135e-05,
+      "loss": 1.6875,
+      "step": 2725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9090670442664868e-05,
+      "loss": 1.0625,
+      "step": 2726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.909001347653e-05,
+      "loss": 1.1328,
+      "step": 2727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.908935628447307e-05,
+      "loss": 0.668,
+      "step": 2728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.908869886651042e-05,
+      "loss": 1.0938,
+      "step": 2729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.908804122265838e-05,
+      "loss": 0.3145,
+      "step": 2730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9087383352933303e-05,
+      "loss": 0.418,
+      "step": 2731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9086725257351533e-05,
+      "loss": 1.2109,
+      "step": 2732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9086066935929435e-05,
+      "loss": 0.625,
+      "step": 2733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9085408388683366e-05,
+      "loss": 1.6719,
+      "step": 2734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.908474961562969e-05,
+      "loss": 1.2969,
+      "step": 2735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9084090616784783e-05,
+      "loss": 1.4609,
+      "step": 2736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9083431392165025e-05,
+      "loss": 0.7188,
+      "step": 2737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9082771941786796e-05,
+      "loss": 1.4453,
+      "step": 2738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.908211226566649e-05,
+      "loss": 1.2266,
+      "step": 2739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9081452363820504e-05,
+      "loss": 1.2578,
+      "step": 2740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9080792236265235e-05,
+      "loss": 0.6602,
+      "step": 2741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.908013188301709e-05,
+      "loss": 1.1562,
+      "step": 2742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9079471304092485e-05,
+      "loss": 0.9844,
+      "step": 2743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9078810499507836e-05,
+      "loss": 0.5664,
+      "step": 2744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9078149469279567e-05,
+      "loss": 1.1875,
+      "step": 2745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9077488213424102e-05,
+      "loss": 1.1016,
+      "step": 2746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9076826731957885e-05,
+      "loss": 0.6055,
+      "step": 2747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9076165024897347e-05,
+      "loss": 1.0781,
+      "step": 2748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.907550309225894e-05,
+      "loss": 1.4766,
+      "step": 2749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9074840934059113e-05,
+      "loss": 0.8789,
+      "step": 2750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9074178550314325e-05,
+      "loss": 1.0391,
+      "step": 2751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9073515941041042e-05,
+      "loss": 1.6094,
+      "step": 2752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9072853106255726e-05,
+      "loss": 1.4453,
+      "step": 2753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.907219004597485e-05,
+      "loss": 1.3672,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.90715267602149e-05,
+      "loss": 1.0234,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9070863248992358e-05,
+      "loss": 1.0156,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9070199512323717e-05,
+      "loss": 0.5586,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9069535550225467e-05,
+      "loss": 0.8242,
+      "step": 2758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9068871362714118e-05,
+      "loss": 1.0703,
+      "step": 2759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9068206949806173e-05,
+      "loss": 0.6445,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9067542311518147e-05,
+      "loss": 1.2422,
+      "step": 2761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.906687744786656e-05,
+      "loss": 1.2578,
+      "step": 2762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9066212358867934e-05,
+      "loss": 0.5273,
+      "step": 2763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9065547044538796e-05,
+      "loss": 1.1719,
+      "step": 2764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9064881504895688e-05,
+      "loss": 1.3125,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.906421573995515e-05,
+      "loss": 1.3906,
+      "step": 2766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9063549749733726e-05,
+      "loss": 0.3867,
+      "step": 2767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.906288353424797e-05,
+      "loss": 2.0625,
+      "step": 2768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9062217093514438e-05,
+      "loss": 1.0469,
+      "step": 2769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.90615504275497e-05,
+      "loss": 1.25,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9060883536370318e-05,
+      "loss": 0.9258,
+      "step": 2771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9060216419992872e-05,
+      "loss": 2.2188,
+      "step": 2772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.905954907843394e-05,
+      "loss": 1.8984,
+      "step": 2773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9058881511710103e-05,
+      "loss": 1.0781,
+      "step": 2774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9058213719837962e-05,
+      "loss": 1.4766,
+      "step": 2775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9057545702834107e-05,
+      "loss": 1.0391,
+      "step": 2776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.905687746071515e-05,
+      "loss": 1.4844,
+      "step": 2777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9056208993497687e-05,
+      "loss": 0.9766,
+      "step": 2778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.905554030119834e-05,
+      "loss": 0.7148,
+      "step": 2779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9054871383833726e-05,
+      "loss": 1.1562,
+      "step": 2780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9054202241420474e-05,
+      "loss": 0.8125,
+      "step": 2781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.905353287397521e-05,
+      "loss": 1.0391,
+      "step": 2782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9052863281514572e-05,
+      "loss": 0.7812,
+      "step": 2783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9052193464055197e-05,
+      "loss": 0.4531,
+      "step": 2784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9051523421613745e-05,
+      "loss": 1.5469,
+      "step": 2785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.905085315420686e-05,
+      "loss": 1.3906,
+      "step": 2786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.90501826618512e-05,
+      "loss": 0.9219,
+      "step": 2787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9049511944563432e-05,
+      "loss": 0.7812,
+      "step": 2788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.904884100236023e-05,
+      "loss": 0.7812,
+      "step": 2789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.904816983525826e-05,
+      "loss": 1.7891,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9047498443274213e-05,
+      "loss": 1.5156,
+      "step": 2791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9046826826424768e-05,
+      "loss": 0.6836,
+      "step": 2792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.904615498472662e-05,
+      "loss": 0.9062,
+      "step": 2793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.904548291819647e-05,
+      "loss": 1.4062,
+      "step": 2794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9044810626851017e-05,
+      "loss": 1.0938,
+      "step": 2795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9044138110706973e-05,
+      "loss": 1.0078,
+      "step": 2796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.904346536978105e-05,
+      "loss": 0.9531,
+      "step": 2797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.904279240408997e-05,
+      "loss": 1.1172,
+      "step": 2798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9042119213650458e-05,
+      "loss": 0.9883,
+      "step": 2799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9041445798479247e-05,
+      "loss": 0.8672,
+      "step": 2800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.904077215859307e-05,
+      "loss": 0.6992,
+      "step": 2801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9040098294008675e-05,
+      "loss": 1.0703,
+      "step": 2802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9039424204742807e-05,
+      "loss": 0.6719,
+      "step": 2803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.903874989081222e-05,
+      "loss": 2.0938,
+      "step": 2804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9038075352233672e-05,
+      "loss": 1.8281,
+      "step": 2805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.903740058902393e-05,
+      "loss": 0.6406,
+      "step": 2806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.903672560119977e-05,
+      "loss": 1.7891,
+      "step": 2807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9036050388777957e-05,
+      "loss": 0.9258,
+      "step": 2808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9035374951775276e-05,
+      "loss": 1.2344,
+      "step": 2809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9034699290208517e-05,
+      "loss": 1.6328,
+      "step": 2810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.903402340409447e-05,
+      "loss": 1.1719,
+      "step": 2811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9033347293449938e-05,
+      "loss": 0.625,
+      "step": 2812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.903267095829172e-05,
+      "loss": 0.6836,
+      "step": 2813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.903199439863663e-05,
+      "loss": 1.7578,
+      "step": 2814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9031317614501476e-05,
+      "loss": 0.6367,
+      "step": 2815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9030640605903087e-05,
+      "loss": 1.3125,
+      "step": 2816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9029963372858287e-05,
+      "loss": 1.5781,
+      "step": 2817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9029285915383905e-05,
+      "loss": 0.6016,
+      "step": 2818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.902860823349678e-05,
+      "loss": 1.0547,
+      "step": 2819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9027930327213754e-05,
+      "loss": 1.4609,
+      "step": 2820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9027252196551677e-05,
+      "loss": 0.5117,
+      "step": 2821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.90265738415274e-05,
+      "loss": 1.3203,
+      "step": 2822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9025895262157788e-05,
+      "loss": 1.0625,
+      "step": 2823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.902521645845971e-05,
+      "loss": 0.7695,
+      "step": 2824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.902453743045002e-05,
+      "loss": 1.4688,
+      "step": 2825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9023858178145612e-05,
+      "loss": 1.5234,
+      "step": 2826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9023178701563357e-05,
+      "loss": 1.1328,
+      "step": 2827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.902249900072015e-05,
+      "loss": 0.8672,
+      "step": 2828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.902181907563288e-05,
+      "loss": 0.9258,
+      "step": 2829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.902113892631845e-05,
+      "loss": 0.9961,
+      "step": 2830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9020458552793756e-05,
+      "loss": 0.7734,
+      "step": 2831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9019777955075717e-05,
+      "loss": 0.9805,
+      "step": 2832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9019097133181244e-05,
+      "loss": 0.5547,
+      "step": 2833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9018416087127257e-05,
+      "loss": 1.375,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9017734816930686e-05,
+      "loss": 0.8359,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9017053322608462e-05,
+      "loss": 0.9062,
+      "step": 2836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9016371604177523e-05,
+      "loss": 1.1641,
+      "step": 2837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9015689661654808e-05,
+      "loss": 1.2344,
+      "step": 2838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9015007495057272e-05,
+      "loss": 1.2109,
+      "step": 2839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9014325104401866e-05,
+      "loss": 1.1328,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.901364248970555e-05,
+      "loss": 0.3711,
+      "step": 2841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9012959650985295e-05,
+      "loss": 0.6523,
+      "step": 2842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9012276588258065e-05,
+      "loss": 1.7969,
+      "step": 2843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.901159330154084e-05,
+      "loss": 1.3516,
+      "step": 2844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.90109097908506e-05,
+      "loss": 1.2188,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9010226056204337e-05,
+      "loss": 1.3984,
+      "step": 2846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9009542097619043e-05,
+      "loss": 1.3438,
+      "step": 2847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9008857915111715e-05,
+      "loss": 1.1328,
+      "step": 2848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.900817350869936e-05,
+      "loss": 0.875,
+      "step": 2849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.900748887839899e-05,
+      "loss": 1.125,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.900680402422761e-05,
+      "loss": 1.6406,
+      "step": 2851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9006118946202254e-05,
+      "loss": 0.5742,
+      "step": 2852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9005433644339945e-05,
+      "loss": 0.9414,
+      "step": 2853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.900474811865771e-05,
+      "loss": 0.6289,
+      "step": 2854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9004062369172594e-05,
+      "loss": 1.1172,
+      "step": 2855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9003376395901637e-05,
+      "loss": 1.6797,
+      "step": 2856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9002690198861892e-05,
+      "loss": 0.3945,
+      "step": 2857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9002003778070408e-05,
+      "loss": 0.4883,
+      "step": 2858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9001317133544245e-05,
+      "loss": 0.4941,
+      "step": 2859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9000630265300478e-05,
+      "loss": 0.3086,
+      "step": 2860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8999943173356167e-05,
+      "loss": 0.5352,
+      "step": 2861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8999255857728395e-05,
+      "loss": 0.6406,
+      "step": 2862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8998568318434244e-05,
+      "loss": 1.2031,
+      "step": 2863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8997880555490802e-05,
+      "loss": 0.918,
+      "step": 2864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.899719256891516e-05,
+      "loss": 0.8438,
+      "step": 2865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.899650435872442e-05,
+      "loss": 1.1719,
+      "step": 2866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8995815924935685e-05,
+      "loss": 0.8086,
+      "step": 2867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8995127267566067e-05,
+      "loss": 0.4316,
+      "step": 2868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.899443838663268e-05,
+      "loss": 0.6953,
+      "step": 2869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8993749282152646e-05,
+      "loss": 1.4766,
+      "step": 2870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8993059954143093e-05,
+      "loss": 1.4922,
+      "step": 2871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8992370402621153e-05,
+      "loss": 0.6289,
+      "step": 2872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.899168062760396e-05,
+      "loss": 1.4844,
+      "step": 2873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8990990629108664e-05,
+      "loss": 1.6484,
+      "step": 2874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8990300407152414e-05,
+      "loss": 1.0625,
+      "step": 2875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8989609961752355e-05,
+      "loss": 1.4141,
+      "step": 2876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.898891929292566e-05,
+      "loss": 1.5391,
+      "step": 2877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8988228400689487e-05,
+      "loss": 1.2031,
+      "step": 2878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.898753728506101e-05,
+      "loss": 0.7109,
+      "step": 2879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8986845946057403e-05,
+      "loss": 0.6641,
+      "step": 2880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.898615438369585e-05,
+      "loss": 0.4512,
+      "step": 2881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8985462597993544e-05,
+      "loss": 1.3359,
+      "step": 2882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.898477058896767e-05,
+      "loss": 0.7773,
+      "step": 2883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8984078356635433e-05,
+      "loss": 0.6016,
+      "step": 2884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8983385901014036e-05,
+      "loss": 1.3828,
+      "step": 2885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.898269322212069e-05,
+      "loss": 1.0625,
+      "step": 2886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.898200031997261e-05,
+      "loss": 0.8281,
+      "step": 2887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8981307194587015e-05,
+      "loss": 1.4609,
+      "step": 2888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8980613845981137e-05,
+      "loss": 1.2031,
+      "step": 2889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8979920274172204e-05,
+      "loss": 1.2344,
+      "step": 2890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8979226479177455e-05,
+      "loss": 0.3203,
+      "step": 2891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8978532461014133e-05,
+      "loss": 0.9219,
+      "step": 2892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8977838219699493e-05,
+      "loss": 0.7578,
+      "step": 2893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.897714375525078e-05,
+      "loss": 1.2188,
+      "step": 2894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.897644906768526e-05,
+      "loss": 1.3359,
+      "step": 2895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8975754157020196e-05,
+      "loss": 1.2578,
+      "step": 2896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8975059023272862e-05,
+      "loss": 0.9609,
+      "step": 2897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.897436366646053e-05,
+      "loss": 1.1641,
+      "step": 2898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8973668086600493e-05,
+      "loss": 1.6641,
+      "step": 2899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8972972283710026e-05,
+      "loss": 0.6953,
+      "step": 2900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.897227625780643e-05,
+      "loss": 0.6758,
+      "step": 2901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8971580008907e-05,
+      "loss": 1.1172,
+      "step": 2902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8970883537029045e-05,
+      "loss": 1.1406,
+      "step": 2903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8970186842189872e-05,
+      "loss": 1.3203,
+      "step": 2904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8969489924406793e-05,
+      "loss": 0.4141,
+      "step": 2905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8968792783697138e-05,
+      "loss": 1.1016,
+      "step": 2906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8968095420078226e-05,
+      "loss": 0.6953,
+      "step": 2907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8967397833567393e-05,
+      "loss": 0.7617,
+      "step": 2908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8966700024181973e-05,
+      "loss": 1.0469,
+      "step": 2909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8966001991939316e-05,
+      "loss": 1.1719,
+      "step": 2910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.896530373685676e-05,
+      "loss": 1.2734,
+      "step": 2911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8964605258951675e-05,
+      "loss": 0.75,
+      "step": 2912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8963906558241407e-05,
+      "loss": 0.7031,
+      "step": 2913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8963207634743327e-05,
+      "loss": 1.4453,
+      "step": 2914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8962508488474806e-05,
+      "loss": 0.582,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8961809119453218e-05,
+      "loss": 1.5156,
+      "step": 2916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8961109527695948e-05,
+      "loss": 1.375,
+      "step": 2917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8960409713220383e-05,
+      "loss": 1.3203,
+      "step": 2918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8959709676043913e-05,
+      "loss": 1.7656,
+      "step": 2919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8959009416183944e-05,
+      "loss": 0.7109,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8958308933657868e-05,
+      "loss": 1.2891,
+      "step": 2921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8957608228483107e-05,
+      "loss": 1.1172,
+      "step": 2922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8956907300677073e-05,
+      "loss": 0.9883,
+      "step": 2923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.895620615025718e-05,
+      "loss": 0.6992,
+      "step": 2924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8955504777240864e-05,
+      "loss": 0.9023,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.895480318164555e-05,
+      "loss": 1.0859,
+      "step": 2926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8954101363488676e-05,
+      "loss": 1.0234,
+      "step": 2927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.895339932278769e-05,
+      "loss": 1.7188,
+      "step": 2928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8952697059560034e-05,
+      "loss": 1.3672,
+      "step": 2929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8951994573823166e-05,
+      "loss": 1.0859,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.895129186559454e-05,
+      "loss": 0.9766,
+      "step": 2931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.895058893489163e-05,
+      "loss": 0.8672,
+      "step": 2932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8949885781731903e-05,
+      "loss": 0.6992,
+      "step": 2933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8949182406132832e-05,
+      "loss": 0.8438,
+      "step": 2934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.89484788081119e-05,
+      "loss": 1.2344,
+      "step": 2935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8947774987686597e-05,
+      "loss": 1.4141,
+      "step": 2936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.894707094487441e-05,
+      "loss": 1.0156,
+      "step": 2937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8946366679692843e-05,
+      "loss": 0.7656,
+      "step": 2938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8945662192159395e-05,
+      "loss": 1.6641,
+      "step": 2939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.894495748229158e-05,
+      "loss": 0.957,
+      "step": 2940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.894425255010691e-05,
+      "loss": 0.7969,
+      "step": 2941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8943547395622904e-05,
+      "loss": 1.4375,
+      "step": 2942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.894284201885709e-05,
+      "loss": 1.3281,
+      "step": 2943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8942136419827003e-05,
+      "loss": 1.125,
+      "step": 2944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.894143059855017e-05,
+      "loss": 1.3125,
+      "step": 2945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.894072455504414e-05,
+      "loss": 1.3438,
+      "step": 2946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8940018289326462e-05,
+      "loss": 0.9414,
+      "step": 2947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8939311801414687e-05,
+      "loss": 0.5625,
+      "step": 2948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8938605091326373e-05,
+      "loss": 0.8281,
+      "step": 2949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8937898159079088e-05,
+      "loss": 1.2109,
+      "step": 2950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.89371910046904e-05,
+      "loss": 1.2031,
+      "step": 2951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.893648362817788e-05,
+      "loss": 1.1016,
+      "step": 2952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8935776029559116e-05,
+      "loss": 0.4102,
+      "step": 2953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.893506820885169e-05,
+      "loss": 0.4746,
+      "step": 2954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.89343601660732e-05,
+      "loss": 0.6836,
+      "step": 2955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.893365190124124e-05,
+      "loss": 1.125,
+      "step": 2956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8932943414373407e-05,
+      "loss": 0.8047,
+      "step": 2957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.893223470548732e-05,
+      "loss": 0.9219,
+      "step": 2958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8931525774600587e-05,
+      "loss": 0.4336,
+      "step": 2959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8930816621730832e-05,
+      "loss": 0.7109,
+      "step": 2960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8930107246895673e-05,
+      "loss": 0.9688,
+      "step": 2961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.892939765011275e-05,
+      "loss": 0.8672,
+      "step": 2962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.892868783139969e-05,
+      "loss": 0.6602,
+      "step": 2963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.892797779077414e-05,
+      "loss": 0.5352,
+      "step": 2964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.892726752825375e-05,
+      "loss": 0.2451,
+      "step": 2965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8926557043856166e-05,
+      "loss": 0.832,
+      "step": 2966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8925846337599052e-05,
+      "loss": 1.3047,
+      "step": 2967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.892513540950007e-05,
+      "loss": 0.5117,
+      "step": 2968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8924424259576882e-05,
+      "loss": 0.7969,
+      "step": 2969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8923712887847175e-05,
+      "loss": 0.8164,
+      "step": 2970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8923001294328622e-05,
+      "loss": 0.6641,
+      "step": 2971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8922289479038914e-05,
+      "loss": 1.1328,
+      "step": 2972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8921577441995733e-05,
+      "loss": 1.4141,
+      "step": 2973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8920865183216785e-05,
+      "loss": 0.6797,
+      "step": 2974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.892015270271977e-05,
+      "loss": 0.957,
+      "step": 2975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8919440000522396e-05,
+      "loss": 1.3594,
+      "step": 2976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8918727076642373e-05,
+      "loss": 1.5,
+      "step": 2977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8918013931097423e-05,
+      "loss": 1.0078,
+      "step": 2978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8917300563905272e-05,
+      "loss": 0.9102,
+      "step": 2979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8916586975083647e-05,
+      "loss": 1.2891,
+      "step": 2980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8915873164650285e-05,
+      "loss": 1.4922,
+      "step": 2981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8915159132622926e-05,
+      "loss": 1.0,
+      "step": 2982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8914444879019314e-05,
+      "loss": 1.0859,
+      "step": 2983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8913730403857205e-05,
+      "loss": 0.7109,
+      "step": 2984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8913015707154358e-05,
+      "loss": 0.7734,
+      "step": 2985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8912300788928532e-05,
+      "loss": 0.6992,
+      "step": 2986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8911585649197495e-05,
+      "loss": 1.375,
+      "step": 2987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8910870287979022e-05,
+      "loss": 0.9258,
+      "step": 2988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8910154705290896e-05,
+      "loss": 1.2656,
+      "step": 2989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8909438901150898e-05,
+      "loss": 0.9766,
+      "step": 2990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.890872287557682e-05,
+      "loss": 1.2969,
+      "step": 2991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8908006628586456e-05,
+      "loss": 1.1797,
+      "step": 2992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.890729016019761e-05,
+      "loss": 0.9609,
+      "step": 2993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8906573470428088e-05,
+      "loss": 1.1406,
+      "step": 2994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8905856559295707e-05,
+      "loss": 1.25,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8905139426818275e-05,
+      "loss": 1.0312,
+      "step": 2996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8904422073013622e-05,
+      "loss": 1.0469,
+      "step": 2997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.890370449789958e-05,
+      "loss": 0.9609,
+      "step": 2998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8902986701493978e-05,
+      "loss": 0.8867,
+      "step": 2999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8902268683814658e-05,
+      "loss": 1.1875,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8901550444879463e-05,
+      "loss": 1.3672,
+      "step": 3001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.890083198470625e-05,
+      "loss": 0.9727,
+      "step": 3002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8900113303312867e-05,
+      "loss": 1.2422,
+      "step": 3003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8899394400717183e-05,
+      "loss": 0.5547,
+      "step": 3004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8898675276937067e-05,
+      "loss": 1.6562,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8897955931990387e-05,
+      "loss": 0.8594,
+      "step": 3006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8897236365895022e-05,
+      "loss": 0.6562,
+      "step": 3007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8896516578668855e-05,
+      "loss": 0.418,
+      "step": 3008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.889579657032978e-05,
+      "loss": 0.498,
+      "step": 3009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.889507634089569e-05,
+      "loss": 1.0781,
+      "step": 3010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.889435589038448e-05,
+      "loss": 0.8281,
+      "step": 3011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8893635218814066e-05,
+      "loss": 1.4297,
+      "step": 3012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8892914326202354e-05,
+      "loss": 0.8711,
+      "step": 3013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.889219321256726e-05,
+      "loss": 0.3691,
+      "step": 3014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.889147187792671e-05,
+      "loss": 0.7109,
+      "step": 3015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8890750322298627e-05,
+      "loss": 0.4414,
+      "step": 3016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.889002854570095e-05,
+      "loss": 0.9102,
+      "step": 3017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888930654815161e-05,
+      "loss": 1.4688,
+      "step": 3018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888858432966856e-05,
+      "loss": 0.498,
+      "step": 3019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888786189026975e-05,
+      "loss": 1.875,
+      "step": 3020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888713922997313e-05,
+      "loss": 1.9375,
+      "step": 3021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888641634879666e-05,
+      "loss": 1.9766,
+      "step": 3022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888569324675831e-05,
+      "loss": 0.9609,
+      "step": 3023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8884969923876056e-05,
+      "loss": 1.4844,
+      "step": 3024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8884246380167862e-05,
+      "loss": 0.543,
+      "step": 3025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8883522615651725e-05,
+      "loss": 1.0625,
+      "step": 3026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888279863034563e-05,
+      "loss": 1.3438,
+      "step": 3027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8882074424267564e-05,
+      "loss": 1.3125,
+      "step": 3028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8881349997435535e-05,
+      "loss": 1.1484,
+      "step": 3029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.888062534986754e-05,
+      "loss": 1.2578,
+      "step": 3030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8879900481581597e-05,
+      "loss": 1.1875,
+      "step": 3031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.887917539259571e-05,
+      "loss": 1.4141,
+      "step": 3032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8878450082927916e-05,
+      "loss": 1.0234,
+      "step": 3033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8877724552596234e-05,
+      "loss": 0.9102,
+      "step": 3034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8876998801618693e-05,
+      "loss": 0.5781,
+      "step": 3035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8876272830013332e-05,
+      "loss": 1.2109,
+      "step": 3036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.88755466377982e-05,
+      "loss": 1.5234,
+      "step": 3037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.887482022499134e-05,
+      "loss": 1.2969,
+      "step": 3038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.887409359161081e-05,
+      "loss": 1.6094,
+      "step": 3039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8873366737674666e-05,
+      "loss": 0.7227,
+      "step": 3040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8872639663200973e-05,
+      "loss": 0.7344,
+      "step": 3041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8871912368207805e-05,
+      "loss": 1.1172,
+      "step": 3042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8871184852713238e-05,
+      "loss": 1.0625,
+      "step": 3043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.887045711673535e-05,
+      "loss": 1.9531,
+      "step": 3044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.886972916029223e-05,
+      "loss": 0.8633,
+      "step": 3045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8869000983401974e-05,
+      "loss": 1.0547,
+      "step": 3046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8868272586082673e-05,
+      "loss": 0.4844,
+      "step": 3047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8867543968352432e-05,
+      "loss": 1.0156,
+      "step": 3048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8866815130229365e-05,
+      "loss": 0.8125,
+      "step": 3049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8866086071731585e-05,
+      "loss": 1.0234,
+      "step": 3050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8865356792877208e-05,
+      "loss": 0.6758,
+      "step": 3051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8864627293684364e-05,
+      "loss": 0.8945,
+      "step": 3052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.886389757417118e-05,
+      "loss": 0.3691,
+      "step": 3053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8863167634355794e-05,
+      "loss": 1.5312,
+      "step": 3054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.886243747425635e-05,
+      "loss": 1.4453,
+      "step": 3055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8861707093890992e-05,
+      "loss": 1.9297,
+      "step": 3056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8860976493277872e-05,
+      "loss": 0.7617,
+      "step": 3057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8860245672435153e-05,
+      "loss": 0.6133,
+      "step": 3058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8859514631380994e-05,
+      "loss": 0.5078,
+      "step": 3059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8858783370133567e-05,
+      "loss": 0.5039,
+      "step": 3060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8858051888711052e-05,
+      "loss": 0.8359,
+      "step": 3061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8857320187131612e-05,
+      "loss": 1.1484,
+      "step": 3062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8856588265413453e-05,
+      "loss": 1.2109,
+      "step": 3063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8855856123574755e-05,
+      "loss": 1.1484,
+      "step": 3064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.885512376163371e-05,
+      "loss": 0.9062,
+      "step": 3065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.885439117960853e-05,
+      "loss": 0.7812,
+      "step": 3066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.885365837751742e-05,
+      "loss": 0.9375,
+      "step": 3067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8852925355378594e-05,
+      "loss": 0.6484,
+      "step": 3068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8852192113210263e-05,
+      "loss": 1.0156,
+      "step": 3069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.885145865103066e-05,
+      "loss": 1.2031,
+      "step": 3070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.885072496885801e-05,
+      "loss": 1.0469,
+      "step": 3071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.884999106671055e-05,
+      "loss": 0.6953,
+      "step": 3072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8849256944606517e-05,
+      "loss": 1.5078,
+      "step": 3073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8848522602564156e-05,
+      "loss": 1.1562,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.884778804060172e-05,
+      "loss": 1.1094,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8847053258737472e-05,
+      "loss": 1.5938,
+      "step": 3076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8846318256989666e-05,
+      "loss": 1.3438,
+      "step": 3077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.884558303537657e-05,
+      "loss": 1.1406,
+      "step": 3078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8844847593916465e-05,
+      "loss": 0.4883,
+      "step": 3079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.884411193262762e-05,
+      "loss": 0.793,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8843376051528325e-05,
+      "loss": 0.9141,
+      "step": 3081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8842639950636866e-05,
+      "loss": 1.1172,
+      "step": 3082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8841903629971537e-05,
+      "loss": 1.0547,
+      "step": 3083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8841167089550642e-05,
+      "loss": 1.5703,
+      "step": 3084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.884043032939249e-05,
+      "loss": 0.7812,
+      "step": 3085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8839693349515385e-05,
+      "loss": 1.1641,
+      "step": 3086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8838956149937646e-05,
+      "loss": 0.6367,
+      "step": 3087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8838218730677598e-05,
+      "loss": 0.6953,
+      "step": 3088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8837481091753565e-05,
+      "loss": 0.377,
+      "step": 3089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8836743233183884e-05,
+      "loss": 1.1172,
+      "step": 3090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.883600515498689e-05,
+      "loss": 0.5977,
+      "step": 3091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8835266857180933e-05,
+      "loss": 1.2109,
+      "step": 3092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8834528339784354e-05,
+      "loss": 1.7656,
+      "step": 3093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8833789602815512e-05,
+      "loss": 1.1172,
+      "step": 3094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8833050646292774e-05,
+      "loss": 1.0312,
+      "step": 3095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8832311470234494e-05,
+      "loss": 1.0938,
+      "step": 3096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8831572074659047e-05,
+      "loss": 1.1094,
+      "step": 3097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8830832459584817e-05,
+      "loss": 0.6172,
+      "step": 3098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.883009262503018e-05,
+      "loss": 1.1875,
+      "step": 3099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8829352571013528e-05,
+      "loss": 0.7539,
+      "step": 3100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8828612297553243e-05,
+      "loss": 1.6875,
+      "step": 3101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8827871804667738e-05,
+      "loss": 1.4688,
+      "step": 3102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8827131092375414e-05,
+      "loss": 0.9453,
+      "step": 3103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8826390160694672e-05,
+      "loss": 1.3125,
+      "step": 3104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8825649009643933e-05,
+      "loss": 1.1484,
+      "step": 3105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8824907639241622e-05,
+      "loss": 1.0938,
+      "step": 3106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8824166049506157e-05,
+      "loss": 1.5,
+      "step": 3107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.882342424045597e-05,
+      "loss": 1.2812,
+      "step": 3108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8822682212109504e-05,
+      "loss": 0.625,
+      "step": 3109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8821939964485196e-05,
+      "loss": 0.668,
+      "step": 3110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8821197497601496e-05,
+      "loss": 1.8047,
+      "step": 3111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8820454811476857e-05,
+      "loss": 0.3242,
+      "step": 3112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8819711906129733e-05,
+      "loss": 0.75,
+      "step": 3113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8818968781578598e-05,
+      "loss": 0.6562,
+      "step": 3114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8818225437841912e-05,
+      "loss": 1.0156,
+      "step": 3115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8817481874938155e-05,
+      "loss": 0.9023,
+      "step": 3116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8816738092885805e-05,
+      "loss": 0.3789,
+      "step": 3117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8815994091703352e-05,
+      "loss": 0.7422,
+      "step": 3118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.881524987140928e-05,
+      "loss": 0.9609,
+      "step": 3119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8814505432022092e-05,
+      "loss": 1.2578,
+      "step": 3120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.881376077356029e-05,
+      "loss": 1.4297,
+      "step": 3121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8813015896042377e-05,
+      "loss": 1.5469,
+      "step": 3122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.881227079948687e-05,
+      "loss": 0.6484,
+      "step": 3123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8811525483912287e-05,
+      "loss": 1.5,
+      "step": 3124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8810779949337155e-05,
+      "loss": 1.1797,
+      "step": 3125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8810034195779994e-05,
+      "loss": 1.2266,
+      "step": 3126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.880928822325935e-05,
+      "loss": 1.0234,
+      "step": 3127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8808542031793754e-05,
+      "loss": 0.9102,
+      "step": 3128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8807795621401757e-05,
+      "loss": 0.9688,
+      "step": 3129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8807048992101908e-05,
+      "loss": 1.0781,
+      "step": 3130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8806302143912763e-05,
+      "loss": 1.6328,
+      "step": 3131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8805555076852894e-05,
+      "loss": 0.6914,
+      "step": 3132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8804807790940852e-05,
+      "loss": 1.1797,
+      "step": 3133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8804060286195224e-05,
+      "loss": 1.5781,
+      "step": 3134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8803312562634577e-05,
+      "loss": 1.0156,
+      "step": 3135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8802564620277504e-05,
+      "loss": 1.6641,
+      "step": 3136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.880181645914259e-05,
+      "loss": 1.4062,
+      "step": 3137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8801068079248427e-05,
+      "loss": 1.4453,
+      "step": 3138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8800319480613622e-05,
+      "loss": 0.2793,
+      "step": 3139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.879957066325678e-05,
+      "loss": 0.6016,
+      "step": 3140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8798821627196504e-05,
+      "loss": 1.0703,
+      "step": 3141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8798072372451418e-05,
+      "loss": 0.957,
+      "step": 3142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.879732289904014e-05,
+      "loss": 1.3359,
+      "step": 3143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.87965732069813e-05,
+      "loss": 0.8242,
+      "step": 3144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8795823296293528e-05,
+      "loss": 1.8438,
+      "step": 3145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8795073166995468e-05,
+      "loss": 1.1797,
+      "step": 3146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8794322819105753e-05,
+      "loss": 0.8516,
+      "step": 3147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8793572252643043e-05,
+      "loss": 0.8594,
+      "step": 3148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8792821467625987e-05,
+      "loss": 1.2344,
+      "step": 3149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8792070464073245e-05,
+      "loss": 0.8633,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8791319242003486e-05,
+      "loss": 1.3359,
+      "step": 3151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8790567801435374e-05,
+      "loss": 1.4375,
+      "step": 3152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.878981614238759e-05,
+      "loss": 1.3828,
+      "step": 3153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8789064264878815e-05,
+      "loss": 1.4062,
+      "step": 3154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8788312168927735e-05,
+      "loss": 1.1719,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8787559854553045e-05,
+      "loss": 1.0312,
+      "step": 3156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8786807321773444e-05,
+      "loss": 1.2422,
+      "step": 3157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8786054570607628e-05,
+      "loss": 1.0625,
+      "step": 3158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.878530160107431e-05,
+      "loss": 1.0859,
+      "step": 3159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8784548413192207e-05,
+      "loss": 1.2891,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8783795006980037e-05,
+      "loss": 0.3809,
+      "step": 3161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8783041382456523e-05,
+      "loss": 0.4473,
+      "step": 3162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8782287539640397e-05,
+      "loss": 0.6289,
+      "step": 3163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8781533478550393e-05,
+      "loss": 1.375,
+      "step": 3164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8780779199205257e-05,
+      "loss": 0.8008,
+      "step": 3165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.878002470162373e-05,
+      "loss": 1.3594,
+      "step": 3166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8779269985824568e-05,
+      "loss": 0.7695,
+      "step": 3167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.877851505182653e-05,
+      "loss": 0.9961,
+      "step": 3168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8777759899648377e-05,
+      "loss": 0.3965,
+      "step": 3169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.877700452930887e-05,
+      "loss": 0.582,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8776248940826796e-05,
+      "loss": 0.4844,
+      "step": 3171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8775493134220926e-05,
+      "loss": 1.6016,
+      "step": 3172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.877473710951005e-05,
+      "loss": 1.5625,
+      "step": 3173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8773980866712957e-05,
+      "loss": 1.1406,
+      "step": 3174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8773224405848434e-05,
+      "loss": 0.668,
+      "step": 3175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.877246772693529e-05,
+      "loss": 1.2969,
+      "step": 3176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8771710829992334e-05,
+      "loss": 1.7109,
+      "step": 3177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.877095371503837e-05,
+      "loss": 0.7656,
+      "step": 3178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8770196382092223e-05,
+      "loss": 0.6797,
+      "step": 3179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8769438831172707e-05,
+      "loss": 1.4688,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.876868106229866e-05,
+      "loss": 0.7891,
+      "step": 3181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8767923075488908e-05,
+      "loss": 0.6758,
+      "step": 3182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.876716487076229e-05,
+      "loss": 0.8906,
+      "step": 3183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8766406448137658e-05,
+      "loss": 0.9531,
+      "step": 3184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8765647807633854e-05,
+      "loss": 0.6523,
+      "step": 3185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8764888949269734e-05,
+      "loss": 1.3672,
+      "step": 3186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8764129873064162e-05,
+      "loss": 0.918,
+      "step": 3187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8763370579036e-05,
+      "loss": 1.4062,
+      "step": 3188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.876261106720412e-05,
+      "loss": 1.1875,
+      "step": 3189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8761851337587407e-05,
+      "loss": 0.9336,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8761091390204732e-05,
+      "loss": 1.1719,
+      "step": 3191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8760331225074986e-05,
+      "loss": 1.0078,
+      "step": 3192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8759570842217064e-05,
+      "loss": 1.3516,
+      "step": 3193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8758810241649868e-05,
+      "loss": 1.1406,
+      "step": 3194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.875804942339229e-05,
+      "loss": 0.6094,
+      "step": 3195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8757288387463253e-05,
+      "loss": 0.9844,
+      "step": 3196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8756527133881665e-05,
+      "loss": 1.1719,
+      "step": 3197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8755765662666443e-05,
+      "loss": 1.0625,
+      "step": 3198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8755003973836518e-05,
+      "loss": 0.6172,
+      "step": 3199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8754242067410818e-05,
+      "loss": 1.0625,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8753479943408283e-05,
+      "loss": 0.7188,
+      "step": 3201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.875271760184785e-05,
+      "loss": 0.4219,
+      "step": 3202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8751955042748472e-05,
+      "loss": 1.2656,
+      "step": 3203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.875119226612909e-05,
+      "loss": 0.4609,
+      "step": 3204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8750429272008678e-05,
+      "loss": 0.5117,
+      "step": 3205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.874966606040619e-05,
+      "loss": 0.6055,
+      "step": 3206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.874890263134059e-05,
+      "loss": 0.5312,
+      "step": 3207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8748138984830864e-05,
+      "loss": 1.0625,
+      "step": 3208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.874737512089598e-05,
+      "loss": 1.6172,
+      "step": 3209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8746611039554935e-05,
+      "loss": 1.3047,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8745846740826707e-05,
+      "loss": 1.0625,
+      "step": 3211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.87450822247303e-05,
+      "loss": 1.3125,
+      "step": 3212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8744317491284716e-05,
+      "loss": 0.8438,
+      "step": 3213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8743552540508956e-05,
+      "loss": 1.4766,
+      "step": 3214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8742787372422034e-05,
+      "loss": 1.4062,
+      "step": 3215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.874202198704297e-05,
+      "loss": 1.1172,
+      "step": 3216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.874125638439078e-05,
+      "loss": 1.6094,
+      "step": 3217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8740490564484505e-05,
+      "loss": 1.3359,
+      "step": 3218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8739724527343162e-05,
+      "loss": 0.4727,
+      "step": 3219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8738958272985803e-05,
+      "loss": 0.6328,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.873819180143147e-05,
+      "loss": 0.7734,
+      "step": 3221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8737425112699207e-05,
+      "loss": 0.6094,
+      "step": 3222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8736658206808073e-05,
+      "loss": 0.6562,
+      "step": 3223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.873589108377713e-05,
+      "loss": 0.7656,
+      "step": 3224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.873512374362544e-05,
+      "loss": 1.1953,
+      "step": 3225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8734356186372078e-05,
+      "loss": 0.9414,
+      "step": 3226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8733588412036118e-05,
+      "loss": 1.4062,
+      "step": 3227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.873282042063665e-05,
+      "loss": 1.0938,
+      "step": 3228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.873205221219275e-05,
+      "loss": 0.9922,
+      "step": 3229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.873128378672352e-05,
+      "loss": 1.1406,
+      "step": 3230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8730515144248048e-05,
+      "loss": 1.1562,
+      "step": 3231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.872974628478545e-05,
+      "loss": 1.125,
+      "step": 3232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8728977208354825e-05,
+      "loss": 1.5234,
+      "step": 3233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8728207914975297e-05,
+      "loss": 0.5273,
+      "step": 3234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.872743840466598e-05,
+      "loss": 0.957,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8726668677445995e-05,
+      "loss": 1.7109,
+      "step": 3236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8725898733334486e-05,
+      "loss": 0.7578,
+      "step": 3237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8725128572350575e-05,
+      "loss": 0.5312,
+      "step": 3238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8724358194513413e-05,
+      "loss": 1.5,
+      "step": 3239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8723587599842142e-05,
+      "loss": 1.5078,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8722816788355916e-05,
+      "loss": 1.1172,
+      "step": 3241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.872204576007389e-05,
+      "loss": 1.3906,
+      "step": 3242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8721274515015235e-05,
+      "loss": 1.3438,
+      "step": 3243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8720503053199107e-05,
+      "loss": 1.4062,
+      "step": 3244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.871973137464469e-05,
+      "loss": 1.0234,
+      "step": 3245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.871895947937116e-05,
+      "loss": 1.1719,
+      "step": 3246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.87181873673977e-05,
+      "loss": 1.125,
+      "step": 3247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8717415038743506e-05,
+      "loss": 1.0312,
+      "step": 3248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8716642493427762e-05,
+      "loss": 0.8086,
+      "step": 3249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.871586973146968e-05,
+      "loss": 0.8242,
+      "step": 3250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.871509675288846e-05,
+      "loss": 0.9062,
+      "step": 3251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8714323557703316e-05,
+      "loss": 0.5859,
+      "step": 3252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.871355014593346e-05,
+      "loss": 0.5859,
+      "step": 3253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8712776517598126e-05,
+      "loss": 0.4863,
+      "step": 3254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.871200267271653e-05,
+      "loss": 1.2656,
+      "step": 3255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8711228611307904e-05,
+      "loss": 0.4727,
+      "step": 3256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8710454333391498e-05,
+      "loss": 0.7227,
+      "step": 3257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8709679838986548e-05,
+      "loss": 1.0625,
+      "step": 3258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8708905128112304e-05,
+      "loss": 1.1641,
+      "step": 3259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.870813020078802e-05,
+      "loss": 1.0703,
+      "step": 3260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8707355057032958e-05,
+      "loss": 1.5469,
+      "step": 3261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8706579696866383e-05,
+      "loss": 0.3203,
+      "step": 3262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8705804120307563e-05,
+      "loss": 0.7461,
+      "step": 3263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.870502832737578e-05,
+      "loss": 0.7227,
+      "step": 3264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.870425231809031e-05,
+      "loss": 1.1172,
+      "step": 3265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8703476092470437e-05,
+      "loss": 0.6758,
+      "step": 3266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.870269965053546e-05,
+      "loss": 0.9375,
+      "step": 3267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8701922992304676e-05,
+      "loss": 1.8828,
+      "step": 3268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8701146117797386e-05,
+      "loss": 0.8438,
+      "step": 3269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8700369027032894e-05,
+      "loss": 2.0,
+      "step": 3270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8699591720030523e-05,
+      "loss": 0.5625,
+      "step": 3271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8698814196809585e-05,
+      "loss": 1.3516,
+      "step": 3272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8698036457389407e-05,
+      "loss": 1.8594,
+      "step": 3273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.869725850178932e-05,
+      "loss": 0.9766,
+      "step": 3274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8696480330028656e-05,
+      "loss": 1.8594,
+      "step": 3275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.869570194212676e-05,
+      "loss": 1.4141,
+      "step": 3276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.869492333810297e-05,
+      "loss": 0.7109,
+      "step": 3277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8694144517976648e-05,
+      "loss": 1.875,
+      "step": 3278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8693365481767143e-05,
+      "loss": 1.125,
+      "step": 3279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8692586229493822e-05,
+      "loss": 1.5,
+      "step": 3280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8691806761176047e-05,
+      "loss": 0.7617,
+      "step": 3281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8691027076833195e-05,
+      "loss": 1.1172,
+      "step": 3282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8690247176484645e-05,
+      "loss": 1.1172,
+      "step": 3283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8689467060149772e-05,
+      "loss": 1.2422,
+      "step": 3284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.868868672784798e-05,
+      "loss": 0.8164,
+      "step": 3285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8687906179598646e-05,
+      "loss": 1.0938,
+      "step": 3286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8687125415421184e-05,
+      "loss": 1.125,
+      "step": 3287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.868634443533499e-05,
+      "loss": 0.498,
+      "step": 3288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8685563239359478e-05,
+      "loss": 1.0078,
+      "step": 3289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8684781827514065e-05,
+      "loss": 0.8828,
+      "step": 3290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.868400019981817e-05,
+      "loss": 0.7969,
+      "step": 3291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.868321835629122e-05,
+      "loss": 0.8594,
+      "step": 3292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8682436296952646e-05,
+      "loss": 0.7969,
+      "step": 3293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8681654021821885e-05,
+      "loss": 1.3594,
+      "step": 3294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8680871530918384e-05,
+      "loss": 1.375,
+      "step": 3295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8680088824261583e-05,
+      "loss": 1.0625,
+      "step": 3296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8679305901870944e-05,
+      "loss": 1.3203,
+      "step": 3297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.867852276376592e-05,
+      "loss": 0.9453,
+      "step": 3298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8677739409965975e-05,
+      "loss": 1.1484,
+      "step": 3299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.867695584049058e-05,
+      "loss": 0.7734,
+      "step": 3300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8676172055359213e-05,
+      "loss": 1.4531,
+      "step": 3301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.867538805459135e-05,
+      "loss": 1.5234,
+      "step": 3302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.867460383820647e-05,
+      "loss": 1.0234,
+      "step": 3303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8673819406224082e-05,
+      "loss": 1.2188,
+      "step": 3304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8673034758663665e-05,
+      "loss": 1.4141,
+      "step": 3305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8672249895544728e-05,
+      "loss": 1.125,
+      "step": 3306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8671464816886774e-05,
+      "loss": 1.2031,
+      "step": 3307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.867067952270932e-05,
+      "loss": 1.4062,
+      "step": 3308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8669894013031886e-05,
+      "loss": 0.8477,
+      "step": 3309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8669108287873982e-05,
+      "loss": 0.6211,
+      "step": 3310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.866832234725515e-05,
+      "loss": 0.9766,
+      "step": 3311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8667536191194915e-05,
+      "loss": 0.4609,
+      "step": 3312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.866674981971282e-05,
+      "loss": 0.918,
+      "step": 3313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.866596323282841e-05,
+      "loss": 0.6875,
+      "step": 3314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8665176430561236e-05,
+      "loss": 1.4141,
+      "step": 3315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8664389412930846e-05,
+      "loss": 0.9336,
+      "step": 3316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.866360217995681e-05,
+      "loss": 0.5195,
+      "step": 3317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8662814731658683e-05,
+      "loss": 1.0625,
+      "step": 3318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8662027068056047e-05,
+      "loss": 1.3125,
+      "step": 3319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8661239189168468e-05,
+      "loss": 1.1719,
+      "step": 3320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8660451095015536e-05,
+      "loss": 0.4277,
+      "step": 3321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8659662785616838e-05,
+      "loss": 0.9766,
+      "step": 3322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.865887426099196e-05,
+      "loss": 0.8516,
+      "step": 3323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8658085521160506e-05,
+      "loss": 0.8633,
+      "step": 3324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8657296566142078e-05,
+      "loss": 1.3828,
+      "step": 3325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8656507395956284e-05,
+      "loss": 1.7422,
+      "step": 3326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8655718010622736e-05,
+      "loss": 1.9297,
+      "step": 3327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8654928410161054e-05,
+      "loss": 1.4297,
+      "step": 3328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8654138594590866e-05,
+      "loss": 1.375,
+      "step": 3329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.86533485639318e-05,
+      "loss": 1.0312,
+      "step": 3330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8652558318203485e-05,
+      "loss": 0.6055,
+      "step": 3331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8651767857425574e-05,
+      "loss": 1.2891,
+      "step": 3332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8650977181617706e-05,
+      "loss": 1.2734,
+      "step": 3333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.865018629079953e-05,
+      "loss": 1.3125,
+      "step": 3334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.864939518499071e-05,
+      "loss": 0.8945,
+      "step": 3335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.86486038642109e-05,
+      "loss": 1.2031,
+      "step": 3336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.864781232847977e-05,
+      "loss": 0.957,
+      "step": 3337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8647020577816997e-05,
+      "loss": 1.3203,
+      "step": 3338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8646228612242253e-05,
+      "loss": 0.4902,
+      "step": 3339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.864543643177523e-05,
+      "loss": 0.9492,
+      "step": 3340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8644644036435605e-05,
+      "loss": 1.9609,
+      "step": 3341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.864385142624308e-05,
+      "loss": 1.2109,
+      "step": 3342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.864305860121735e-05,
+      "loss": 1.5312,
+      "step": 3343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8642265561378127e-05,
+      "loss": 0.9023,
+      "step": 3344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8641472306745112e-05,
+      "loss": 0.9961,
+      "step": 3345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8640678837338027e-05,
+      "loss": 1.4062,
+      "step": 3346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8639885153176594e-05,
+      "loss": 0.8398,
+      "step": 3347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8639091254280532e-05,
+      "loss": 0.6328,
+      "step": 3348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8638297140669576e-05,
+      "loss": 0.5898,
+      "step": 3349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8637502812363463e-05,
+      "loss": 0.8086,
+      "step": 3350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.863670826938194e-05,
+      "loss": 0.7461,
+      "step": 3351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8635913511744745e-05,
+      "loss": 0.6406,
+      "step": 3352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8635118539471638e-05,
+      "loss": 1.0156,
+      "step": 3353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8634323352582372e-05,
+      "loss": 0.9961,
+      "step": 3354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8633527951096716e-05,
+      "loss": 1.3438,
+      "step": 3355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8632732335034432e-05,
+      "loss": 1.0469,
+      "step": 3356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.86319365044153e-05,
+      "loss": 0.5234,
+      "step": 3357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8631140459259097e-05,
+      "loss": 1.1719,
+      "step": 3358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.863034419958561e-05,
+      "loss": 1.2031,
+      "step": 3359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.862954772541463e-05,
+      "loss": 0.6289,
+      "step": 3360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8628751036765946e-05,
+      "loss": 1.125,
+      "step": 3361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8627954133659362e-05,
+      "loss": 1.1641,
+      "step": 3362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.862715701611469e-05,
+      "loss": 0.4727,
+      "step": 3363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.862635968415173e-05,
+      "loss": 1.0469,
+      "step": 3364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8625562137790308e-05,
+      "loss": 1.0703,
+      "step": 3365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8624764377050245e-05,
+      "loss": 0.9766,
+      "step": 3366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.862396640195136e-05,
+      "loss": 1.125,
+      "step": 3367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.86231682125135e-05,
+      "loss": 1.6328,
+      "step": 3368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8622369808756492e-05,
+      "loss": 2.0469,
+      "step": 3369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8621571190700187e-05,
+      "loss": 2.0469,
+      "step": 3370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8620772358364424e-05,
+      "loss": 1.7109,
+      "step": 3371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8619973311769068e-05,
+      "loss": 0.9062,
+      "step": 3372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8619174050933967e-05,
+      "loss": 0.3652,
+      "step": 3373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8618374575879e-05,
+      "loss": 1.125,
+      "step": 3374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.861757488662402e-05,
+      "loss": 1.5078,
+      "step": 3375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.861677498318892e-05,
+      "loss": 0.9453,
+      "step": 3376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.861597486559357e-05,
+      "loss": 0.8477,
+      "step": 3377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8615174533857853e-05,
+      "loss": 1.0781,
+      "step": 3378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8614373988001668e-05,
+      "loss": 1.5938,
+      "step": 3379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.861357322804491e-05,
+      "loss": 0.875,
+      "step": 3380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.861277225400748e-05,
+      "loss": 1.2578,
+      "step": 3381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8611971065909285e-05,
+      "loss": 1.2891,
+      "step": 3382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8611169663770238e-05,
+      "loss": 0.3359,
+      "step": 3383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8610368047610258e-05,
+      "loss": 1.5391,
+      "step": 3384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8609566217449267e-05,
+      "loss": 1.3438,
+      "step": 3385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8608764173307195e-05,
+      "loss": 1.3906,
+      "step": 3386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8607961915203974e-05,
+      "loss": 0.8594,
+      "step": 3387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8607159443159545e-05,
+      "loss": 1.7344,
+      "step": 3388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.860635675719385e-05,
+      "loss": 1.0078,
+      "step": 3389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.860555385732684e-05,
+      "loss": 1.5469,
+      "step": 3390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.860475074357847e-05,
+      "loss": 0.6797,
+      "step": 3391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.86039474159687e-05,
+      "loss": 1.0703,
+      "step": 3392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8603143874517502e-05,
+      "loss": 1.75,
+      "step": 3393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8602340119244837e-05,
+      "loss": 0.7266,
+      "step": 3394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8601536150170688e-05,
+      "loss": 0.5898,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8600731967315035e-05,
+      "loss": 1.0469,
+      "step": 3396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8599927570697866e-05,
+      "loss": 1.3906,
+      "step": 3397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8599122960339172e-05,
+      "loss": 1.2891,
+      "step": 3398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.859831813625895e-05,
+      "loss": 1.0234,
+      "step": 3399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8597513098477207e-05,
+      "loss": 0.8633,
+      "step": 3400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8596707847013947e-05,
+      "loss": 1.6016,
+      "step": 3401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8595902381889187e-05,
+      "loss": 1.2188,
+      "step": 3402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8595096703122944e-05,
+      "loss": 0.4512,
+      "step": 3403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.859429081073524e-05,
+      "loss": 0.8594,
+      "step": 3404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8593484704746107e-05,
+      "loss": 1.5781,
+      "step": 3405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8592678385175584e-05,
+      "loss": 1.6172,
+      "step": 3406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.85918718520437e-05,
+      "loss": 1.3984,
+      "step": 3407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8591065105370512e-05,
+      "loss": 0.7109,
+      "step": 3408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8590258145176062e-05,
+      "loss": 0.4258,
+      "step": 3409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8589450971480416e-05,
+      "loss": 1.8516,
+      "step": 3410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8588643584303626e-05,
+      "loss": 0.9727,
+      "step": 3411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8587835983665763e-05,
+      "loss": 0.5078,
+      "step": 3412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.85870281695869e-05,
+      "loss": 0.7695,
+      "step": 3413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8586220142087114e-05,
+      "loss": 0.8281,
+      "step": 3414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8585411901186484e-05,
+      "loss": 0.4688,
+      "step": 3415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8584603446905096e-05,
+      "loss": 1.3828,
+      "step": 3416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8583794779263054e-05,
+      "loss": 1.2891,
+      "step": 3417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8582985898280447e-05,
+      "loss": 0.5977,
+      "step": 3418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8582176803977382e-05,
+      "loss": 1.1641,
+      "step": 3419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8581367496373966e-05,
+      "loss": 1.9141,
+      "step": 3420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.858055797549032e-05,
+      "loss": 0.8594,
+      "step": 3421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8579748241346557e-05,
+      "loss": 1.2109,
+      "step": 3422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.85789382939628e-05,
+      "loss": 0.1836,
+      "step": 3423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.857812813335919e-05,
+      "loss": 1.8984,
+      "step": 3424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8577317759555852e-05,
+      "loss": 1.6172,
+      "step": 3425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8576507172572934e-05,
+      "loss": 1.3984,
+      "step": 3426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8575696372430576e-05,
+      "loss": 1.5859,
+      "step": 3427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8574885359148935e-05,
+      "loss": 1.4609,
+      "step": 3428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8574074132748167e-05,
+      "loss": 1.7188,
+      "step": 3429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8573262693248427e-05,
+      "loss": 0.9258,
+      "step": 3430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8572451040669892e-05,
+      "loss": 1.4453,
+      "step": 3431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.857163917503273e-05,
+      "loss": 1.2891,
+      "step": 3432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8570827096357122e-05,
+      "loss": 0.7969,
+      "step": 3433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8570014804663246e-05,
+      "loss": 1.0156,
+      "step": 3434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8569202299971298e-05,
+      "loss": 1.3594,
+      "step": 3435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8568389582301463e-05,
+      "loss": 1.3594,
+      "step": 3436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8567576651673945e-05,
+      "loss": 0.7227,
+      "step": 3437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.856676350810895e-05,
+      "loss": 0.5469,
+      "step": 3438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8565950151626685e-05,
+      "loss": 0.7969,
+      "step": 3439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8565136582247366e-05,
+      "loss": 1.0938,
+      "step": 3440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8564322799991215e-05,
+      "loss": 0.8477,
+      "step": 3441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.856350880487846e-05,
+      "loss": 1.1953,
+      "step": 3442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.856269459692932e-05,
+      "loss": 1.0625,
+      "step": 3443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8561880176164043e-05,
+      "loss": 1.0391,
+      "step": 3444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8561065542602867e-05,
+      "loss": 1.3984,
+      "step": 3445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.856025069626604e-05,
+      "loss": 0.7148,
+      "step": 3446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.855943563717381e-05,
+      "loss": 1.1797,
+      "step": 3447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8558620365346435e-05,
+      "loss": 0.9102,
+      "step": 3448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8557804880804183e-05,
+      "loss": 1.3359,
+      "step": 3449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.855698918356732e-05,
+      "loss": 0.8203,
+      "step": 3450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8556173273656118e-05,
+      "loss": 0.9414,
+      "step": 3451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.855535715109085e-05,
+      "loss": 0.5273,
+      "step": 3452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.855454081589181e-05,
+      "loss": 0.416,
+      "step": 3453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8553724268079284e-05,
+      "loss": 1.0703,
+      "step": 3454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.855290750767356e-05,
+      "loss": 0.9883,
+      "step": 3455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8552090534694943e-05,
+      "loss": 0.3457,
+      "step": 3456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8551273349163738e-05,
+      "loss": 1.2656,
+      "step": 3457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8550455951100254e-05,
+      "loss": 0.8828,
+      "step": 3458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.854963834052481e-05,
+      "loss": 0.8828,
+      "step": 3459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8548820517457722e-05,
+      "loss": 1.5156,
+      "step": 3460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8548002481919315e-05,
+      "loss": 1.3438,
+      "step": 3461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8547184233929928e-05,
+      "loss": 0.7617,
+      "step": 3462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.854636577350989e-05,
+      "loss": 1.4375,
+      "step": 3463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8545547100679545e-05,
+      "loss": 1.7734,
+      "step": 3464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8544728215459244e-05,
+      "loss": 1.9922,
+      "step": 3465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8543909117869336e-05,
+      "loss": 1.25,
+      "step": 3466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8543089807930178e-05,
+      "loss": 1.2812,
+      "step": 3467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8542270285662134e-05,
+      "loss": 1.5234,
+      "step": 3468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8541450551085573e-05,
+      "loss": 0.4824,
+      "step": 3469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.854063060422087e-05,
+      "loss": 1.3906,
+      "step": 3470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8539810445088396e-05,
+      "loss": 1.4688,
+      "step": 3471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.853899007370855e-05,
+      "loss": 1.125,
+      "step": 3472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8538169490101703e-05,
+      "loss": 1.1328,
+      "step": 3473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8537348694288266e-05,
+      "loss": 0.3789,
+      "step": 3474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8536527686288627e-05,
+      "loss": 1.6172,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8535706466123202e-05,
+      "loss": 1.0,
+      "step": 3476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8534885033812393e-05,
+      "loss": 1.3359,
+      "step": 3477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.853406338937662e-05,
+      "loss": 0.5859,
+      "step": 3478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.85332415328363e-05,
+      "loss": 1.1797,
+      "step": 3479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8532419464211868e-05,
+      "loss": 0.5977,
+      "step": 3480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.853159718352374e-05,
+      "loss": 0.6289,
+      "step": 3481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8530774690792368e-05,
+      "loss": 0.8398,
+      "step": 3482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.852995198603819e-05,
+      "loss": 0.668,
+      "step": 3483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.852912906928165e-05,
+      "loss": 0.5195,
+      "step": 3484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8528305940543203e-05,
+      "loss": 0.4336,
+      "step": 3485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8527482599843308e-05,
+      "loss": 0.9102,
+      "step": 3486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8526659047202423e-05,
+      "loss": 0.8672,
+      "step": 3487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8525835282641024e-05,
+      "loss": 0.2578,
+      "step": 3488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.852501130617958e-05,
+      "loss": 0.9844,
+      "step": 3489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.852418711783857e-05,
+      "loss": 0.5625,
+      "step": 3490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8523362717638482e-05,
+      "loss": 1.4531,
+      "step": 3491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.85225381055998e-05,
+      "loss": 1.0469,
+      "step": 3492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8521713281743024e-05,
+      "loss": 1.2812,
+      "step": 3493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.852088824608865e-05,
+      "loss": 0.7344,
+      "step": 3494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8520062998657183e-05,
+      "loss": 0.7227,
+      "step": 3495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8519237539469143e-05,
+      "loss": 1.5078,
+      "step": 3496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8518411868545028e-05,
+      "loss": 1.1875,
+      "step": 3497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.851758598590538e-05,
+      "loss": 0.9648,
+      "step": 3498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.851675989157071e-05,
+      "loss": 0.707,
+      "step": 3499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8515933585561554e-05,
+      "loss": 0.9883,
+      "step": 3500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8515107067898453e-05,
+      "loss": 1.4297,
+      "step": 3501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.851428033860194e-05,
+      "loss": 0.9922,
+      "step": 3502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.851345339769257e-05,
+      "loss": 1.1875,
+      "step": 3503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.85126262451909e-05,
+      "loss": 0.9805,
+      "step": 3504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8511798881117474e-05,
+      "loss": 1.8125,
+      "step": 3505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8510971305492868e-05,
+      "loss": 0.5742,
+      "step": 3506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8510143518337643e-05,
+      "loss": 0.6758,
+      "step": 3507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8509315519672376e-05,
+      "loss": 1.1094,
+      "step": 3508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8508487309517645e-05,
+      "loss": 0.582,
+      "step": 3509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8507658887894034e-05,
+      "loss": 1.3828,
+      "step": 3510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8506830254822133e-05,
+      "loss": 0.6602,
+      "step": 3511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.850600141032254e-05,
+      "loss": 0.6211,
+      "step": 3512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8505172354415848e-05,
+      "loss": 1.1719,
+      "step": 3513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8504343087122667e-05,
+      "loss": 0.707,
+      "step": 3514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8503513608463606e-05,
+      "loss": 0.9453,
+      "step": 3515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.850268391845928e-05,
+      "loss": 1.1094,
+      "step": 3516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.850185401713031e-05,
+      "loss": 0.459,
+      "step": 3517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8501023904497327e-05,
+      "loss": 1.3516,
+      "step": 3518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.850019358058096e-05,
+      "loss": 0.6914,
+      "step": 3519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.849936304540184e-05,
+      "loss": 0.7305,
+      "step": 3520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8498532298980618e-05,
+      "loss": 0.7266,
+      "step": 3521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8497701341337934e-05,
+      "loss": 0.7148,
+      "step": 3522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8496870172494445e-05,
+      "loss": 1.2891,
+      "step": 3523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8496038792470806e-05,
+      "loss": 0.3848,
+      "step": 3524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8495207201287684e-05,
+      "loss": 1.9609,
+      "step": 3525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8494375398965744e-05,
+      "loss": 1.0625,
+      "step": 3526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8493543385525656e-05,
+      "loss": 1.7812,
+      "step": 3527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8492711160988105e-05,
+      "loss": 0.3516,
+      "step": 3528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8491878725373776e-05,
+      "loss": 1.3125,
+      "step": 3529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8491046078703353e-05,
+      "loss": 1.4531,
+      "step": 3530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8490213220997527e-05,
+      "loss": 0.6133,
+      "step": 3531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.848938015227701e-05,
+      "loss": 1.8047,
+      "step": 3532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.84885468725625e-05,
+      "loss": 0.6797,
+      "step": 3533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8487713381874706e-05,
+      "loss": 1.6406,
+      "step": 3534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8486879680234345e-05,
+      "loss": 1.2188,
+      "step": 3535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.848604576766214e-05,
+      "loss": 1.4062,
+      "step": 3536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.848521164417881e-05,
+      "loss": 0.8906,
+      "step": 3537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.848437730980509e-05,
+      "loss": 1.2422,
+      "step": 3538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8483542764561725e-05,
+      "loss": 0.3203,
+      "step": 3539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8482708008469445e-05,
+      "loss": 1.8125,
+      "step": 3540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8481873041549002e-05,
+      "loss": 0.9844,
+      "step": 3541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8481037863821143e-05,
+      "loss": 1.2109,
+      "step": 3542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.848020247530663e-05,
+      "loss": 1.3906,
+      "step": 3543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.847936687602623e-05,
+      "loss": 0.9453,
+      "step": 3544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8478531066000703e-05,
+      "loss": 1.7578,
+      "step": 3545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.847769504525083e-05,
+      "loss": 0.8477,
+      "step": 3546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8476858813797374e-05,
+      "loss": 1.3906,
+      "step": 3547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8476022371661138e-05,
+      "loss": 1.1328,
+      "step": 3548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.84751857188629e-05,
+      "loss": 1.2188,
+      "step": 3549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.847434885542345e-05,
+      "loss": 0.5664,
+      "step": 3550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8473511781363603e-05,
+      "loss": 0.8008,
+      "step": 3551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8472674496704148e-05,
+      "loss": 0.4199,
+      "step": 3552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.84718370014659e-05,
+      "loss": 1.1406,
+      "step": 3553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8470999295669675e-05,
+      "loss": 0.7734,
+      "step": 3554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8470161379336297e-05,
+      "loss": 1.1719,
+      "step": 3555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8469323252486585e-05,
+      "loss": 0.9531,
+      "step": 3556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.846848491514137e-05,
+      "loss": 1.2109,
+      "step": 3557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.846764636732149e-05,
+      "loss": 0.5195,
+      "step": 3558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.846680760904779e-05,
+      "loss": 0.9297,
+      "step": 3559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8465968640341107e-05,
+      "loss": 0.6562,
+      "step": 3560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.84651294612223e-05,
+      "loss": 0.9102,
+      "step": 3561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8464290071712226e-05,
+      "loss": 0.6719,
+      "step": 3562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8463450471831745e-05,
+      "loss": 0.7031,
+      "step": 3563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8462610661601723e-05,
+      "loss": 0.2852,
+      "step": 3564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8461770641043033e-05,
+      "loss": 0.9688,
+      "step": 3565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8460930410176555e-05,
+      "loss": 1.5078,
+      "step": 3566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.846008996902317e-05,
+      "loss": 1.3203,
+      "step": 3567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8459249317603767e-05,
+      "loss": 0.7773,
+      "step": 3568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.845840845593924e-05,
+      "loss": 1.375,
+      "step": 3569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8457567384050485e-05,
+      "loss": 1.1484,
+      "step": 3570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.845672610195841e-05,
+      "loss": 0.3965,
+      "step": 3571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8455884609683917e-05,
+      "loss": 0.2236,
+      "step": 3572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8455042907247934e-05,
+      "loss": 1.1094,
+      "step": 3573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8454200994671366e-05,
+      "loss": 0.5703,
+      "step": 3574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8453358871975144e-05,
+      "loss": 1.0625,
+      "step": 3575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8452516539180202e-05,
+      "loss": 0.6445,
+      "step": 3576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8451673996307465e-05,
+      "loss": 1.2969,
+      "step": 3577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.845083124337788e-05,
+      "loss": 0.4082,
+      "step": 3578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8449988280412397e-05,
+      "loss": 0.8438,
+      "step": 3579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.844914510743196e-05,
+      "loss": 1.3672,
+      "step": 3580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8448301724457526e-05,
+      "loss": 0.5703,
+      "step": 3581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8447458131510055e-05,
+      "loss": 0.7891,
+      "step": 3582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.844661432861052e-05,
+      "loss": 0.6016,
+      "step": 3583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8445770315779887e-05,
+      "loss": 1.1797,
+      "step": 3584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8444926093039134e-05,
+      "loss": 0.9023,
+      "step": 3585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8444081660409244e-05,
+      "loss": 0.7031,
+      "step": 3586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8443237017911207e-05,
+      "loss": 0.7578,
+      "step": 3587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.844239216556601e-05,
+      "loss": 1.4531,
+      "step": 3588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8441547103394654e-05,
+      "loss": 1.1328,
+      "step": 3589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8440701831418143e-05,
+      "loss": 1.8359,
+      "step": 3590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8439856349657487e-05,
+      "loss": 0.7852,
+      "step": 3591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8439010658133695e-05,
+      "loss": 0.8594,
+      "step": 3592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.843816475686778e-05,
+      "loss": 1.1016,
+      "step": 3593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8437318645880782e-05,
+      "loss": 1.1406,
+      "step": 3594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8436472325193718e-05,
+      "loss": 0.5547,
+      "step": 3595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.843562579482763e-05,
+      "loss": 1.3438,
+      "step": 3596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8434779054803548e-05,
+      "loss": 1.1953,
+      "step": 3597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8433932105142526e-05,
+      "loss": 0.8047,
+      "step": 3598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8433084945865606e-05,
+      "loss": 1.0625,
+      "step": 3599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8432237576993854e-05,
+      "loss": 0.793,
+      "step": 3600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8431389998548318e-05,
+      "loss": 1.0078,
+      "step": 3601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8430542210550067e-05,
+      "loss": 0.9766,
+      "step": 3602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8429694213020178e-05,
+      "loss": 1.0156,
+      "step": 3603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8428846005979725e-05,
+      "loss": 0.6875,
+      "step": 3604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8427997589449783e-05,
+      "loss": 0.5547,
+      "step": 3605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.842714896345145e-05,
+      "loss": 0.3496,
+      "step": 3606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8426300128005804e-05,
+      "loss": 1.5547,
+      "step": 3607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.842545108313395e-05,
+      "loss": 0.6172,
+      "step": 3608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8424601828856988e-05,
+      "loss": 1.0078,
+      "step": 3609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.842375236519603e-05,
+      "loss": 1.3281,
+      "step": 3610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.842290269217218e-05,
+      "loss": 1.0469,
+      "step": 3611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8422052809806557e-05,
+      "loss": 1.8125,
+      "step": 3612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.842120271812029e-05,
+      "loss": 1.4766,
+      "step": 3613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8420352417134503e-05,
+      "loss": 0.9766,
+      "step": 3614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.841950190687033e-05,
+      "loss": 0.5195,
+      "step": 3615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8418651187348912e-05,
+      "loss": 1.2812,
+      "step": 3616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8417800258591388e-05,
+      "loss": 0.8398,
+      "step": 3617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.841694912061891e-05,
+      "loss": 1.0156,
+      "step": 3618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8416097773452628e-05,
+      "loss": 1.5,
+      "step": 3619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8415246217113706e-05,
+      "loss": 0.5391,
+      "step": 3620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.841439445162331e-05,
+      "loss": 0.9297,
+      "step": 3621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.84135424770026e-05,
+      "loss": 1.3359,
+      "step": 3622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8412690293272764e-05,
+      "loss": 1.2812,
+      "step": 3623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8411837900454973e-05,
+      "loss": 1.2188,
+      "step": 3624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8410985298570413e-05,
+      "loss": 1.0625,
+      "step": 3625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.841013248764028e-05,
+      "loss": 1.5781,
+      "step": 3626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.840927946768576e-05,
+      "loss": 0.7422,
+      "step": 3627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8408426238728067e-05,
+      "loss": 1.5859,
+      "step": 3628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8407572800788394e-05,
+      "loss": 0.6445,
+      "step": 3629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.840671915388796e-05,
+      "loss": 0.7773,
+      "step": 3630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.840586529804798e-05,
+      "loss": 0.9688,
+      "step": 3631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8405011233289673e-05,
+      "loss": 1.1875,
+      "step": 3632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.840415695963427e-05,
+      "loss": 1.2422,
+      "step": 3633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8403302477103e-05,
+      "loss": 1.3047,
+      "step": 3634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.84024477857171e-05,
+      "loss": 1.2031,
+      "step": 3635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8401592885497813e-05,
+      "loss": 1.1797,
+      "step": 3636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.840073777646639e-05,
+      "loss": 0.7227,
+      "step": 3637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.839988245864408e-05,
+      "loss": 0.3418,
+      "step": 3638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8399026932052144e-05,
+      "loss": 0.6719,
+      "step": 3639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.839817119671184e-05,
+      "loss": 0.7422,
+      "step": 3640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8397315252644438e-05,
+      "loss": 1.5156,
+      "step": 3641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8396459099871212e-05,
+      "loss": 1.5078,
+      "step": 3642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8395602738413446e-05,
+      "loss": 0.7695,
+      "step": 3643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8394746168292417e-05,
+      "loss": 1.0078,
+      "step": 3644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.839388938952942e-05,
+      "loss": 0.7422,
+      "step": 3645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.839303240214574e-05,
+      "loss": 0.9297,
+      "step": 3646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8392175206162684e-05,
+      "loss": 0.9453,
+      "step": 3647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8391317801601557e-05,
+      "loss": 1.0469,
+      "step": 3648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8390460188483662e-05,
+      "loss": 0.8438,
+      "step": 3649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8389602366830325e-05,
+      "loss": 0.8477,
+      "step": 3650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8388744336662856e-05,
+      "loss": 1.625,
+      "step": 3651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8387886098002585e-05,
+      "loss": 1.4766,
+      "step": 3652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8387027650870843e-05,
+      "loss": 1.1172,
+      "step": 3653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8386168995288964e-05,
+      "loss": 0.9883,
+      "step": 3654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8385310131278288e-05,
+      "loss": 0.334,
+      "step": 3655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8384451058860163e-05,
+      "loss": 0.625,
+      "step": 3656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.838359177805594e-05,
+      "loss": 1.2578,
+      "step": 3657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8382732288886977e-05,
+      "loss": 0.4668,
+      "step": 3658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8381872591374634e-05,
+      "loss": 1.0078,
+      "step": 3659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8381012685540277e-05,
+      "loss": 0.3242,
+      "step": 3660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8380152571405278e-05,
+      "loss": 0.3223,
+      "step": 3661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8379292248991013e-05,
+      "loss": 1.6953,
+      "step": 3662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.837843171831887e-05,
+      "loss": 1.0156,
+      "step": 3663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.837757097941023e-05,
+      "loss": 1.8594,
+      "step": 3664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.837671003228649e-05,
+      "loss": 0.6562,
+      "step": 3665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8375848876969046e-05,
+      "loss": 0.4805,
+      "step": 3666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8374987513479305e-05,
+      "loss": 0.6133,
+      "step": 3667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8374125941838668e-05,
+      "loss": 1.4375,
+      "step": 3668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.837326416206855e-05,
+      "loss": 0.8984,
+      "step": 3669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8372402174190373e-05,
+      "loss": 0.3594,
+      "step": 3670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8371539978225563e-05,
+      "loss": 1.4062,
+      "step": 3671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.837067757419554e-05,
+      "loss": 1.4297,
+      "step": 3672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8369814962121746e-05,
+      "loss": 0.6875,
+      "step": 3673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8368952142025618e-05,
+      "loss": 1.5312,
+      "step": 3674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83680891139286e-05,
+      "loss": 1.1172,
+      "step": 3675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.836722587785214e-05,
+      "loss": 1.6953,
+      "step": 3676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8366362433817696e-05,
+      "loss": 1.1328,
+      "step": 3677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8365498781846725e-05,
+      "loss": 0.9805,
+      "step": 3678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8364634921960692e-05,
+      "loss": 1.6875,
+      "step": 3679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8363770854181072e-05,
+      "loss": 0.4648,
+      "step": 3680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8362906578529337e-05,
+      "loss": 1.4297,
+      "step": 3681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8362042095026966e-05,
+      "loss": 0.4648,
+      "step": 3682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.836117740369545e-05,
+      "loss": 0.6758,
+      "step": 3683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8360312504556267e-05,
+      "loss": 1.0156,
+      "step": 3684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.835944739763093e-05,
+      "loss": 1.75,
+      "step": 3685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8358582082940932e-05,
+      "loss": 1.1562,
+      "step": 3686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8357716560507782e-05,
+      "loss": 1.0078,
+      "step": 3687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8356850830352985e-05,
+      "loss": 0.6875,
+      "step": 3688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8355984892498065e-05,
+      "loss": 0.6406,
+      "step": 3689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.835511874696454e-05,
+      "loss": 0.918,
+      "step": 3690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8354252393773942e-05,
+      "loss": 1.6719,
+      "step": 3691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8353385832947797e-05,
+      "loss": 1.5,
+      "step": 3692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8352519064507647e-05,
+      "loss": 0.6797,
+      "step": 3693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.835165208847503e-05,
+      "loss": 1.9297,
+      "step": 3694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8350784904871498e-05,
+      "loss": 1.2188,
+      "step": 3695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.83499175137186e-05,
+      "loss": 1.1172,
+      "step": 3696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8349049915037896e-05,
+      "loss": 0.7656,
+      "step": 3697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8348182108850956e-05,
+      "loss": 0.8086,
+      "step": 3698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8347314095179335e-05,
+      "loss": 0.6094,
+      "step": 3699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8346445874044615e-05,
+      "loss": 0.8555,
+      "step": 3700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8345577445468376e-05,
+      "loss": 1.7578,
+      "step": 3701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8344708809472195e-05,
+      "loss": 1.0859,
+      "step": 3702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8343839966077667e-05,
+      "loss": 0.5117,
+      "step": 3703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8342970915306383e-05,
+      "loss": 0.7383,
+      "step": 3704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8342101657179942e-05,
+      "loss": 1.125,
+      "step": 3705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8341232191719952e-05,
+      "loss": 1.2656,
+      "step": 3706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8340362518948017e-05,
+      "loss": 0.9102,
+      "step": 3707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8339492638885762e-05,
+      "loss": 1.2266,
+      "step": 3708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.833862255155479e-05,
+      "loss": 0.8398,
+      "step": 3709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8337752256976743e-05,
+      "loss": 0.4453,
+      "step": 3710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.833688175517324e-05,
+      "loss": 0.5273,
+      "step": 3711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8336011046165924e-05,
+      "loss": 0.2891,
+      "step": 3712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8335140129976426e-05,
+      "loss": 1.9453,
+      "step": 3713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8334269006626403e-05,
+      "loss": 2.0469,
+      "step": 3714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8333397676137502e-05,
+      "loss": 0.9258,
+      "step": 3715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8332526138531373e-05,
+      "loss": 0.9453,
+      "step": 3716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.833165439382968e-05,
+      "loss": 0.4863,
+      "step": 3717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8330782442054093e-05,
+      "loss": 0.7188,
+      "step": 3718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8329910283226282e-05,
+      "loss": 1.4531,
+      "step": 3719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8329037917367922e-05,
+      "loss": 1.4375,
+      "step": 3720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8328165344500696e-05,
+      "loss": 1.6406,
+      "step": 3721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8327292564646286e-05,
+      "loss": 0.7383,
+      "step": 3722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8326419577826388e-05,
+      "loss": 0.832,
+      "step": 3723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8325546384062704e-05,
+      "loss": 1.5391,
+      "step": 3724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8324672983376928e-05,
+      "loss": 1.5234,
+      "step": 3725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.832379937579077e-05,
+      "loss": 0.9102,
+      "step": 3726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8322925561325944e-05,
+      "loss": 1.3594,
+      "step": 3727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.832205154000417e-05,
+      "loss": 0.9609,
+      "step": 3728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8321177311847164e-05,
+      "loss": 1.3281,
+      "step": 3729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.832030287687666e-05,
+      "loss": 1.2891,
+      "step": 3730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8319428235114387e-05,
+      "loss": 1.4219,
+      "step": 3731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.831855338658209e-05,
+      "loss": 0.6211,
+      "step": 3732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.83176783313015e-05,
+      "loss": 1.25,
+      "step": 3733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8316803069294375e-05,
+      "loss": 0.7344,
+      "step": 3734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.831592760058247e-05,
+      "loss": 1.4375,
+      "step": 3735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8315051925187535e-05,
+      "loss": 1.3984,
+      "step": 3736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.831417604313134e-05,
+      "loss": 1.1094,
+      "step": 3737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8313299954435658e-05,
+      "loss": 1.0703,
+      "step": 3738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8312423659122256e-05,
+      "loss": 0.832,
+      "step": 3739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8311547157212914e-05,
+      "loss": 0.8906,
+      "step": 3740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.831067044872942e-05,
+      "loss": 1.0859,
+      "step": 3741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8309793533693562e-05,
+      "loss": 1.3906,
+      "step": 3742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8308916412127137e-05,
+      "loss": 1.0469,
+      "step": 3743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8308039084051938e-05,
+      "loss": 1.1094,
+      "step": 3744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8307161549489775e-05,
+      "loss": 1.2109,
+      "step": 3745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8306283808462463e-05,
+      "loss": 1.2109,
+      "step": 3746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8305405860991806e-05,
+      "loss": 0.9375,
+      "step": 3747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8304527707099633e-05,
+      "loss": 0.9023,
+      "step": 3748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8303649346807767e-05,
+      "loss": 1.0234,
+      "step": 3749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8302770780138037e-05,
+      "loss": 0.7383,
+      "step": 3750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.830189200711228e-05,
+      "loss": 1.6562,
+      "step": 3751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.830101302775234e-05,
+      "loss": 1.7812,
+      "step": 3752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8300133842080056e-05,
+      "loss": 0.5703,
+      "step": 3753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8299254450117286e-05,
+      "loss": 1.1875,
+      "step": 3754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8298374851885887e-05,
+      "loss": 1.2109,
+      "step": 3755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.829749504740771e-05,
+      "loss": 2.0781,
+      "step": 3756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8296615036704636e-05,
+      "loss": 1.9766,
+      "step": 3757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8295734819798525e-05,
+      "loss": 1.6562,
+      "step": 3758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.829485439671126e-05,
+      "loss": 0.2734,
+      "step": 3759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.829397376746472e-05,
+      "loss": 0.5742,
+      "step": 3760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8293092932080792e-05,
+      "loss": 0.6484,
+      "step": 3761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8292211890581372e-05,
+      "loss": 0.8086,
+      "step": 3762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8291330642988356e-05,
+      "loss": 0.5586,
+      "step": 3763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.829044918932364e-05,
+      "loss": 0.5586,
+      "step": 3764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8289567529609137e-05,
+      "loss": 0.625,
+      "step": 3765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8288685663866764e-05,
+      "loss": 1.625,
+      "step": 3766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.828780359211843e-05,
+      "loss": 0.6328,
+      "step": 3767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.828692131438606e-05,
+      "loss": 0.8828,
+      "step": 3768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.828603883069159e-05,
+      "loss": 0.5703,
+      "step": 3769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8285156141056944e-05,
+      "loss": 2.1719,
+      "step": 3770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8284273245504063e-05,
+      "loss": 1.9609,
+      "step": 3771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.828339014405489e-05,
+      "loss": 1.6875,
+      "step": 3772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8282506836731376e-05,
+      "loss": 0.1729,
+      "step": 3773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.828162332355547e-05,
+      "loss": 1.6641,
+      "step": 3774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.828073960454914e-05,
+      "loss": 1.0391,
+      "step": 3775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8279855679734336e-05,
+      "loss": 0.9062,
+      "step": 3776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8278971549133038e-05,
+      "loss": 0.6836,
+      "step": 3777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8278087212767217e-05,
+      "loss": 1.4766,
+      "step": 3778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.827720267065885e-05,
+      "loss": 1.3047,
+      "step": 3779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8276317922829924e-05,
+      "loss": 0.9961,
+      "step": 3780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8275432969302426e-05,
+      "loss": 0.6445,
+      "step": 3781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.827454781009835e-05,
+      "loss": 1.1328,
+      "step": 3782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.82736624452397e-05,
+      "loss": 0.6172,
+      "step": 3783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.827277687474848e-05,
+      "loss": 0.375,
+      "step": 3784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8271891098646696e-05,
+      "loss": 0.9531,
+      "step": 3785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8271005116956363e-05,
+      "loss": 1.0156,
+      "step": 3786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8270118929699506e-05,
+      "loss": 1.0391,
+      "step": 3787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8269232536898148e-05,
+      "loss": 0.5312,
+      "step": 3788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8268345938574318e-05,
+      "loss": 1.0469,
+      "step": 3789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.826745913475005e-05,
+      "loss": 0.4551,
+      "step": 3790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8266572125447388e-05,
+      "loss": 0.8594,
+      "step": 3791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8265684910688378e-05,
+      "loss": 1.1797,
+      "step": 3792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8264797490495065e-05,
+      "loss": 1.1641,
+      "step": 3793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.826390986488951e-05,
+      "loss": 1.1016,
+      "step": 3794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8263022033893774e-05,
+      "loss": 0.7734,
+      "step": 3795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8262133997529922e-05,
+      "loss": 1.5234,
+      "step": 3796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8261245755820022e-05,
+      "loss": 1.6328,
+      "step": 3797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8260357308786157e-05,
+      "loss": 0.6523,
+      "step": 3798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8259468656450398e-05,
+      "loss": 1.1172,
+      "step": 3799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8258579798834842e-05,
+      "loss": 0.9219,
+      "step": 3800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8257690735961577e-05,
+      "loss": 1.2812,
+      "step": 3801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.82568014678527e-05,
+      "loss": 1.0625,
+      "step": 3802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8255911994530307e-05,
+      "loss": 0.9609,
+      "step": 3803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8255022316016514e-05,
+      "loss": 0.7344,
+      "step": 3804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.825413243233343e-05,
+      "loss": 0.4473,
+      "step": 3805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8253242343503165e-05,
+      "loss": 1.6328,
+      "step": 3806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.825235204954785e-05,
+      "loss": 1.1094,
+      "step": 3807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8251461550489612e-05,
+      "loss": 1.3984,
+      "step": 3808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8250570846350576e-05,
+      "loss": 1.1641,
+      "step": 3809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8249679937152887e-05,
+      "loss": 0.5586,
+      "step": 3810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8248788822918683e-05,
+      "loss": 0.6523,
+      "step": 3811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8247897503670114e-05,
+      "loss": 1.0781,
+      "step": 3812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.824700597942933e-05,
+      "loss": 0.5586,
+      "step": 3813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.824611425021849e-05,
+      "loss": 0.7734,
+      "step": 3814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.824522231605976e-05,
+      "loss": 0.7266,
+      "step": 3815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8244330176975306e-05,
+      "loss": 0.6562,
+      "step": 3816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8243437832987298e-05,
+      "loss": 1.625,
+      "step": 3817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.824254528411792e-05,
+      "loss": 0.9727,
+      "step": 3818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8241652530389352e-05,
+      "loss": 0.4727,
+      "step": 3819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8240759571823782e-05,
+      "loss": 1.2578,
+      "step": 3820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8239866408443402e-05,
+      "loss": 0.8086,
+      "step": 3821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8238973040270415e-05,
+      "loss": 1.2891,
+      "step": 3822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.823807946732702e-05,
+      "loss": 0.7109,
+      "step": 3823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.823718568963543e-05,
+      "loss": 1.6172,
+      "step": 3824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8236291707217856e-05,
+      "loss": 1.3906,
+      "step": 3825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.823539752009652e-05,
+      "loss": 0.875,
+      "step": 3826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.823450312829364e-05,
+      "loss": 1.1953,
+      "step": 3827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8233608531831452e-05,
+      "loss": 1.4062,
+      "step": 3828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8232713730732188e-05,
+      "loss": 0.3906,
+      "step": 3829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8231818725018085e-05,
+      "loss": 1.4453,
+      "step": 3830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.823092351471139e-05,
+      "loss": 1.0234,
+      "step": 3831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.823002809983435e-05,
+      "loss": 1.0078,
+      "step": 3832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.822913248040922e-05,
+      "loss": 0.7578,
+      "step": 3833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8228236656458264e-05,
+      "loss": 1.3203,
+      "step": 3834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8227340628003736e-05,
+      "loss": 0.7344,
+      "step": 3835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.822644439506792e-05,
+      "loss": 1.8125,
+      "step": 3836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.822554795767308e-05,
+      "loss": 1.0156,
+      "step": 3837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.82246513158415e-05,
+      "loss": 0.8711,
+      "step": 3838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8223754469595467e-05,
+      "loss": 0.4648,
+      "step": 3839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8222857418957265e-05,
+      "loss": 1.3828,
+      "step": 3840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8221960163949193e-05,
+      "loss": 1.125,
+      "step": 3841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8221062704593556e-05,
+      "loss": 1.2109,
+      "step": 3842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8220165040912653e-05,
+      "loss": 0.9531,
+      "step": 3843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8219267172928793e-05,
+      "loss": 1.125,
+      "step": 3844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8218369100664296e-05,
+      "loss": 1.0781,
+      "step": 3845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8217470824141483e-05,
+      "loss": 1.1016,
+      "step": 3846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8216572343382675e-05,
+      "loss": 1.1016,
+      "step": 3847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8215673658410208e-05,
+      "loss": 1.2656,
+      "step": 3848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.821477476924641e-05,
+      "loss": 0.9492,
+      "step": 3849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8213875675913634e-05,
+      "loss": 1.1562,
+      "step": 3850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8212976378434217e-05,
+      "loss": 1.4062,
+      "step": 3851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.821207687683051e-05,
+      "loss": 1.1094,
+      "step": 3852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8211177171124874e-05,
+      "loss": 0.9102,
+      "step": 3853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8210277261339664e-05,
+      "loss": 1.0469,
+      "step": 3854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.820937714749725e-05,
+      "loss": 0.9141,
+      "step": 3855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8208476829620004e-05,
+      "loss": 0.5117,
+      "step": 3856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8207576307730305e-05,
+      "loss": 2.3438,
+      "step": 3857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8206675581850528e-05,
+      "loss": 1.3828,
+      "step": 3858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8205774652003062e-05,
+      "loss": 1.0391,
+      "step": 3859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8204873518210297e-05,
+      "loss": 0.7617,
+      "step": 3860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8203972180494634e-05,
+      "loss": 1.7656,
+      "step": 3861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8203070638878473e-05,
+      "loss": 1.5391,
+      "step": 3862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.820216889338422e-05,
+      "loss": 1.1953,
+      "step": 3863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8201266944034285e-05,
+      "loss": 0.8438,
+      "step": 3864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.820036479085109e-05,
+      "loss": 0.5625,
+      "step": 3865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.819946243385705e-05,
+      "loss": 1.0312,
+      "step": 3866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.81985598730746e-05,
+      "loss": 1.5078,
+      "step": 3867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8197657108526163e-05,
+      "loss": 1.2734,
+      "step": 3868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8196754140234184e-05,
+      "loss": 0.832,
+      "step": 3869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8195850968221105e-05,
+      "loss": 1.5625,
+      "step": 3870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8194947592509368e-05,
+      "loss": 1.2344,
+      "step": 3871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.819404401312143e-05,
+      "loss": 1.1172,
+      "step": 3872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8193140230079746e-05,
+      "loss": 1.1172,
+      "step": 3873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.819223624340678e-05,
+      "loss": 1.2734,
+      "step": 3874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8191332053124996e-05,
+      "loss": 1.2109,
+      "step": 3875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8190427659256875e-05,
+      "loss": 0.7461,
+      "step": 3876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8189523061824883e-05,
+      "loss": 1.2344,
+      "step": 3877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.818861826085151e-05,
+      "loss": 1.2188,
+      "step": 3878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.818771325635925e-05,
+      "loss": 1.1719,
+      "step": 3879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.818680804837058e-05,
+      "loss": 1.0312,
+      "step": 3880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.818590263690801e-05,
+      "loss": 1.0703,
+      "step": 3881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.818499702199404e-05,
+      "loss": 1.1406,
+      "step": 3882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8184091203651176e-05,
+      "loss": 0.8711,
+      "step": 3883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8183185181901933e-05,
+      "loss": 1.7344,
+      "step": 3884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8182278956768827e-05,
+      "loss": 0.8672,
+      "step": 3885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8181372528274387e-05,
+      "loss": 1.2109,
+      "step": 3886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8180465896441135e-05,
+      "loss": 0.7383,
+      "step": 3887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8179559061291608e-05,
+      "loss": 0.707,
+      "step": 3888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8178652022848342e-05,
+      "loss": 0.7617,
+      "step": 3889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8177744781133883e-05,
+      "loss": 0.5898,
+      "step": 3890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8176837336170775e-05,
+      "loss": 0.9727,
+      "step": 3891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8175929687981575e-05,
+      "loss": 0.9805,
+      "step": 3892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8175021836588843e-05,
+      "loss": 1.4219,
+      "step": 3893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8174113782015137e-05,
+      "loss": 1.0,
+      "step": 3894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8173205524283034e-05,
+      "loss": 1.4531,
+      "step": 3895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8172297063415097e-05,
+      "loss": 1.4531,
+      "step": 3896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8171388399433915e-05,
+      "loss": 0.9141,
+      "step": 3897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8170479532362065e-05,
+      "loss": 1.5234,
+      "step": 3898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.816957046222214e-05,
+      "loss": 0.7461,
+      "step": 3899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8168661189036735e-05,
+      "loss": 0.5859,
+      "step": 3900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8167751712828442e-05,
+      "loss": 1.6094,
+      "step": 3901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.816684203361987e-05,
+      "loss": 1.7109,
+      "step": 3902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.816593215143363e-05,
+      "loss": 1.1172,
+      "step": 3903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.816502206629233e-05,
+      "loss": 0.6797,
+      "step": 3904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8164111778218596e-05,
+      "loss": 1.25,
+      "step": 3905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8163201287235048e-05,
+      "loss": 1.2031,
+      "step": 3906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8162290593364313e-05,
+      "loss": 0.4395,
+      "step": 3907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.816137969662903e-05,
+      "loss": 0.4531,
+      "step": 3908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8160468597051836e-05,
+      "loss": 0.4297,
+      "step": 3909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.815955729465538e-05,
+      "loss": 1.4688,
+      "step": 3910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8158645789462302e-05,
+      "loss": 0.7109,
+      "step": 3911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8157734081495266e-05,
+      "loss": 1.8672,
+      "step": 3912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8156822170776925e-05,
+      "loss": 0.7891,
+      "step": 3913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8155910057329945e-05,
+      "loss": 0.8945,
+      "step": 3914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8154997741176997e-05,
+      "loss": 1.5078,
+      "step": 3915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8154085222340757e-05,
+      "loss": 1.1328,
+      "step": 3916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.81531725008439e-05,
+      "loss": 1.6172,
+      "step": 3917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8152259576709115e-05,
+      "loss": 1.375,
+      "step": 3918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.815134644995909e-05,
+      "loss": 0.8086,
+      "step": 3919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.815043312061652e-05,
+      "loss": 0.9961,
+      "step": 3920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8149519588704102e-05,
+      "loss": 1.6719,
+      "step": 3921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8148605854244546e-05,
+      "loss": 1.3984,
+      "step": 3922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.814769191726056e-05,
+      "loss": 2.25,
+      "step": 3923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8146777777774852e-05,
+      "loss": 0.4043,
+      "step": 3924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.814586343581015e-05,
+      "loss": 0.8672,
+      "step": 3925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.814494889138918e-05,
+      "loss": 0.9766,
+      "step": 3926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8144034144534667e-05,
+      "loss": 0.9844,
+      "step": 3927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.814311919526935e-05,
+      "loss": 0.6133,
+      "step": 3928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.814220404361596e-05,
+      "loss": 0.4531,
+      "step": 3929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8141288689597253e-05,
+      "loss": 1.2109,
+      "step": 3930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8140373133235978e-05,
+      "loss": 0.7695,
+      "step": 3931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8139457374554883e-05,
+      "loss": 1.9688,
+      "step": 3932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8138541413576733e-05,
+      "loss": 1.1797,
+      "step": 3933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8137625250324296e-05,
+      "loss": 1.2891,
+      "step": 3934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8136708884820334e-05,
+      "loss": 0.9727,
+      "step": 3935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.813579231708763e-05,
+      "loss": 0.7461,
+      "step": 3936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.813487554714896e-05,
+      "loss": 0.418,
+      "step": 3937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.813395857502711e-05,
+      "loss": 1.3906,
+      "step": 3938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8133041400744873e-05,
+      "loss": 1.0312,
+      "step": 3939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.813212402432504e-05,
+      "loss": 0.6641,
+      "step": 3940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.813120644579041e-05,
+      "loss": 0.875,
+      "step": 3941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8130288665163798e-05,
+      "loss": 1.1953,
+      "step": 3942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8129370682468006e-05,
+      "loss": 1.0938,
+      "step": 3943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.812845249772585e-05,
+      "loss": 0.9688,
+      "step": 3944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8127534110960156e-05,
+      "loss": 1.1797,
+      "step": 3945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.812661552219374e-05,
+      "loss": 1.5078,
+      "step": 3946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8125696731449444e-05,
+      "loss": 1.4766,
+      "step": 3947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8124777738750092e-05,
+      "loss": 0.8203,
+      "step": 3948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8123858544118532e-05,
+      "loss": 1.1016,
+      "step": 3949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.812293914757761e-05,
+      "loss": 1.0703,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.812201954915017e-05,
+      "loss": 1.0,
+      "step": 3951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8121099748859076e-05,
+      "loss": 1.2109,
+      "step": 3952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8120179746727183e-05,
+      "loss": 1.8047,
+      "step": 3953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8119259542777356e-05,
+      "loss": 1.1328,
+      "step": 3954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.811833913703247e-05,
+      "loss": 0.4453,
+      "step": 3955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.81174185295154e-05,
+      "loss": 0.6953,
+      "step": 3956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.811649772024902e-05,
+      "loss": 0.25,
+      "step": 3957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8115576709256226e-05,
+      "loss": 0.793,
+      "step": 3958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8114655496559897e-05,
+      "loss": 0.4688,
+      "step": 3959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.811373408218294e-05,
+      "loss": 1.3203,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8112812466148248e-05,
+      "loss": 0.4824,
+      "step": 3961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8111890648478732e-05,
+      "loss": 1.2031,
+      "step": 3962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8110968629197295e-05,
+      "loss": 1.25,
+      "step": 3963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8110046408326866e-05,
+      "loss": 0.5312,
+      "step": 3964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8109123985890354e-05,
+      "loss": 1.375,
+      "step": 3965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8108201361910683e-05,
+      "loss": 0.8867,
+      "step": 3966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.81072785364108e-05,
+      "loss": 1.0781,
+      "step": 3967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.810635550941362e-05,
+      "loss": 1.5,
+      "step": 3968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8105432280942098e-05,
+      "loss": 1.7422,
+      "step": 3969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8104508851019177e-05,
+      "loss": 0.7305,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8103585219667806e-05,
+      "loss": 0.625,
+      "step": 3971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8102661386910944e-05,
+      "loss": 0.918,
+      "step": 3972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.810173735277154e-05,
+      "loss": 1.2656,
+      "step": 3973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.810081311727258e-05,
+      "loss": 0.5273,
+      "step": 3974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8099888680437016e-05,
+      "loss": 1.1406,
+      "step": 3975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8098964042287838e-05,
+      "loss": 0.8711,
+      "step": 3976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8098039202848015e-05,
+      "loss": 1.0234,
+      "step": 3977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8097114162140543e-05,
+      "loss": 0.3555,
+      "step": 3978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8096188920188403e-05,
+      "loss": 0.8086,
+      "step": 3979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.80952634770146e-05,
+      "loss": 1.0625,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8094337832642127e-05,
+      "loss": 0.793,
+      "step": 3981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8093411987093998e-05,
+      "loss": 0.6602,
+      "step": 3982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8092485940393212e-05,
+      "loss": 0.6406,
+      "step": 3983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.80915596925628e-05,
+      "loss": 1.4219,
+      "step": 3984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8090633243625774e-05,
+      "loss": 1.3672,
+      "step": 3985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8089706593605162e-05,
+      "loss": 0.7422,
+      "step": 3986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.808877974252399e-05,
+      "loss": 1.5,
+      "step": 3987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.80878526904053e-05,
+      "loss": 1.4141,
+      "step": 3988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.808692543727213e-05,
+      "loss": 1.1094,
+      "step": 3989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8085997983147528e-05,
+      "loss": 0.7734,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.808507032805454e-05,
+      "loss": 1.2656,
+      "step": 3991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.808414247201623e-05,
+      "loss": 1.1797,
+      "step": 3992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.808321441505565e-05,
+      "loss": 0.9688,
+      "step": 3993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.808228615719587e-05,
+      "loss": 1.3047,
+      "step": 3994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.808135769845996e-05,
+      "loss": 1.4375,
+      "step": 3995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8080429038871003e-05,
+      "loss": 1.25,
+      "step": 3996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8079500178452066e-05,
+      "loss": 0.8945,
+      "step": 3997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8078571117226244e-05,
+      "loss": 0.9961,
+      "step": 3998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8077641855216624e-05,
+      "loss": 1.0312,
+      "step": 3999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8076712392446308e-05,
+      "loss": 0.8203,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8075782728938385e-05,
+      "loss": 1.0469,
+      "step": 4001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8074852864715974e-05,
+      "loss": 0.6836,
+      "step": 4002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.807392279980218e-05,
+      "loss": 1.6484,
+      "step": 4003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8072992534220115e-05,
+      "loss": 0.957,
+      "step": 4004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8072062067992905e-05,
+      "loss": 0.5508,
+      "step": 4005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8071131401143672e-05,
+      "loss": 0.4004,
+      "step": 4006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.807020053369555e-05,
+      "loss": 0.5586,
+      "step": 4007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8069269465671672e-05,
+      "loss": 0.8281,
+      "step": 4008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.806833819709518e-05,
+      "loss": 0.5664,
+      "step": 4009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.806740672798922e-05,
+      "loss": 0.4141,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8066475058376944e-05,
+      "loss": 0.7383,
+      "step": 4011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8065543188281505e-05,
+      "loss": 0.4629,
+      "step": 4012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.806461111772606e-05,
+      "loss": 1.0234,
+      "step": 4013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.806367884673378e-05,
+      "loss": 0.5977,
+      "step": 4014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.806274637532784e-05,
+      "loss": 1.2188,
+      "step": 4015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8061813703531403e-05,
+      "loss": 1.375,
+      "step": 4016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.806088083136766e-05,
+      "loss": 1.3438,
+      "step": 4017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8059947758859794e-05,
+      "loss": 1.0938,
+      "step": 4018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8059014486030987e-05,
+      "loss": 1.0156,
+      "step": 4019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.805808101290445e-05,
+      "loss": 1.0234,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.805714733950337e-05,
+      "loss": 1.1406,
+      "step": 4021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.805621346585096e-05,
+      "loss": 1.0859,
+      "step": 4022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.805527939197043e-05,
+      "loss": 0.7969,
+      "step": 4023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8054345117884988e-05,
+      "loss": 1.4844,
+      "step": 4024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8053410643617863e-05,
+      "loss": 1.25,
+      "step": 4025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8052475969192275e-05,
+      "loss": 1.3906,
+      "step": 4026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.805154109463146e-05,
+      "loss": 0.4746,
+      "step": 4027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8050606019958647e-05,
+      "loss": 0.9609,
+      "step": 4028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.804967074519708e-05,
+      "loss": 1.4688,
+      "step": 4029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.804873527037e-05,
+      "loss": 0.5352,
+      "step": 4030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8047799595500663e-05,
+      "loss": 1.0625,
+      "step": 4031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.804686372061232e-05,
+      "loss": 0.6914,
+      "step": 4032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8045927645728232e-05,
+      "loss": 1.0078,
+      "step": 4033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8044991370871665e-05,
+      "loss": 0.8867,
+      "step": 4034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.804405489606589e-05,
+      "loss": 0.6836,
+      "step": 4035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8043118221334177e-05,
+      "loss": 1.7422,
+      "step": 4036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.804218134669981e-05,
+      "loss": 1.5,
+      "step": 4037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8041244272186075e-05,
+      "loss": 1.5781,
+      "step": 4038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.804030699781626e-05,
+      "loss": 0.9375,
+      "step": 4039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.803936952361366e-05,
+      "loss": 0.4473,
+      "step": 4040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.803843184960158e-05,
+      "loss": 1.4219,
+      "step": 4041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8037493975803313e-05,
+      "loss": 0.7344,
+      "step": 4042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8036555902242182e-05,
+      "loss": 1.0781,
+      "step": 4043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.803561762894149e-05,
+      "loss": 1.6484,
+      "step": 4044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8034679155924567e-05,
+      "loss": 1.0938,
+      "step": 4045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8033740483214728e-05,
+      "loss": 1.0938,
+      "step": 4046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.803280161083531e-05,
+      "loss": 1.2891,
+      "step": 4047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8031862538809647e-05,
+      "loss": 0.8672,
+      "step": 4048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8030923267161076e-05,
+      "loss": 0.8516,
+      "step": 4049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.802998379591294e-05,
+      "loss": 1.3359,
+      "step": 4050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8029044125088592e-05,
+      "loss": 1.4688,
+      "step": 4051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8028104254711385e-05,
+      "loss": 1.4375,
+      "step": 4052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8027164184804683e-05,
+      "loss": 0.9805,
+      "step": 4053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8026223915391843e-05,
+      "loss": 1.3438,
+      "step": 4054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8025283446496235e-05,
+      "loss": 1.0312,
+      "step": 4055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8024342778141238e-05,
+      "loss": 0.9844,
+      "step": 4056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8023401910350232e-05,
+      "loss": 1.3672,
+      "step": 4057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8022460843146594e-05,
+      "loss": 0.5039,
+      "step": 4058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.802151957655372e-05,
+      "loss": 0.6133,
+      "step": 4059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8020578110594996e-05,
+      "loss": 0.6133,
+      "step": 4060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8019636445293834e-05,
+      "loss": 0.75,
+      "step": 4061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.801869458067363e-05,
+      "loss": 1.0938,
+      "step": 4062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.801775251675779e-05,
+      "loss": 0.4902,
+      "step": 4063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.801681025356973e-05,
+      "loss": 0.8242,
+      "step": 4064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8015867791132874e-05,
+      "loss": 0.7188,
+      "step": 4065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.801492512947064e-05,
+      "loss": 1.6016,
+      "step": 4066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.801398226860646e-05,
+      "loss": 1.8516,
+      "step": 4067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.801303920856377e-05,
+      "loss": 1.3359,
+      "step": 4068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8012095949366006e-05,
+      "loss": 1.8047,
+      "step": 4069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8011152491036607e-05,
+      "loss": 0.6602,
+      "step": 4070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8010208833599026e-05,
+      "loss": 1.0234,
+      "step": 4071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.800926497707672e-05,
+      "loss": 1.3047,
+      "step": 4072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.800832092149314e-05,
+      "loss": 0.9336,
+      "step": 4073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.800737666687175e-05,
+      "loss": 1.6719,
+      "step": 4074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.800643221323603e-05,
+      "loss": 1.0859,
+      "step": 4075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8005487560609444e-05,
+      "loss": 0.6328,
+      "step": 4076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8004542709015466e-05,
+      "loss": 1.1406,
+      "step": 4077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8003597658477586e-05,
+      "loss": 0.8633,
+      "step": 4078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8002652409019293e-05,
+      "loss": 0.8047,
+      "step": 4079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8001706960664074e-05,
+      "loss": 0.6055,
+      "step": 4080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8000761313435434e-05,
+      "loss": 0.9492,
+      "step": 4081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7999815467356874e-05,
+      "loss": 1.2812,
+      "step": 4082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7998869422451895e-05,
+      "loss": 0.2578,
+      "step": 4083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7997923178744022e-05,
+      "loss": 1.4297,
+      "step": 4084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7996976736256764e-05,
+      "loss": 1.1484,
+      "step": 4085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7996030095013645e-05,
+      "loss": 1.0469,
+      "step": 4086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7995083255038195e-05,
+      "loss": 0.8242,
+      "step": 4087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7994136216353946e-05,
+      "loss": 1.0625,
+      "step": 4088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7993188978984434e-05,
+      "loss": 0.5078,
+      "step": 4089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7992241542953204e-05,
+      "loss": 1.5469,
+      "step": 4090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7991293908283803e-05,
+      "loss": 0.9805,
+      "step": 4091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.799034607499978e-05,
+      "loss": 1.0312,
+      "step": 4092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7989398043124696e-05,
+      "loss": 1.375,
+      "step": 4093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.798844981268211e-05,
+      "loss": 0.9141,
+      "step": 4094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7987501383695594e-05,
+      "loss": 0.9727,
+      "step": 4095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7986552756188715e-05,
+      "loss": 1.5938,
+      "step": 4096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7985603930185055e-05,
+      "loss": 1.5,
+      "step": 4097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7984654905708195e-05,
+      "loss": 0.5117,
+      "step": 4098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7983705682781714e-05,
+      "loss": 1.2969,
+      "step": 4099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.798275626142922e-05,
+      "loss": 0.7852,
+      "step": 4100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.798180664167429e-05,
+      "loss": 1.0156,
+      "step": 4101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.798085682354054e-05,
+      "loss": 1.5547,
+      "step": 4102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7979906807051576e-05,
+      "loss": 0.9102,
+      "step": 4103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7978956592231e-05,
+      "loss": 0.7578,
+      "step": 4104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.797800617910244e-05,
+      "loss": 0.668,
+      "step": 4105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7977055567689512e-05,
+      "loss": 0.5664,
+      "step": 4106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.797610475801584e-05,
+      "loss": 0.707,
+      "step": 4107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7975153750105058e-05,
+      "loss": 0.8516,
+      "step": 4108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7974202543980802e-05,
+      "loss": 0.4219,
+      "step": 4109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7973251139666713e-05,
+      "loss": 0.3984,
+      "step": 4110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7972299537186436e-05,
+      "loss": 0.7227,
+      "step": 4111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7971347736563626e-05,
+      "loss": 1.2031,
+      "step": 4112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7970395737821933e-05,
+      "loss": 1.3828,
+      "step": 4113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7969443540985024e-05,
+      "loss": 0.7734,
+      "step": 4114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7968491146076562e-05,
+      "loss": 0.8867,
+      "step": 4115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7967538553120217e-05,
+      "loss": 1.3047,
+      "step": 4116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7966585762139664e-05,
+      "loss": 1.2656,
+      "step": 4117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7965632773158583e-05,
+      "loss": 0.8398,
+      "step": 4118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7964679586200664e-05,
+      "loss": 0.8242,
+      "step": 4119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7963726201289594e-05,
+      "loss": 0.5586,
+      "step": 4120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7962772618449065e-05,
+      "loss": 0.6914,
+      "step": 4121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.796181883770279e-05,
+      "loss": 1.4375,
+      "step": 4122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7960864859074457e-05,
+      "loss": 1.9766,
+      "step": 4123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7959910682587787e-05,
+      "loss": 1.0391,
+      "step": 4124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7958956308266492e-05,
+      "loss": 0.9258,
+      "step": 4125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7958001736134294e-05,
+      "loss": 0.6719,
+      "step": 4126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7957046966214915e-05,
+      "loss": 0.2871,
+      "step": 4127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7956091998532086e-05,
+      "loss": 1.4141,
+      "step": 4128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.795513683310954e-05,
+      "loss": 1.0156,
+      "step": 4129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.795418146997102e-05,
+      "loss": 1.3594,
+      "step": 4130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7953225909140267e-05,
+      "loss": 0.7812,
+      "step": 4131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.795227015064103e-05,
+      "loss": 1.1328,
+      "step": 4132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7951314194497068e-05,
+      "loss": 0.9805,
+      "step": 4133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.795035804073214e-05,
+      "loss": 1.5312,
+      "step": 4134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.794940168937e-05,
+      "loss": 1.5781,
+      "step": 4135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.794844514043443e-05,
+      "loss": 1.1562,
+      "step": 4136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7947488393949197e-05,
+      "loss": 1.0156,
+      "step": 4137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.794653144993808e-05,
+      "loss": 1.5078,
+      "step": 4138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7945574308424864e-05,
+      "loss": 0.6992,
+      "step": 4139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.794461696943334e-05,
+      "loss": 0.8711,
+      "step": 4140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7943659432987298e-05,
+      "loss": 1.0234,
+      "step": 4141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.794270169911054e-05,
+      "loss": 1.2109,
+      "step": 4142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7941743767826865e-05,
+      "loss": 1.0703,
+      "step": 4143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.794078563916008e-05,
+      "loss": 1.0703,
+      "step": 4144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7939827313134007e-05,
+      "loss": 0.668,
+      "step": 4145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7938868789772457e-05,
+      "loss": 1.1875,
+      "step": 4146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7937910069099254e-05,
+      "loss": 0.6328,
+      "step": 4147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.793695115113823e-05,
+      "loss": 1.4766,
+      "step": 4148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.793599203591321e-05,
+      "loss": 1.2344,
+      "step": 4149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7935032723448043e-05,
+      "loss": 0.7812,
+      "step": 4150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.793407321376656e-05,
+      "loss": 1.4609,
+      "step": 4151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7933113506892615e-05,
+      "loss": 0.5,
+      "step": 4152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7932153602850058e-05,
+      "loss": 0.4707,
+      "step": 4153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7931193501662747e-05,
+      "loss": 0.543,
+      "step": 4154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7930233203354545e-05,
+      "loss": 1.1016,
+      "step": 4155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.792927270794932e-05,
+      "loss": 0.6758,
+      "step": 4156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.792831201547094e-05,
+      "loss": 1.125,
+      "step": 4157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7927351125943286e-05,
+      "loss": 0.8242,
+      "step": 4158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7926390039390238e-05,
+      "loss": 0.459,
+      "step": 4159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7925428755835684e-05,
+      "loss": 0.3984,
+      "step": 4160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7924467275303513e-05,
+      "loss": 0.6797,
+      "step": 4161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.792350559781763e-05,
+      "loss": 0.2988,
+      "step": 4162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.792254372340192e-05,
+      "loss": 0.9141,
+      "step": 4163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7921581652080308e-05,
+      "loss": 1.1406,
+      "step": 4164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7920619383876688e-05,
+      "loss": 1.1797,
+      "step": 4165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.791965691881499e-05,
+      "loss": 1.7031,
+      "step": 4166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.791869425691913e-05,
+      "loss": 1.2578,
+      "step": 4167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.791773139821303e-05,
+      "loss": 0.6016,
+      "step": 4168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.791676834272062e-05,
+      "loss": 0.5508,
+      "step": 4169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7915805090465842e-05,
+      "loss": 1.3281,
+      "step": 4170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.791484164147264e-05,
+      "loss": 1.7266,
+      "step": 4171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7913877995764948e-05,
+      "loss": 0.5312,
+      "step": 4172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7912914153366724e-05,
+      "loss": 0.9766,
+      "step": 4173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7911950114301923e-05,
+      "loss": 0.5664,
+      "step": 4174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.79109858785945e-05,
+      "loss": 1.1016,
+      "step": 4175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7910021446268423e-05,
+      "loss": 0.5,
+      "step": 4176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7909056817347662e-05,
+      "loss": 1.4297,
+      "step": 4177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7908091991856195e-05,
+      "loss": 1.2578,
+      "step": 4178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7907126969817997e-05,
+      "loss": 1.0156,
+      "step": 4179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7906161751257053e-05,
+      "loss": 0.6133,
+      "step": 4180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7905196336197357e-05,
+      "loss": 1.6641,
+      "step": 4181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7904230724662896e-05,
+      "loss": 1.1016,
+      "step": 4182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7903264916677673e-05,
+      "loss": 0.4902,
+      "step": 4183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7902298912265693e-05,
+      "loss": 1.375,
+      "step": 4184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7901332711450964e-05,
+      "loss": 0.8555,
+      "step": 4185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7900366314257503e-05,
+      "loss": 1.6016,
+      "step": 4186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7899399720709324e-05,
+      "loss": 1.2891,
+      "step": 4187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.789843293083045e-05,
+      "loss": 0.9062,
+      "step": 4188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7897465944644913e-05,
+      "loss": 1.3281,
+      "step": 4189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7896498762176744e-05,
+      "loss": 0.9102,
+      "step": 4190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.789553138344998e-05,
+      "loss": 0.8008,
+      "step": 4191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.789456380848867e-05,
+      "loss": 0.6328,
+      "step": 4192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.789359603731686e-05,
+      "loss": 1.2109,
+      "step": 4193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7892628069958594e-05,
+      "loss": 1.1641,
+      "step": 4194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7891659906437944e-05,
+      "loss": 1.3359,
+      "step": 4195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7890691546778965e-05,
+      "loss": 1.6562,
+      "step": 4196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7889722991005722e-05,
+      "loss": 0.9453,
+      "step": 4197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.788875423914229e-05,
+      "loss": 1.3125,
+      "step": 4198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.788778529121275e-05,
+      "loss": 0.6094,
+      "step": 4199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.788681614724118e-05,
+      "loss": 0.8047,
+      "step": 4200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7885846807251662e-05,
+      "loss": 1.3359,
+      "step": 4201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.78848772712683e-05,
+      "loss": 1.3281,
+      "step": 4202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7883907539315184e-05,
+      "loss": 0.668,
+      "step": 4203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7882937611416414e-05,
+      "loss": 0.8047,
+      "step": 4204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.78819674875961e-05,
+      "loss": 0.7852,
+      "step": 4205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7880997167878348e-05,
+      "loss": 0.4375,
+      "step": 4206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7880026652287282e-05,
+      "loss": 1.2656,
+      "step": 4207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7879055940847016e-05,
+      "loss": 0.6328,
+      "step": 4208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.787808503358168e-05,
+      "loss": 0.4961,
+      "step": 4209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7877113930515408e-05,
+      "loss": 0.7188,
+      "step": 4210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7876142631672325e-05,
+      "loss": 0.8047,
+      "step": 4211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.787517113707658e-05,
+      "loss": 0.4023,
+      "step": 4212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.787419944675232e-05,
+      "loss": 0.5234,
+      "step": 4213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7873227560723686e-05,
+      "loss": 0.4199,
+      "step": 4214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7872255479014843e-05,
+      "loss": 0.7734,
+      "step": 4215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7871283201649942e-05,
+      "loss": 0.7812,
+      "step": 4216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7870310728653155e-05,
+      "loss": 1.0234,
+      "step": 4217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.786933806004865e-05,
+      "loss": 1.0156,
+      "step": 4218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.78683651958606e-05,
+      "loss": 1.2344,
+      "step": 4219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.786739213611319e-05,
+      "loss": 0.8789,
+      "step": 4220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7866418880830593e-05,
+      "loss": 1.3125,
+      "step": 4221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.786544543003701e-05,
+      "loss": 1.9922,
+      "step": 4222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7864471783756627e-05,
+      "loss": 1.4375,
+      "step": 4223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7863497942013643e-05,
+      "loss": 0.8438,
+      "step": 4224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.786252390483227e-05,
+      "loss": 0.8281,
+      "step": 4225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.786154967223671e-05,
+      "loss": 1.125,
+      "step": 4226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7860575244251174e-05,
+      "loss": 0.9375,
+      "step": 4227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7859600620899885e-05,
+      "loss": 0.832,
+      "step": 4228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7858625802207066e-05,
+      "loss": 1.0547,
+      "step": 4229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7857650788196944e-05,
+      "loss": 1.4688,
+      "step": 4230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7856675578893754e-05,
+      "loss": 1.0938,
+      "step": 4231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.785570017432173e-05,
+      "loss": 0.9805,
+      "step": 4232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7854724574505117e-05,
+      "loss": 1.1875,
+      "step": 4233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7853748779468158e-05,
+      "loss": 0.8047,
+      "step": 4234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.785277278923511e-05,
+      "loss": 1.0312,
+      "step": 4235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7851796603830235e-05,
+      "loss": 0.4668,
+      "step": 4236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7850820223277787e-05,
+      "loss": 0.7383,
+      "step": 4237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.784984364760203e-05,
+      "loss": 1.0234,
+      "step": 4238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7848866876827245e-05,
+      "loss": 0.5273,
+      "step": 4239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.78478899109777e-05,
+      "loss": 0.5273,
+      "step": 4240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7846912750077683e-05,
+      "loss": 0.9297,
+      "step": 4241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7845935394151478e-05,
+      "loss": 0.3379,
+      "step": 4242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7844957843223378e-05,
+      "loss": 0.8203,
+      "step": 4243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.784398009731767e-05,
+      "loss": 0.7656,
+      "step": 4244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7843002156458668e-05,
+      "loss": 0.5,
+      "step": 4245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.784202402067067e-05,
+      "loss": 1.1562,
+      "step": 4246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7841045689977986e-05,
+      "loss": 1.4844,
+      "step": 4247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7840067164404936e-05,
+      "loss": 1.0703,
+      "step": 4248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7839088443975833e-05,
+      "loss": 0.5742,
+      "step": 4249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7838109528715007e-05,
+      "loss": 1.1562,
+      "step": 4250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7837130418646787e-05,
+      "loss": 0.6953,
+      "step": 4251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7836151113795512e-05,
+      "loss": 1.8359,
+      "step": 4252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7835171614185513e-05,
+      "loss": 0.5547,
+      "step": 4253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.783419191984114e-05,
+      "loss": 1.1172,
+      "step": 4254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.783321203078674e-05,
+      "loss": 1.9453,
+      "step": 4255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7832231947046667e-05,
+      "loss": 1.4062,
+      "step": 4256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.783125166864528e-05,
+      "loss": 0.793,
+      "step": 4257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7830271195606947e-05,
+      "loss": 1.2656,
+      "step": 4258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.782929052795603e-05,
+      "loss": 1.7734,
+      "step": 4259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7828309665716903e-05,
+      "loss": 1.5234,
+      "step": 4260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.782732860891395e-05,
+      "loss": 0.8086,
+      "step": 4261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.782634735757155e-05,
+      "loss": 0.4922,
+      "step": 4262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.782536591171409e-05,
+      "loss": 0.1895,
+      "step": 4263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7824384271365966e-05,
+      "loss": 0.4785,
+      "step": 4264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.782340243655157e-05,
+      "loss": 0.582,
+      "step": 4265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7822420407295317e-05,
+      "loss": 0.9961,
+      "step": 4266
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7821438183621597e-05,
+      "loss": 0.8125,
+      "step": 4267
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7820455765554832e-05,
+      "loss": 0.8438,
+      "step": 4268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.781947315311944e-05,
+      "loss": 0.7461,
+      "step": 4269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7818490346339837e-05,
+      "loss": 1.6406,
+      "step": 4270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7817507345240454e-05,
+      "loss": 1.5156,
+      "step": 4271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7816524149845726e-05,
+      "loss": 0.6484,
+      "step": 4272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7815540760180075e-05,
+      "loss": 1.1719,
+      "step": 4273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7814557176267956e-05,
+      "loss": 1.1016,
+      "step": 4274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.781357339813381e-05,
+      "loss": 0.6758,
+      "step": 4275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.781258942580209e-05,
+      "loss": 0.6914,
+      "step": 4276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7811605259297244e-05,
+      "loss": 0.5859,
+      "step": 4277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7810620898643742e-05,
+      "loss": 0.7031,
+      "step": 4278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7809636343866044e-05,
+      "loss": 0.8555,
+      "step": 4279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.780865159498862e-05,
+      "loss": 0.7891,
+      "step": 4280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.780766665203595e-05,
+      "loss": 1.1094,
+      "step": 4281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7806681515032503e-05,
+      "loss": 1.8281,
+      "step": 4282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7805696184002773e-05,
+      "loss": 1.1484,
+      "step": 4283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7804710658971246e-05,
+      "loss": 1.2969,
+      "step": 4284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7803724939962415e-05,
+      "loss": 1.3906,
+      "step": 4285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7802739027000783e-05,
+      "loss": 0.8242,
+      "step": 4286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7801752920110845e-05,
+      "loss": 0.6133,
+      "step": 4287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7800766619317126e-05,
+      "loss": 1.0547,
+      "step": 4288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.779978012464412e-05,
+      "loss": 1.0312,
+      "step": 4289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7798793436116354e-05,
+      "loss": 0.8203,
+      "step": 4290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7797806553758358e-05,
+      "loss": 1.1406,
+      "step": 4291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.779681947759465e-05,
+      "loss": 0.9258,
+      "step": 4292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7795832207649762e-05,
+      "loss": 1.375,
+      "step": 4293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.779484474394824e-05,
+      "loss": 1.1719,
+      "step": 4294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7793857086514615e-05,
+      "loss": 0.9766,
+      "step": 4295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.779286923537345e-05,
+      "loss": 0.8633,
+      "step": 4296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.779188119054928e-05,
+      "loss": 0.9297,
+      "step": 4297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7790892952066673e-05,
+      "loss": 1.4141,
+      "step": 4298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7789904519950183e-05,
+      "loss": 1.0,
+      "step": 4299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7788915894224384e-05,
+      "loss": 0.6523,
+      "step": 4300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7787927074913844e-05,
+      "loss": 0.9141,
+      "step": 4301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7786938062043135e-05,
+      "loss": 1.5312,
+      "step": 4302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7785948855636844e-05,
+      "loss": 1.0,
+      "step": 4303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.778495945571955e-05,
+      "loss": 0.9688,
+      "step": 4304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.778396986231585e-05,
+      "loss": 1.4141,
+      "step": 4305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7782980075450335e-05,
+      "loss": 1.0703,
+      "step": 4306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.778199009514761e-05,
+      "loss": 1.1016,
+      "step": 4307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7780999921432274e-05,
+      "loss": 0.6016,
+      "step": 4308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.778000955432894e-05,
+      "loss": 0.6719,
+      "step": 4309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7779018993862222e-05,
+      "loss": 0.6094,
+      "step": 4310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7778028240056735e-05,
+      "loss": 0.9141,
+      "step": 4311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7777037292937106e-05,
+      "loss": 1.0859,
+      "step": 4312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7776046152527967e-05,
+      "loss": 0.5508,
+      "step": 4313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7775054818853953e-05,
+      "loss": 0.9492,
+      "step": 4314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7774063291939692e-05,
+      "loss": 1.9297,
+      "step": 4315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.777307157180984e-05,
+      "loss": 0.7422,
+      "step": 4316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7772079658489035e-05,
+      "loss": 0.8633,
+      "step": 4317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7771087552001937e-05,
+      "loss": 0.6523,
+      "step": 4318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.77700952523732e-05,
+      "loss": 0.2812,
+      "step": 4319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7769102759627486e-05,
+      "loss": 0.7383,
+      "step": 4320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7768110073789466e-05,
+      "loss": 1.0078,
+      "step": 4321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7767117194883807e-05,
+      "loss": 0.6758,
+      "step": 4322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7766124122935192e-05,
+      "loss": 0.7656,
+      "step": 4323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7765130857968294e-05,
+      "loss": 0.6719,
+      "step": 4324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.776413740000781e-05,
+      "loss": 0.875,
+      "step": 4325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7763143749078422e-05,
+      "loss": 1.125,
+      "step": 4326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7762149905204834e-05,
+      "loss": 0.9219,
+      "step": 4327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.776115586841174e-05,
+      "loss": 0.6562,
+      "step": 4328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.776016163872385e-05,
+      "loss": 0.9102,
+      "step": 4329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7759167216165872e-05,
+      "loss": 0.8242,
+      "step": 4330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.775817260076252e-05,
+      "loss": 1.0,
+      "step": 4331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7757177792538517e-05,
+      "loss": 1.1016,
+      "step": 4332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7756182791518593e-05,
+      "loss": 0.5234,
+      "step": 4333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7755187597727465e-05,
+      "loss": 0.668,
+      "step": 4334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.775419221118988e-05,
+      "loss": 0.4727,
+      "step": 4335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7753196631930567e-05,
+      "loss": 0.9492,
+      "step": 4336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7752200859974275e-05,
+      "loss": 1.625,
+      "step": 4337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7751204895345753e-05,
+      "loss": 0.8594,
+      "step": 4338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7750208738069755e-05,
+      "loss": 1.2656,
+      "step": 4339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7749212388171034e-05,
+      "loss": 1.1797,
+      "step": 4340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7748215845674362e-05,
+      "loss": 1.0938,
+      "step": 4341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7747219110604502e-05,
+      "loss": 1.5234,
+      "step": 4342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7746222182986225e-05,
+      "loss": 1.0078,
+      "step": 4343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.774522506284431e-05,
+      "loss": 0.9297,
+      "step": 4344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.774422775020354e-05,
+      "loss": 0.793,
+      "step": 4345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7743230245088704e-05,
+      "loss": 0.4961,
+      "step": 4346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7742232547524593e-05,
+      "loss": 1.2891,
+      "step": 4347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7741234657536e-05,
+      "loss": 0.9961,
+      "step": 4348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.774023657514773e-05,
+      "loss": 0.8555,
+      "step": 4349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7739238300384583e-05,
+      "loss": 1.0938,
+      "step": 4350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.773823983327138e-05,
+      "loss": 1.4297,
+      "step": 4351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7737241173832932e-05,
+      "loss": 1.2734,
+      "step": 4352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.773624232209406e-05,
+      "loss": 0.6133,
+      "step": 4353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7735243278079585e-05,
+      "loss": 1.3203,
+      "step": 4354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7734244041814345e-05,
+      "loss": 0.6562,
+      "step": 4355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.773324461332317e-05,
+      "loss": 0.8281,
+      "step": 4356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7732244992630897e-05,
+      "loss": 0.3613,
+      "step": 4357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7731245179762377e-05,
+      "loss": 1.0469,
+      "step": 4358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7730245174742455e-05,
+      "loss": 1.2266,
+      "step": 4359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7729244977595988e-05,
+      "loss": 1.5625,
+      "step": 4360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7728244588347833e-05,
+      "loss": 1.0938,
+      "step": 4361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7727244007022847e-05,
+      "loss": 0.875,
+      "step": 4362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772624323364591e-05,
+      "loss": 0.6172,
+      "step": 4363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7725242268241893e-05,
+      "loss": 0.4141,
+      "step": 4364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7724241110835665e-05,
+      "loss": 0.5117,
+      "step": 4365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7723239761452116e-05,
+      "loss": 1.125,
+      "step": 4366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.772223822011613e-05,
+      "loss": 0.5508,
+      "step": 4367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7721236486852604e-05,
+      "loss": 1.6406,
+      "step": 4368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7720234561686433e-05,
+      "loss": 0.8164,
+      "step": 4369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7719232444642514e-05,
+      "loss": 0.707,
+      "step": 4370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.771823013574576e-05,
+      "loss": 1.5859,
+      "step": 4371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7717227635021076e-05,
+      "loss": 0.5117,
+      "step": 4372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.771622494249338e-05,
+      "loss": 0.7734,
+      "step": 4373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7715222058187602e-05,
+      "loss": 0.9219,
+      "step": 4374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7714218982128658e-05,
+      "loss": 1.1875,
+      "step": 4375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7713215714341475e-05,
+      "loss": 0.5312,
+      "step": 4376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7712212254850996e-05,
+      "loss": 0.8281,
+      "step": 4377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7711208603682156e-05,
+      "loss": 0.668,
+      "step": 4378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7710204760859904e-05,
+      "loss": 0.7578,
+      "step": 4379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7709200726409187e-05,
+      "loss": 0.4941,
+      "step": 4380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7708196500354956e-05,
+      "loss": 1.2656,
+      "step": 4381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7707192082722176e-05,
+      "loss": 0.7852,
+      "step": 4382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.770618747353581e-05,
+      "loss": 1.8359,
+      "step": 4383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.770518267282082e-05,
+      "loss": 1.2656,
+      "step": 4384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7704177680602187e-05,
+      "loss": 1.2422,
+      "step": 4385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7703172496904882e-05,
+      "loss": 0.9414,
+      "step": 4386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7702167121753896e-05,
+      "loss": 1.6719,
+      "step": 4387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7701161555174205e-05,
+      "loss": 1.3281,
+      "step": 4388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7700155797190812e-05,
+      "loss": 0.5586,
+      "step": 4389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7699149847828706e-05,
+      "loss": 1.2969,
+      "step": 4390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7698143707112896e-05,
+      "loss": 1.4141,
+      "step": 4391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7697137375068383e-05,
+      "loss": 0.5977,
+      "step": 4392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769613085172018e-05,
+      "loss": 0.8359,
+      "step": 4393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7695124137093307e-05,
+      "loss": 0.7109,
+      "step": 4394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7694117231212778e-05,
+      "loss": 0.8711,
+      "step": 4395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769311013410362e-05,
+      "loss": 1.0625,
+      "step": 4396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7692102845790867e-05,
+      "loss": 1.1328,
+      "step": 4397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.769109536629955e-05,
+      "loss": 0.8867,
+      "step": 4398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7690087695654715e-05,
+      "loss": 0.9258,
+      "step": 4399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7689079833881398e-05,
+      "loss": 1.4844,
+      "step": 4400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7688071781004654e-05,
+      "loss": 1.8594,
+      "step": 4401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7687063537049535e-05,
+      "loss": 0.5703,
+      "step": 4402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7686055102041097e-05,
+      "loss": 1.8203,
+      "step": 4403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7685046476004412e-05,
+      "loss": 1.1016,
+      "step": 4404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7684037658964538e-05,
+      "loss": 0.5781,
+      "step": 4405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7683028650946555e-05,
+      "loss": 0.4902,
+      "step": 4406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.768201945197554e-05,
+      "loss": 0.4043,
+      "step": 4407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7681010062076574e-05,
+      "loss": 0.9414,
+      "step": 4408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7680000481274742e-05,
+      "loss": 0.875,
+      "step": 4409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7678990709595137e-05,
+      "loss": 0.793,
+      "step": 4410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767798074706286e-05,
+      "loss": 1.1719,
+      "step": 4411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7676970593703013e-05,
+      "loss": 0.4688,
+      "step": 4412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7675960249540692e-05,
+      "loss": 0.7656,
+      "step": 4413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767494971460102e-05,
+      "loss": 0.6133,
+      "step": 4414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7673938988909107e-05,
+      "loss": 0.4141,
+      "step": 4415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767292807249007e-05,
+      "loss": 0.2412,
+      "step": 4416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7671916965369044e-05,
+      "loss": 0.9492,
+      "step": 4417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.767090566757115e-05,
+      "loss": 0.8477,
+      "step": 4418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7669894179121527e-05,
+      "loss": 0.4121,
+      "step": 4419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.766888250004531e-05,
+      "loss": 1.7266,
+      "step": 4420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.766787063036765e-05,
+      "loss": 1.7812,
+      "step": 4421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.766685857011369e-05,
+      "loss": 1.9141,
+      "step": 4422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7665846319308587e-05,
+      "loss": 0.543,
+      "step": 4423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7664833877977498e-05,
+      "loss": 0.6016,
+      "step": 4424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7663821246145585e-05,
+      "loss": 1.8906,
+      "step": 4425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.766280842383802e-05,
+      "loss": 1.6875,
+      "step": 4426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.766179541107997e-05,
+      "loss": 0.2344,
+      "step": 4427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7660782207896616e-05,
+      "loss": 1.0234,
+      "step": 4428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7659768814313138e-05,
+      "loss": 0.832,
+      "step": 4429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7658755230354725e-05,
+      "loss": 2.0625,
+      "step": 4430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7657741456046568e-05,
+      "loss": 0.8633,
+      "step": 4431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.765672749141386e-05,
+      "loss": 1.3984,
+      "step": 4432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7655713336481805e-05,
+      "loss": 1.4062,
+      "step": 4433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.765469899127561e-05,
+      "loss": 0.5859,
+      "step": 4434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.765368445582048e-05,
+      "loss": 1.4688,
+      "step": 4435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7652669730141635e-05,
+      "loss": 1.0,
+      "step": 4436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7651654814264295e-05,
+      "loss": 0.8242,
+      "step": 4437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7650639708213683e-05,
+      "loss": 0.9805,
+      "step": 4438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7649624412015027e-05,
+      "loss": 1.5312,
+      "step": 4439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7648608925693564e-05,
+      "loss": 0.6602,
+      "step": 4440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7647593249274533e-05,
+      "loss": 1.2188,
+      "step": 4441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7646577382783173e-05,
+      "loss": 0.8125,
+      "step": 4442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7645561326244737e-05,
+      "loss": 0.7891,
+      "step": 4443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7644545079684472e-05,
+      "loss": 0.8164,
+      "step": 4444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7643528643127642e-05,
+      "loss": 0.9102,
+      "step": 4445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7642512016599503e-05,
+      "loss": 0.9023,
+      "step": 4446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.764149520012533e-05,
+      "loss": 1.1172,
+      "step": 4447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7640478193730392e-05,
+      "loss": 0.9062,
+      "step": 4448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7639460997439964e-05,
+      "loss": 0.5742,
+      "step": 4449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7638443611279326e-05,
+      "loss": 1.0703,
+      "step": 4450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7637426035273765e-05,
+      "loss": 1.4531,
+      "step": 4451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7636408269448575e-05,
+      "loss": 1.6641,
+      "step": 4452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7635390313829048e-05,
+      "loss": 0.7148,
+      "step": 4453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7634372168440482e-05,
+      "loss": 0.9453,
+      "step": 4454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.763335383330819e-05,
+      "loss": 0.5234,
+      "step": 4455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7632335308457473e-05,
+      "loss": 1.1953,
+      "step": 4456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7631316593913647e-05,
+      "loss": 0.3516,
+      "step": 4457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7630297689702037e-05,
+      "loss": 1.0703,
+      "step": 4458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.762927859584796e-05,
+      "loss": 0.8672,
+      "step": 4459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7628259312376745e-05,
+      "loss": 0.7852,
+      "step": 4460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.762723983931373e-05,
+      "loss": 0.5156,
+      "step": 4461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.762622017668425e-05,
+      "loss": 0.5,
+      "step": 4462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.762520032451365e-05,
+      "loss": 0.6211,
+      "step": 4463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7624180282827267e-05,
+      "loss": 0.543,
+      "step": 4464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7623160051650465e-05,
+      "loss": 0.8984,
+      "step": 4465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7622139631008597e-05,
+      "loss": 0.6328,
+      "step": 4466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.762111902092702e-05,
+      "loss": 1.2734,
+      "step": 4467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7620098221431108e-05,
+      "loss": 1.3594,
+      "step": 4468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7619077232546224e-05,
+      "loss": 1.0391,
+      "step": 4469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7618056054297752e-05,
+      "loss": 1.5312,
+      "step": 4470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7617034686711065e-05,
+      "loss": 0.6094,
+      "step": 4471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.761601312981155e-05,
+      "loss": 0.7539,
+      "step": 4472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.76149913836246e-05,
+      "loss": 0.6836,
+      "step": 4473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.76139694481756e-05,
+      "loss": 1.6484,
+      "step": 4474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.761294732348996e-05,
+      "loss": 0.832,
+      "step": 4475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7611925009593077e-05,
+      "loss": 0.5273,
+      "step": 4476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7610902506510368e-05,
+      "loss": 1.2969,
+      "step": 4477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.760987981426723e-05,
+      "loss": 1.2188,
+      "step": 4478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.76088569328891e-05,
+      "loss": 1.1719,
+      "step": 4479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7607833862401386e-05,
+      "loss": 0.9922,
+      "step": 4480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7606810602829523e-05,
+      "loss": 1.1328,
+      "step": 4481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7605787154198937e-05,
+      "loss": 0.9688,
+      "step": 4482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.760476351653507e-05,
+      "loss": 1.7891,
+      "step": 4483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.760373968986336e-05,
+      "loss": 1.5156,
+      "step": 4484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.760271567420926e-05,
+      "loss": 1.0391,
+      "step": 4485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7601691469598213e-05,
+      "loss": 1.1562,
+      "step": 4486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7600667076055673e-05,
+      "loss": 0.6914,
+      "step": 4487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7599642493607107e-05,
+      "loss": 1.0781,
+      "step": 4488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.759861772227798e-05,
+      "loss": 0.7656,
+      "step": 4489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.759759276209375e-05,
+      "loss": 1.4453,
+      "step": 4490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7596567613079906e-05,
+      "loss": 1.5547,
+      "step": 4491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.759554227526192e-05,
+      "loss": 1.0391,
+      "step": 4492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7594516748665276e-05,
+      "loss": 1.4922,
+      "step": 4493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7593491033315458e-05,
+      "loss": 0.9062,
+      "step": 4494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.759246512923797e-05,
+      "loss": 1.1016,
+      "step": 4495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7591439036458296e-05,
+      "loss": 0.6523,
+      "step": 4496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.759041275500195e-05,
+      "loss": 1.2109,
+      "step": 4497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7589386284894436e-05,
+      "loss": 1.3672,
+      "step": 4498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7588359626161263e-05,
+      "loss": 0.9805,
+      "step": 4499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7587332778827944e-05,
+      "loss": 1.2891,
+      "step": 4500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7586305742920013e-05,
+      "loss": 1.3125,
+      "step": 4501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7585278518462984e-05,
+      "loss": 0.957,
+      "step": 4502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.758425110548239e-05,
+      "loss": 1.1484,
+      "step": 4503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7583223504003768e-05,
+      "loss": 0.7656,
+      "step": 4504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.758219571405266e-05,
+      "loss": 0.9258,
+      "step": 4505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7581167735654603e-05,
+      "loss": 0.4629,
+      "step": 4506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.758013956883515e-05,
+      "loss": 1.5469,
+      "step": 4507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7579111213619865e-05,
+      "loss": 0.543,
+      "step": 4508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.757808267003429e-05,
+      "loss": 0.3262,
+      "step": 4509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7577053938104e-05,
+      "loss": 0.3398,
+      "step": 4510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7576025017854558e-05,
+      "loss": 0.8242,
+      "step": 4511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7574995909311536e-05,
+      "loss": 0.9453,
+      "step": 4512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7573966612500514e-05,
+      "loss": 1.5703,
+      "step": 4513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.757293712744707e-05,
+      "loss": 0.7969,
+      "step": 4514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.75719074541768e-05,
+      "loss": 0.582,
+      "step": 4515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7570877592715284e-05,
+      "loss": 0.3652,
+      "step": 4516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7569847543088125e-05,
+      "loss": 0.4512,
+      "step": 4517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7568817305320924e-05,
+      "loss": 0.4434,
+      "step": 4518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.756778687943928e-05,
+      "loss": 0.9727,
+      "step": 4519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.756675626546881e-05,
+      "loss": 1.4844,
+      "step": 4520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7565725463435122e-05,
+      "loss": 0.5117,
+      "step": 4521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7564694473363842e-05,
+      "loss": 1.2266,
+      "step": 4522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7563663295280593e-05,
+      "loss": 1.1875,
+      "step": 4523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7562631929211e-05,
+      "loss": 0.4395,
+      "step": 4524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.75616003751807e-05,
+      "loss": 0.6992,
+      "step": 4525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.756056863321533e-05,
+      "loss": 1.8672,
+      "step": 4526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7559536703340532e-05,
+      "loss": 1.9062,
+      "step": 4527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7558504585581953e-05,
+      "loss": 0.7734,
+      "step": 4528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.755747227996525e-05,
+      "loss": 0.3203,
+      "step": 4529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7556439786516072e-05,
+      "loss": 1.1719,
+      "step": 4530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7555407105260087e-05,
+      "loss": 1.0625,
+      "step": 4531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.755437423622296e-05,
+      "loss": 0.3242,
+      "step": 4532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7553341179430356e-05,
+      "loss": 0.9844,
+      "step": 4533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7552307934907954e-05,
+      "loss": 0.5234,
+      "step": 4534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.755127450268144e-05,
+      "loss": 0.6914,
+      "step": 4535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.755024088277649e-05,
+      "loss": 0.7578,
+      "step": 4536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7549207075218803e-05,
+      "loss": 1.2266,
+      "step": 4537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.754817308003406e-05,
+      "loss": 1.3984,
+      "step": 4538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.754713889724797e-05,
+      "loss": 1.1875,
+      "step": 4539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7546104526886235e-05,
+      "loss": 0.4922,
+      "step": 4540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7545069968974563e-05,
+      "loss": 0.625,
+      "step": 4541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7544035223538662e-05,
+      "loss": 0.6953,
+      "step": 4542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7543000290604256e-05,
+      "loss": 1.125,
+      "step": 4543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.754196517019706e-05,
+      "loss": 0.5977,
+      "step": 4544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.754092986234281e-05,
+      "loss": 0.8633,
+      "step": 4545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7539894367067228e-05,
+      "loss": 1.0391,
+      "step": 4546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7538858684396056e-05,
+      "loss": 1.1953,
+      "step": 4547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7537822814355033e-05,
+      "loss": 0.8281,
+      "step": 4548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7536786756969902e-05,
+      "loss": 1.0703,
+      "step": 4549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.753575051226642e-05,
+      "loss": 1.2266,
+      "step": 4550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7534714080270333e-05,
+      "loss": 0.8945,
+      "step": 4551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7533677461007407e-05,
+      "loss": 1.1719,
+      "step": 4552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.75326406545034e-05,
+      "loss": 1.1172,
+      "step": 4553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.753160366078409e-05,
+      "loss": 1.1953,
+      "step": 4554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.753056647987524e-05,
+      "loss": 1.0312,
+      "step": 4555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7529529111802635e-05,
+      "loss": 0.418,
+      "step": 4556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.752849155659205e-05,
+      "loss": 1.2422,
+      "step": 4557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7527453814269285e-05,
+      "loss": 1.6094,
+      "step": 4558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7526415884860117e-05,
+      "loss": 0.4434,
+      "step": 4559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7525377768390353e-05,
+      "loss": 1.2891,
+      "step": 4560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.752433946488579e-05,
+      "loss": 0.9531,
+      "step": 4561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.752330097437224e-05,
+      "loss": 1.0156,
+      "step": 4562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.75222622968755e-05,
+      "loss": 1.1094,
+      "step": 4563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.75212234324214e-05,
+      "loss": 1.0703,
+      "step": 4564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.752018438103575e-05,
+      "loss": 2.0938,
+      "step": 4565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7519145142744377e-05,
+      "loss": 1.3906,
+      "step": 4566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7518105717573114e-05,
+      "loss": 0.5898,
+      "step": 4567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7517066105547792e-05,
+      "loss": 1.0078,
+      "step": 4568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7516026306694246e-05,
+      "loss": 1.1328,
+      "step": 4569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7514986321038324e-05,
+      "loss": 1.6016,
+      "step": 4570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.751394614860587e-05,
+      "loss": 1.3047,
+      "step": 4571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.751290578942274e-05,
+      "loss": 1.0312,
+      "step": 4572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7511865243514788e-05,
+      "loss": 1.1953,
+      "step": 4573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.751082451090788e-05,
+      "loss": 0.9883,
+      "step": 4574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.750978359162787e-05,
+      "loss": 0.7344,
+      "step": 4575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7508742485700644e-05,
+      "loss": 1.5781,
+      "step": 4576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7507701193152072e-05,
+      "loss": 0.6289,
+      "step": 4577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7506659714008033e-05,
+      "loss": 1.3594,
+      "step": 4578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7505618048294407e-05,
+      "loss": 1.1562,
+      "step": 4579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7504576196037095e-05,
+      "loss": 1.0234,
+      "step": 4580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7503534157261983e-05,
+      "loss": 1.6797,
+      "step": 4581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7502491931994973e-05,
+      "loss": 0.4355,
+      "step": 4582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.750144952026196e-05,
+      "loss": 0.8398,
+      "step": 4583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7500406922088868e-05,
+      "loss": 1.6328,
+      "step": 4584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.749936413750159e-05,
+      "loss": 0.6562,
+      "step": 4585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.749832116652606e-05,
+      "loss": 1.3125,
+      "step": 4586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7497278009188196e-05,
+      "loss": 0.4004,
+      "step": 4587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7496234665513915e-05,
+      "loss": 0.5117,
+      "step": 4588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7495191135529158e-05,
+      "loss": 1.1016,
+      "step": 4589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.749414741925986e-05,
+      "loss": 1.3906,
+      "step": 4590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7493103516731957e-05,
+      "loss": 1.4688,
+      "step": 4591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7492059427971395e-05,
+      "loss": 0.4434,
+      "step": 4592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.749101515300413e-05,
+      "loss": 1.1172,
+      "step": 4593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7489970691856103e-05,
+      "loss": 1.5156,
+      "step": 4594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7488926044553288e-05,
+      "loss": 1.3203,
+      "step": 4595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7487881211121638e-05,
+      "loss": 0.6367,
+      "step": 4596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7486836191587128e-05,
+      "loss": 0.793,
+      "step": 4597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7485790985975723e-05,
+      "loss": 0.9922,
+      "step": 4598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.748474559431341e-05,
+      "loss": 0.9961,
+      "step": 4599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7483700016626162e-05,
+      "loss": 0.9844,
+      "step": 4600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.748265425293997e-05,
+      "loss": 0.375,
+      "step": 4601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7481608303280825e-05,
+      "loss": 0.4512,
+      "step": 4602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.748056216767472e-05,
+      "loss": 1.5312,
+      "step": 4603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7479515846147664e-05,
+      "loss": 1.3672,
+      "step": 4604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7478469338725657e-05,
+      "loss": 0.9766,
+      "step": 4605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7477422645434705e-05,
+      "loss": 0.5312,
+      "step": 4606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7476375766300827e-05,
+      "loss": 0.5625,
+      "step": 4607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7475328701350038e-05,
+      "loss": 0.6523,
+      "step": 4608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7474281450608366e-05,
+      "loss": 1.3047,
+      "step": 4609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.747323401410184e-05,
+      "loss": 0.4883,
+      "step": 4610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7472186391856486e-05,
+      "loss": 0.3809,
+      "step": 4611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.747113858389835e-05,
+      "loss": 0.4414,
+      "step": 4612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7470090590253466e-05,
+      "loss": 0.9727,
+      "step": 4613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7469042410947887e-05,
+      "loss": 0.5781,
+      "step": 4614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.746799404600766e-05,
+      "loss": 0.5664,
+      "step": 4615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7466945495458844e-05,
+      "loss": 1.4766,
+      "step": 4616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.74658967593275e-05,
+      "loss": 0.6719,
+      "step": 4617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7464847837639692e-05,
+      "loss": 1.5,
+      "step": 4618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7463798730421486e-05,
+      "loss": 0.9492,
+      "step": 4619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.746274943769896e-05,
+      "loss": 1.25,
+      "step": 4620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7461699959498194e-05,
+      "loss": 1.3594,
+      "step": 4621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7460650295845268e-05,
+      "loss": 0.9297,
+      "step": 4622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7459600446766276e-05,
+      "loss": 1.2109,
+      "step": 4623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7458550412287306e-05,
+      "loss": 1.2344,
+      "step": 4624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7457500192434458e-05,
+      "loss": 1.125,
+      "step": 4625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7456449787233832e-05,
+      "loss": 1.0625,
+      "step": 4626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7455399196711536e-05,
+      "loss": 0.4863,
+      "step": 4627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7454348420893678e-05,
+      "loss": 0.6914,
+      "step": 4628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7453297459806377e-05,
+      "loss": 0.9688,
+      "step": 4629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7452246313475755e-05,
+      "loss": 1.6406,
+      "step": 4630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.745119498192793e-05,
+      "loss": 0.9688,
+      "step": 4631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7450143465189043e-05,
+      "loss": 1.1484,
+      "step": 4632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.744909176328522e-05,
+      "loss": 1.0938,
+      "step": 4633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7448039876242604e-05,
+      "loss": 1.3359,
+      "step": 4634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7446987804087332e-05,
+      "loss": 1.3438,
+      "step": 4635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.744593554684556e-05,
+      "loss": 0.4258,
+      "step": 4636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7444883104543432e-05,
+      "loss": 1.3984,
+      "step": 4637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7443830477207117e-05,
+      "loss": 1.4844,
+      "step": 4638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7442777664862766e-05,
+      "loss": 1.3047,
+      "step": 4639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7441724667536547e-05,
+      "loss": 1.0938,
+      "step": 4640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.744067148525464e-05,
+      "loss": 1.25,
+      "step": 4641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.743961811804321e-05,
+      "loss": 1.4688,
+      "step": 4642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7438564565928446e-05,
+      "loss": 0.9922,
+      "step": 4643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7437510828936523e-05,
+      "loss": 1.0312,
+      "step": 4644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.743645690709364e-05,
+      "loss": 1.1328,
+      "step": 4645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7435402800425984e-05,
+      "loss": 0.7148,
+      "step": 4646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7434348508959758e-05,
+      "loss": 0.7266,
+      "step": 4647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7433294032721166e-05,
+      "loss": 0.8359,
+      "step": 4648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.743223937173641e-05,
+      "loss": 0.8516,
+      "step": 4649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7431184526031705e-05,
+      "loss": 0.9648,
+      "step": 4650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7430129495633276e-05,
+      "loss": 1.0781,
+      "step": 4651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7429074280567332e-05,
+      "loss": 1.7812,
+      "step": 4652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7428018880860105e-05,
+      "loss": 0.5273,
+      "step": 4653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7426963296537825e-05,
+      "loss": 1.2578,
+      "step": 4654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.742590752762673e-05,
+      "loss": 0.8281,
+      "step": 4655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.742485157415306e-05,
+      "loss": 1.6719,
+      "step": 4656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7423795436143055e-05,
+      "loss": 1.5,
+      "step": 4657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7422739113622963e-05,
+      "loss": 0.8984,
+      "step": 4658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7421682606619047e-05,
+      "loss": 1.4141,
+      "step": 4659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.742062591515756e-05,
+      "loss": 0.4355,
+      "step": 4660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.741956903926476e-05,
+      "loss": 0.6094,
+      "step": 4661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.741851197896692e-05,
+      "loss": 0.875,
+      "step": 4662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7417454734290312e-05,
+      "loss": 0.7305,
+      "step": 4663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.741639730526121e-05,
+      "loss": 1.0938,
+      "step": 4664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7415339691905903e-05,
+      "loss": 1.4141,
+      "step": 4665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7414281894250665e-05,
+      "loss": 0.9531,
+      "step": 4666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.741322391232179e-05,
+      "loss": 1.75,
+      "step": 4667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.741216574614558e-05,
+      "loss": 0.4688,
+      "step": 4668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7411107395748327e-05,
+      "loss": 1.7031,
+      "step": 4669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7410048861156336e-05,
+      "loss": 0.7852,
+      "step": 4670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.740899014239592e-05,
+      "loss": 0.8242,
+      "step": 4671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7407931239493387e-05,
+      "loss": 0.8945,
+      "step": 4672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.740687215247506e-05,
+      "loss": 1.5234,
+      "step": 4673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7405812881367256e-05,
+      "loss": 1.6953,
+      "step": 4674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7404753426196308e-05,
+      "loss": 1.4688,
+      "step": 4675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.740369378698854e-05,
+      "loss": 0.7266,
+      "step": 4676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7402633963770298e-05,
+      "loss": 1.3828,
+      "step": 4677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7401573956567913e-05,
+      "loss": 1.2109,
+      "step": 4678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7400513765407736e-05,
+      "loss": 0.8555,
+      "step": 4679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7399453390316116e-05,
+      "loss": 1.1328,
+      "step": 4680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7398392831319403e-05,
+      "loss": 0.9375,
+      "step": 4681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7397332088443963e-05,
+      "loss": 0.5312,
+      "step": 4682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7396271161716156e-05,
+      "loss": 1.1016,
+      "step": 4683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7395210051162352e-05,
+      "loss": 0.7266,
+      "step": 4684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.739414875680892e-05,
+      "loss": 0.6016,
+      "step": 4685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7393087278682238e-05,
+      "loss": 1.0781,
+      "step": 4686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.739202561680869e-05,
+      "loss": 1.6406,
+      "step": 4687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7390963771214664e-05,
+      "loss": 0.9375,
+      "step": 4688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.738990174192655e-05,
+      "loss": 0.5156,
+      "step": 4689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.738883952897074e-05,
+      "loss": 0.6133,
+      "step": 4690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7387777132373635e-05,
+      "loss": 0.7852,
+      "step": 4691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.738671455216164e-05,
+      "loss": 0.8242,
+      "step": 4692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7385651788361165e-05,
+      "loss": 1.3203,
+      "step": 4693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7384588840998627e-05,
+      "loss": 0.8203,
+      "step": 4694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.738352571010044e-05,
+      "loss": 1.125,
+      "step": 4695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.738246239569303e-05,
+      "loss": 1.0,
+      "step": 4696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7381398897802825e-05,
+      "loss": 0.7266,
+      "step": 4697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7380335216456254e-05,
+      "loss": 1.0078,
+      "step": 4698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7379271351679747e-05,
+      "loss": 0.9297,
+      "step": 4699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.737820730349976e-05,
+      "loss": 0.8906,
+      "step": 4700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7377143071942727e-05,
+      "loss": 1.0078,
+      "step": 4701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7376078657035103e-05,
+      "loss": 1.1484,
+      "step": 4702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.737501405880335e-05,
+      "loss": 1.7656,
+      "step": 4703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.737394927727391e-05,
+      "loss": 0.4199,
+      "step": 4704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.737288431247326e-05,
+      "loss": 0.5625,
+      "step": 4705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.737181916442787e-05,
+      "loss": 0.3496,
+      "step": 4706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.73707538331642e-05,
+      "loss": 0.5469,
+      "step": 4707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.736968831870874e-05,
+      "loss": 0.7383,
+      "step": 4708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.736862262108797e-05,
+      "loss": 0.5273,
+      "step": 4709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7367556740328374e-05,
+      "loss": 0.2656,
+      "step": 4710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7366490676456442e-05,
+      "loss": 0.8047,
+      "step": 4711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7365424429498675e-05,
+      "loss": 0.5859,
+      "step": 4712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7364357999481566e-05,
+      "loss": 1.3281,
+      "step": 4713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.736329138643163e-05,
+      "loss": 1.0,
+      "step": 4714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7362224590375364e-05,
+      "loss": 1.4375,
+      "step": 4715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7361157611339292e-05,
+      "loss": 1.7656,
+      "step": 4716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7360090449349928e-05,
+      "loss": 1.1641,
+      "step": 4717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.73590231044338e-05,
+      "loss": 1.4922,
+      "step": 4718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.735795557661743e-05,
+      "loss": 1.4531,
+      "step": 4719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7356887865927353e-05,
+      "loss": 1.7422,
+      "step": 4720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.73558199723901e-05,
+      "loss": 1.6094,
+      "step": 4721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7354751896032222e-05,
+      "loss": 1.8594,
+      "step": 4722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7353683636880258e-05,
+      "loss": 1.0156,
+      "step": 4723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7352615194960765e-05,
+      "loss": 1.375,
+      "step": 4724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7351546570300288e-05,
+      "loss": 0.8984,
+      "step": 4725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7350477762925394e-05,
+      "loss": 1.1094,
+      "step": 4726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7349408772862645e-05,
+      "loss": 1.9766,
+      "step": 4727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.734833960013861e-05,
+      "loss": 1.2109,
+      "step": 4728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.734727024477986e-05,
+      "loss": 1.4062,
+      "step": 4729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7346200706812975e-05,
+      "loss": 0.9922,
+      "step": 4730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7345130986264537e-05,
+      "loss": 0.498,
+      "step": 4731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.734406108316113e-05,
+      "loss": 1.0234,
+      "step": 4732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7342990997529346e-05,
+      "loss": 0.3301,
+      "step": 4733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7341920729395784e-05,
+      "loss": 0.6445,
+      "step": 4734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7340850278787042e-05,
+      "loss": 0.7109,
+      "step": 4735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7339779645729725e-05,
+      "loss": 0.9922,
+      "step": 4736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7338708830250442e-05,
+      "loss": 0.7031,
+      "step": 4737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7337637832375814e-05,
+      "loss": 1.2812,
+      "step": 4738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7336566652132445e-05,
+      "loss": 1.4688,
+      "step": 4739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7335495289546968e-05,
+      "loss": 1.3047,
+      "step": 4740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.733442374464601e-05,
+      "loss": 0.8125,
+      "step": 4741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7333352017456198e-05,
+      "loss": 1.3984,
+      "step": 4742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7332280108004177e-05,
+      "loss": 1.0469,
+      "step": 4743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.733120801631658e-05,
+      "loss": 1.2344,
+      "step": 4744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7330135742420053e-05,
+      "loss": 0.875,
+      "step": 4745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.732906328634125e-05,
+      "loss": 1.4844,
+      "step": 4746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.732799064810683e-05,
+      "loss": 1.1016,
+      "step": 4747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7326917827743444e-05,
+      "loss": 1.2422,
+      "step": 4748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7325844825277756e-05,
+      "loss": 0.6523,
+      "step": 4749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.732477164073644e-05,
+      "loss": 0.7891,
+      "step": 4750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7323698274146163e-05,
+      "loss": 1.1797,
+      "step": 4751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7322624725533608e-05,
+      "loss": 0.6406,
+      "step": 4752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7321550994925452e-05,
+      "loss": 0.5625,
+      "step": 4753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7320477082348382e-05,
+      "loss": 1.4609,
+      "step": 4754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.731940298782909e-05,
+      "loss": 1.0625,
+      "step": 4755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.731832871139427e-05,
+      "loss": 0.5547,
+      "step": 4756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.731725425307063e-05,
+      "loss": 0.6133,
+      "step": 4757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7316179612884863e-05,
+      "loss": 0.5781,
+      "step": 4758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7315104790863682e-05,
+      "loss": 0.582,
+      "step": 4759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7314029787033802e-05,
+      "loss": 0.7383,
+      "step": 4760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7312954601421944e-05,
+      "loss": 1.8594,
+      "step": 4761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7311879234054823e-05,
+      "loss": 0.4336,
+      "step": 4762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.731080368495917e-05,
+      "loss": 1.2969,
+      "step": 4763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7309727954161716e-05,
+      "loss": 0.9062,
+      "step": 4764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.73086520416892e-05,
+      "loss": 0.3828,
+      "step": 4765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.730757594756836e-05,
+      "loss": 0.7539,
+      "step": 4766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.730649967182594e-05,
+      "loss": 0.5156,
+      "step": 4767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.730542321448869e-05,
+      "loss": 0.8594,
+      "step": 4768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7304346575583366e-05,
+      "loss": 1.125,
+      "step": 4769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7303269755136726e-05,
+      "loss": 1.5312,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7302192753175534e-05,
+      "loss": 0.7969,
+      "step": 4771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.730111556972655e-05,
+      "loss": 1.5859,
+      "step": 4772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.730003820481656e-05,
+      "loss": 0.7695,
+      "step": 4773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.729896065847233e-05,
+      "loss": 1.0078,
+      "step": 4774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7297882930720646e-05,
+      "loss": 1.4688,
+      "step": 4775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.729680502158829e-05,
+      "loss": 0.9023,
+      "step": 4776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.729572693110206e-05,
+      "loss": 1.7422,
+      "step": 4777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7294648659288738e-05,
+      "loss": 1.0391,
+      "step": 4778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7293570206175136e-05,
+      "loss": 1.875,
+      "step": 4779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.729249157178805e-05,
+      "loss": 1.0703,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7291412756154286e-05,
+      "loss": 1.1719,
+      "step": 4781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7290333759300666e-05,
+      "loss": 0.6367,
+      "step": 4782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7289254581254003e-05,
+      "loss": 0.7383,
+      "step": 4783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7288175222041115e-05,
+      "loss": 0.8945,
+      "step": 4784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7287095681688835e-05,
+      "loss": 0.8242,
+      "step": 4785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7286015960223987e-05,
+      "loss": 1.2969,
+      "step": 4786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7284936057673412e-05,
+      "loss": 0.6445,
+      "step": 4787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7283855974063945e-05,
+      "loss": 0.8945,
+      "step": 4788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7282775709422435e-05,
+      "loss": 0.5977,
+      "step": 4789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7281695263775727e-05,
+      "loss": 1.3906,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7280614637150675e-05,
+      "loss": 1.1172,
+      "step": 4791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7279533829574138e-05,
+      "loss": 1.3516,
+      "step": 4792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7278452841072975e-05,
+      "loss": 1.1484,
+      "step": 4793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7277371671674056e-05,
+      "loss": 0.6797,
+      "step": 4794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7276290321404255e-05,
+      "loss": 0.8398,
+      "step": 4795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.727520879029044e-05,
+      "loss": 1.9922,
+      "step": 4796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.72741270783595e-05,
+      "loss": 1.2109,
+      "step": 4797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7273045185638314e-05,
+      "loss": 1.0,
+      "step": 4798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7271963112153768e-05,
+      "loss": 1.1328,
+      "step": 4799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7270880857932763e-05,
+      "loss": 0.6133,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7269798423002196e-05,
+      "loss": 0.9766,
+      "step": 4801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7268715807388967e-05,
+      "loss": 1.4375,
+      "step": 4802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7267633011119985e-05,
+      "loss": 0.8945,
+      "step": 4803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.726655003422216e-05,
+      "loss": 1.8203,
+      "step": 4804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.726546687672241e-05,
+      "loss": 1.3203,
+      "step": 4805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7264383538647652e-05,
+      "loss": 0.5625,
+      "step": 4806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.726330002002482e-05,
+      "loss": 0.5977,
+      "step": 4807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.726221632088083e-05,
+      "loss": 0.3184,
+      "step": 4808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7261132441242628e-05,
+      "loss": 0.8359,
+      "step": 4809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7260048381137147e-05,
+      "loss": 0.418,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.725896414059133e-05,
+      "loss": 0.75,
+      "step": 4811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.725787971963213e-05,
+      "loss": 1.3594,
+      "step": 4812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7256795118286494e-05,
+      "loss": 0.5781,
+      "step": 4813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7255710336581378e-05,
+      "loss": 0.459,
+      "step": 4814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7254625374543743e-05,
+      "loss": 0.4141,
+      "step": 4815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.725354023220056e-05,
+      "loss": 0.6719,
+      "step": 4816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7252454909578794e-05,
+      "loss": 0.8359,
+      "step": 4817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.725136940670542e-05,
+      "loss": 0.8398,
+      "step": 4818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7250283723607416e-05,
+      "loss": 1.6172,
+      "step": 4819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.724919786031177e-05,
+      "loss": 1.1953,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7248111816845464e-05,
+      "loss": 1.6172,
+      "step": 4821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7247025593235497e-05,
+      "loss": 0.5742,
+      "step": 4822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7245939189508857e-05,
+      "loss": 0.8906,
+      "step": 4823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7244852605692556e-05,
+      "loss": 0.9609,
+      "step": 4824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.724376584181359e-05,
+      "loss": 0.9375,
+      "step": 4825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7242678897898976e-05,
+      "loss": 0.8867,
+      "step": 4826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7241591773975725e-05,
+      "loss": 0.6992,
+      "step": 4827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7240504470070858e-05,
+      "loss": 1.0859,
+      "step": 4828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.72394169862114e-05,
+      "loss": 0.6836,
+      "step": 4829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7238329322424375e-05,
+      "loss": 1.3906,
+      "step": 4830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7237241478736822e-05,
+      "loss": 0.7266,
+      "step": 4831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.723615345517577e-05,
+      "loss": 0.3184,
+      "step": 4832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7235065251768266e-05,
+      "loss": 1.25,
+      "step": 4833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.723397686854136e-05,
+      "loss": 1.1484,
+      "step": 4834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.723288830552209e-05,
+      "loss": 0.8438,
+      "step": 4835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.723179956273752e-05,
+      "loss": 0.5234,
+      "step": 4836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.723071064021471e-05,
+      "loss": 0.9219,
+      "step": 4837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7229621537980724e-05,
+      "loss": 1.3516,
+      "step": 4838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7228532256062624e-05,
+      "loss": 0.8281,
+      "step": 4839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7227442794487487e-05,
+      "loss": 1.2656,
+      "step": 4840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7226353153282395e-05,
+      "loss": 1.0469,
+      "step": 4841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7225263332474426e-05,
+      "loss": 1.2656,
+      "step": 4842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7224173332090658e-05,
+      "loss": 1.2812,
+      "step": 4843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7223083152158198e-05,
+      "loss": 0.4121,
+      "step": 4844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.722199279270413e-05,
+      "loss": 0.7266,
+      "step": 4845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7220902253755555e-05,
+      "loss": 0.9883,
+      "step": 4846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7219811535339575e-05,
+      "loss": 1.4297,
+      "step": 4847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7218720637483305e-05,
+      "loss": 0.6367,
+      "step": 4848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7217629560213858e-05,
+      "loss": 0.9297,
+      "step": 4849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7216538303558344e-05,
+      "loss": 0.7734,
+      "step": 4850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.721544686754389e-05,
+      "loss": 1.7578,
+      "step": 4851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7214355252197625e-05,
+      "loss": 1.0625,
+      "step": 4852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7213263457546675e-05,
+      "loss": 0.6328,
+      "step": 4853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.721217148361818e-05,
+      "loss": 1.8594,
+      "step": 4854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.721107933043927e-05,
+      "loss": 0.5,
+      "step": 4855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.72099869980371e-05,
+      "loss": 0.4844,
+      "step": 4856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7208894486438815e-05,
+      "loss": 1.1797,
+      "step": 4857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7207801795671567e-05,
+      "loss": 1.8125,
+      "step": 4858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7206708925762513e-05,
+      "loss": 0.7734,
+      "step": 4859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.720561587673882e-05,
+      "loss": 0.2773,
+      "step": 4860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.720452264862765e-05,
+      "loss": 0.3711,
+      "step": 4861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7203429241456176e-05,
+      "loss": 0.9766,
+      "step": 4862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.720233565525157e-05,
+      "loss": 0.5508,
+      "step": 4863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7201241890041017e-05,
+      "loss": 0.9219,
+      "step": 4864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7200147945851694e-05,
+      "loss": 0.9844,
+      "step": 4865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7199053822710797e-05,
+      "loss": 0.2715,
+      "step": 4866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7197959520645516e-05,
+      "loss": 0.6836,
+      "step": 4867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7196865039683052e-05,
+      "loss": 1.2266,
+      "step": 4868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7195770379850604e-05,
+      "loss": 1.9531,
+      "step": 4869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7194675541175376e-05,
+      "loss": 2.0625,
+      "step": 4870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.719358052368458e-05,
+      "loss": 0.7109,
+      "step": 4871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7192485327405438e-05,
+      "loss": 0.875,
+      "step": 4872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.719138995236516e-05,
+      "loss": 1.3516,
+      "step": 4873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.719029439859098e-05,
+      "loss": 1.6719,
+      "step": 4874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.718919866611012e-05,
+      "loss": 0.9648,
+      "step": 4875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7188102754949815e-05,
+      "loss": 0.75,
+      "step": 4876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7187006665137307e-05,
+      "loss": 1.8828,
+      "step": 4877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.718591039669983e-05,
+      "loss": 0.5547,
+      "step": 4878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7184813949664637e-05,
+      "loss": 1.3203,
+      "step": 4879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7183717324058976e-05,
+      "loss": 0.832,
+      "step": 4880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.71826205199101e-05,
+      "loss": 1.2266,
+      "step": 4881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7181523537245274e-05,
+      "loss": 1.3125,
+      "step": 4882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.718042637609176e-05,
+      "loss": 0.5273,
+      "step": 4883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.717932903647683e-05,
+      "loss": 1.875,
+      "step": 4884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.717823151842775e-05,
+      "loss": 1.75,
+      "step": 4885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.71771338219718e-05,
+      "loss": 0.7461,
+      "step": 4886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7176035947136265e-05,
+      "loss": 1.0547,
+      "step": 4887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.717493789394843e-05,
+      "loss": 1.7969,
+      "step": 4888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.717383966243559e-05,
+      "loss": 0.9062,
+      "step": 4889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7172741252625033e-05,
+      "loss": 0.4434,
+      "step": 4890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.717164266454406e-05,
+      "loss": 0.8125,
+      "step": 4891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.717054389821998e-05,
+      "loss": 0.6484,
+      "step": 4892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7169444953680093e-05,
+      "loss": 1.5,
+      "step": 4893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7168345830951725e-05,
+      "loss": 0.6562,
+      "step": 4894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7167246530062183e-05,
+      "loss": 0.543,
+      "step": 4895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7166147051038794e-05,
+      "loss": 1.0547,
+      "step": 4896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.716504739390888e-05,
+      "loss": 1.1328,
+      "step": 4897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7163947558699773e-05,
+      "loss": 0.7188,
+      "step": 4898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.716284754543881e-05,
+      "loss": 0.7891,
+      "step": 4899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.716174735415333e-05,
+      "loss": 0.6953,
+      "step": 4900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.716064698487068e-05,
+      "loss": 1.1016,
+      "step": 4901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.71595464376182e-05,
+      "loss": 0.8203,
+      "step": 4902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.715844571242325e-05,
+      "loss": 1.125,
+      "step": 4903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7157344809313187e-05,
+      "loss": 1.1641,
+      "step": 4904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.715624372831537e-05,
+      "loss": 0.3379,
+      "step": 4905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.715514246945717e-05,
+      "loss": 1.0312,
+      "step": 4906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7154041032765946e-05,
+      "loss": 0.9102,
+      "step": 4907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7152939418269084e-05,
+      "loss": 0.375,
+      "step": 4908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7151837625993962e-05,
+      "loss": 0.4199,
+      "step": 4909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7150735655967963e-05,
+      "loss": 0.2852,
+      "step": 4910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.714963350821847e-05,
+      "loss": 0.4941,
+      "step": 4911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7148531182772883e-05,
+      "loss": 1.4531,
+      "step": 4912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7147428679658594e-05,
+      "loss": 0.1904,
+      "step": 4913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.714632599890301e-05,
+      "loss": 0.7227,
+      "step": 4914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.714522314053353e-05,
+      "loss": 0.7305,
+      "step": 4915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.714412010457757e-05,
+      "loss": 1.4062,
+      "step": 4916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.714301689106254e-05,
+      "loss": 2.0156,
+      "step": 4917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.714191350001586e-05,
+      "loss": 1.3672,
+      "step": 4918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.714080993146496e-05,
+      "loss": 0.3711,
+      "step": 4919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7139706185437258e-05,
+      "loss": 0.9219,
+      "step": 4920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7138602261960197e-05,
+      "loss": 0.9297,
+      "step": 4921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7137498161061202e-05,
+      "loss": 1.5469,
+      "step": 4922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7136393882767724e-05,
+      "loss": 1.4453,
+      "step": 4923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7135289427107204e-05,
+      "loss": 0.5469,
+      "step": 4924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.713418479410709e-05,
+      "loss": 0.9766,
+      "step": 4925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7133079983794842e-05,
+      "loss": 0.3008,
+      "step": 4926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7131974996197916e-05,
+      "loss": 0.8125,
+      "step": 4927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7130869831343774e-05,
+      "loss": 1.6406,
+      "step": 4928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.712976448925989e-05,
+      "loss": 0.9727,
+      "step": 4929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7128658969973724e-05,
+      "loss": 0.6562,
+      "step": 4930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7127553273512763e-05,
+      "loss": 1.0781,
+      "step": 4931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7126447399904483e-05,
+      "loss": 0.832,
+      "step": 4932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.712534134917637e-05,
+      "loss": 1.1328,
+      "step": 4933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7124235121355918e-05,
+      "loss": 0.875,
+      "step": 4934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7123128716470613e-05,
+      "loss": 0.9648,
+      "step": 4935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7122022134547958e-05,
+      "loss": 0.9805,
+      "step": 4936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7120915375615458e-05,
+      "loss": 0.9453,
+      "step": 4937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7119808439700617e-05,
+      "loss": 1.3359,
+      "step": 4938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7118701326830944e-05,
+      "loss": 1.8984,
+      "step": 4939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.711759403703396e-05,
+      "loss": 1.0234,
+      "step": 4940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7116486570337187e-05,
+      "loss": 1.5312,
+      "step": 4941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.711537892676814e-05,
+      "loss": 0.8359,
+      "step": 4942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7114271106354365e-05,
+      "loss": 0.8867,
+      "step": 4943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.711316310912338e-05,
+      "loss": 1.0391,
+      "step": 4944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.711205493510273e-05,
+      "loss": 0.9141,
+      "step": 4945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.711094658431995e-05,
+      "loss": 1.1641,
+      "step": 4946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.71098380568026e-05,
+      "loss": 0.9414,
+      "step": 4947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.710872935257822e-05,
+      "loss": 1.2266,
+      "step": 4948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7107620471674373e-05,
+      "loss": 0.498,
+      "step": 4949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7106511414118612e-05,
+      "loss": 1.0234,
+      "step": 4950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.710540217993851e-05,
+      "loss": 0.8438,
+      "step": 4951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7104292769161627e-05,
+      "loss": 1.5859,
+      "step": 4952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7103183181815537e-05,
+      "loss": 1.7266,
+      "step": 4953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7102073417927824e-05,
+      "loss": 1.1875,
+      "step": 4954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.710096347752607e-05,
+      "loss": 1.7578,
+      "step": 4955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.709985336063785e-05,
+      "loss": 0.8125,
+      "step": 4956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7098743067290768e-05,
+      "loss": 0.5273,
+      "step": 4957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.709763259751241e-05,
+      "loss": 0.3359,
+      "step": 4958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7096521951330383e-05,
+      "loss": 1.1562,
+      "step": 4959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7095411128772284e-05,
+      "loss": 0.5156,
+      "step": 4960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7094300129865726e-05,
+      "loss": 0.7266,
+      "step": 4961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7093188954638317e-05,
+      "loss": 0.6055,
+      "step": 4962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.709207760311768e-05,
+      "loss": 0.5234,
+      "step": 4963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.709096607533143e-05,
+      "loss": 1.2031,
+      "step": 4964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7089854371307197e-05,
+      "loss": 0.5117,
+      "step": 4965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.708874249107261e-05,
+      "loss": 1.5156,
+      "step": 4966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7087630434655303e-05,
+      "loss": 0.707,
+      "step": 4967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7086518202082917e-05,
+      "loss": 1.875,
+      "step": 4968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7085405793383094e-05,
+      "loss": 1.8281,
+      "step": 4969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.708429320858348e-05,
+      "loss": 1.0156,
+      "step": 4970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.708318044771173e-05,
+      "loss": 1.3516,
+      "step": 4971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7082067510795497e-05,
+      "loss": 0.6094,
+      "step": 4972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7080954397862446e-05,
+      "loss": 1.6406,
+      "step": 4973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7079841108940235e-05,
+      "loss": 0.5898,
+      "step": 4974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7078727644056545e-05,
+      "loss": 0.5508,
+      "step": 4975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.707761400323904e-05,
+      "loss": 0.5078,
+      "step": 4976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.70765001865154e-05,
+      "loss": 0.9531,
+      "step": 4977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7075386193913314e-05,
+      "loss": 0.6133,
+      "step": 4978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7074272025460462e-05,
+      "loss": 0.75,
+      "step": 4979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7073157681184537e-05,
+      "loss": 1.1797,
+      "step": 4980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7072043161113235e-05,
+      "loss": 0.9062,
+      "step": 4981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7070928465274255e-05,
+      "loss": 1.4375,
+      "step": 4982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7069813593695306e-05,
+      "loss": 0.4863,
+      "step": 4983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7068698546404094e-05,
+      "loss": 0.7305,
+      "step": 4984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7067583323428332e-05,
+      "loss": 0.7344,
+      "step": 4985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.706646792479574e-05,
+      "loss": 1.3359,
+      "step": 4986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7065352350534034e-05,
+      "loss": 0.7461,
+      "step": 4987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7064236600670945e-05,
+      "loss": 1.0703,
+      "step": 4988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7063120675234205e-05,
+      "loss": 0.8281,
+      "step": 4989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7062004574251544e-05,
+      "loss": 1.0312,
+      "step": 4990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7060888297750707e-05,
+      "loss": 0.9414,
+      "step": 4991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7059771845759433e-05,
+      "loss": 1.3906,
+      "step": 4992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7058655218305476e-05,
+      "loss": 0.9219,
+      "step": 4993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.705753841541658e-05,
+      "loss": 1.1172,
+      "step": 4994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.705642143712051e-05,
+      "loss": 1.1484,
+      "step": 4995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7055304283445024e-05,
+      "loss": 1.0781,
+      "step": 4996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7054186954417885e-05,
+      "loss": 0.8477,
+      "step": 4997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.705306945006687e-05,
+      "loss": 0.8867,
+      "step": 4998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7051951770419747e-05,
+      "loss": 0.6445,
+      "step": 4999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7050833915504295e-05,
+      "loss": 0.9336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7049715885348298e-05,
+      "loss": 1.5234,
+      "step": 5001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7048597679979547e-05,
+      "loss": 1.8438,
+      "step": 5002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.704747929942583e-05,
+      "loss": 1.6094,
+      "step": 5003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7046360743714942e-05,
+      "loss": 1.125,
+      "step": 5004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7045242012874686e-05,
+      "loss": 1.1016,
+      "step": 5005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.704412310693287e-05,
+      "loss": 0.7344,
+      "step": 5006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.704300402591729e-05,
+      "loss": 0.7539,
+      "step": 5007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7041884769855777e-05,
+      "loss": 0.9453,
+      "step": 5008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7040765338776135e-05,
+      "loss": 0.2793,
+      "step": 5009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7039645732706195e-05,
+      "loss": 1.5781,
+      "step": 5010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7038525951673783e-05,
+      "loss": 0.4102,
+      "step": 5011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7037405995706722e-05,
+      "loss": 1.6484,
+      "step": 5012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7036285864832854e-05,
+      "loss": 0.2793,
+      "step": 5013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7035165559080014e-05,
+      "loss": 0.4785,
+      "step": 5014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7034045078476054e-05,
+      "loss": 0.4805,
+      "step": 5015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7032924423048816e-05,
+      "loss": 1.75,
+      "step": 5016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.703180359282615e-05,
+      "loss": 0.9062,
+      "step": 5017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7030682587835922e-05,
+      "loss": 1.0703,
+      "step": 5018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7029561408105987e-05,
+      "loss": 0.2871,
+      "step": 5019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.702844005366421e-05,
+      "loss": 1.7578,
+      "step": 5020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7027318524538463e-05,
+      "loss": 1.6562,
+      "step": 5021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7026196820756622e-05,
+      "loss": 0.5586,
+      "step": 5022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7025074942346562e-05,
+      "loss": 2.1562,
+      "step": 5023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7023952889336172e-05,
+      "loss": 0.2383,
+      "step": 5024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7022830661753327e-05,
+      "loss": 0.4707,
+      "step": 5025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7021708259625934e-05,
+      "loss": 1.1484,
+      "step": 5026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.702058568298188e-05,
+      "loss": 0.7578,
+      "step": 5027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7019462931849068e-05,
+      "loss": 0.7969,
+      "step": 5028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.70183400062554e-05,
+      "loss": 0.9023,
+      "step": 5029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.701721690622879e-05,
+      "loss": 1.125,
+      "step": 5030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7016093631797144e-05,
+      "loss": 1.4688,
+      "step": 5031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7014970182988387e-05,
+      "loss": 1.5391,
+      "step": 5032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.701384655983044e-05,
+      "loss": 0.9219,
+      "step": 5033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.701272276235123e-05,
+      "loss": 0.5703,
+      "step": 5034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.701159879057868e-05,
+      "loss": 0.8359,
+      "step": 5035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.701047464454073e-05,
+      "loss": 1.8906,
+      "step": 5036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.700935032426532e-05,
+      "loss": 0.7773,
+      "step": 5037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7008225829780396e-05,
+      "loss": 0.4062,
+      "step": 5038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7007101161113904e-05,
+      "loss": 1.0859,
+      "step": 5039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7005976318293796e-05,
+      "loss": 1.0312,
+      "step": 5040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7004851301348027e-05,
+      "loss": 1.1016,
+      "step": 5041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7003726110304563e-05,
+      "loss": 1.1406,
+      "step": 5042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.700260074519136e-05,
+      "loss": 1.1094,
+      "step": 5043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.70014752060364e-05,
+      "loss": 1.1484,
+      "step": 5044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7000349492867646e-05,
+      "loss": 0.9844,
+      "step": 5045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6999223605713085e-05,
+      "loss": 1.0469,
+      "step": 5046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.699809754460069e-05,
+      "loss": 0.9219,
+      "step": 5047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6996971309558457e-05,
+      "loss": 0.5312,
+      "step": 5048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.699584490061438e-05,
+      "loss": 0.8164,
+      "step": 5049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.699471831779644e-05,
+      "loss": 1.0,
+      "step": 5050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.699359156113265e-05,
+      "loss": 1.4688,
+      "step": 5051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.699246463065101e-05,
+      "loss": 1.8672,
+      "step": 5052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6991337526379527e-05,
+      "loss": 0.4434,
+      "step": 5053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6990210248346214e-05,
+      "loss": 0.3164,
+      "step": 5054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.698908279657909e-05,
+      "loss": 0.8203,
+      "step": 5055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6987955171106182e-05,
+      "loss": 0.1357,
+      "step": 5056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6986827371955503e-05,
+      "loss": 0.9102,
+      "step": 5057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.698569939915509e-05,
+      "loss": 0.2637,
+      "step": 5058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6984571252732985e-05,
+      "loss": 0.7812,
+      "step": 5059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6983442932717213e-05,
+      "loss": 0.6953,
+      "step": 5060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6982314439135826e-05,
+      "loss": 0.543,
+      "step": 5061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.698118577201687e-05,
+      "loss": 0.8008,
+      "step": 5062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6980056931388393e-05,
+      "loss": 0.8281,
+      "step": 5063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6978927917278455e-05,
+      "loss": 1.7812,
+      "step": 5064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6977798729715116e-05,
+      "loss": 0.377,
+      "step": 5065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.697666936872644e-05,
+      "loss": 0.5156,
+      "step": 5066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6975539834340494e-05,
+      "loss": 1.4844,
+      "step": 5067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6974410126585356e-05,
+      "loss": 0.7695,
+      "step": 5068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.69732802454891e-05,
+      "loss": 2.0625,
+      "step": 5069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6972150191079808e-05,
+      "loss": 1.1953,
+      "step": 5070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6971019963385566e-05,
+      "loss": 1.8516,
+      "step": 5071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6969889562434466e-05,
+      "loss": 0.3047,
+      "step": 5072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6968758988254604e-05,
+      "loss": 0.8047,
+      "step": 5073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.696762824087408e-05,
+      "loss": 1.375,
+      "step": 5074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.696649732032099e-05,
+      "loss": 0.75,
+      "step": 5075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.696536622662345e-05,
+      "loss": 1.7812,
+      "step": 5076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6964234959809566e-05,
+      "loss": 0.9453,
+      "step": 5077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.696310351990746e-05,
+      "loss": 0.8672,
+      "step": 5078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.696197190694525e-05,
+      "loss": 0.4805,
+      "step": 5079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6960840120951058e-05,
+      "loss": 0.6953,
+      "step": 5080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6959708161953016e-05,
+      "loss": 0.7148,
+      "step": 5081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6958576029979262e-05,
+      "loss": 1.5156,
+      "step": 5082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6957443725057925e-05,
+      "loss": 0.6172,
+      "step": 5083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6956311247217156e-05,
+      "loss": 0.9375,
+      "step": 5084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.695517859648509e-05,
+      "loss": 0.707,
+      "step": 5085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.695404577288989e-05,
+      "loss": 1.2656,
+      "step": 5086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6952912776459708e-05,
+      "loss": 1.6797,
+      "step": 5087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6951779607222695e-05,
+      "loss": 0.6797,
+      "step": 5088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6950646265207028e-05,
+      "loss": 1.5859,
+      "step": 5089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6949512750440865e-05,
+      "loss": 0.8086,
+      "step": 5090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6948379062952377e-05,
+      "loss": 0.9766,
+      "step": 5091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.694724520276975e-05,
+      "loss": 1.1953,
+      "step": 5092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.694611116992116e-05,
+      "loss": 1.5625,
+      "step": 5093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6944976964434786e-05,
+      "loss": 1.1172,
+      "step": 5094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.694384258633883e-05,
+      "loss": 1.0859,
+      "step": 5095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.694270803566147e-05,
+      "loss": 0.6562,
+      "step": 5096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.694157331243092e-05,
+      "loss": 1.5078,
+      "step": 5097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6940438416675373e-05,
+      "loss": 1.0703,
+      "step": 5098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6939303348423036e-05,
+      "loss": 0.9766,
+      "step": 5099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6938168107702125e-05,
+      "loss": 1.3359,
+      "step": 5100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.693703269454085e-05,
+      "loss": 1.8281,
+      "step": 5101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.693589710896743e-05,
+      "loss": 0.8633,
+      "step": 5102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.693476135101009e-05,
+      "loss": 0.8438,
+      "step": 5103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6933625420697062e-05,
+      "loss": 0.7266,
+      "step": 5104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6932489318056573e-05,
+      "loss": 0.6406,
+      "step": 5105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.693135304311686e-05,
+      "loss": 1.2031,
+      "step": 5106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6930216595906167e-05,
+      "loss": 0.8867,
+      "step": 5107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6929079976452737e-05,
+      "loss": 1.4922,
+      "step": 5108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.692794318478482e-05,
+      "loss": 0.5352,
+      "step": 5109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6926806220930667e-05,
+      "loss": 0.3828,
+      "step": 5110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6925669084918544e-05,
+      "loss": 1.6328,
+      "step": 5111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6924531776776703e-05,
+      "loss": 1.6016,
+      "step": 5112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6923394296533416e-05,
+      "loss": 0.3164,
+      "step": 5113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.692225664421695e-05,
+      "loss": 0.7461,
+      "step": 5114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6921118819855592e-05,
+      "loss": 0.332,
+      "step": 5115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.69199808234776e-05,
+      "loss": 0.6367,
+      "step": 5116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.691884265511128e-05,
+      "loss": 0.9219,
+      "step": 5117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6917704314784907e-05,
+      "loss": 2.0938,
+      "step": 5118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6916565802526778e-05,
+      "loss": 0.8398,
+      "step": 5119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.691542711836518e-05,
+      "loss": 0.6211,
+      "step": 5120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.691428826232843e-05,
+      "loss": 0.8906,
+      "step": 5121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6913149234444818e-05,
+      "loss": 1.0234,
+      "step": 5122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6912010034742663e-05,
+      "loss": 1.0312,
+      "step": 5123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6910870663250273e-05,
+      "loss": 1.6328,
+      "step": 5124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.690973111999597e-05,
+      "loss": 0.6172,
+      "step": 5125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.690859140500807e-05,
+      "loss": 1.0312,
+      "step": 5126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6907451518314903e-05,
+      "loss": 0.8164,
+      "step": 5127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6906311459944804e-05,
+      "loss": 0.9688,
+      "step": 5128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6905171229926104e-05,
+      "loss": 0.9609,
+      "step": 5129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6904030828287137e-05,
+      "loss": 1.2109,
+      "step": 5130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6902890255056253e-05,
+      "loss": 0.5742,
+      "step": 5131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.69017495102618e-05,
+      "loss": 0.8906,
+      "step": 5132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6900608593932127e-05,
+      "loss": 1.1406,
+      "step": 5133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.689946750609559e-05,
+      "loss": 0.3691,
+      "step": 5134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6898326246780547e-05,
+      "loss": 1.3125,
+      "step": 5135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.689718481601537e-05,
+      "loss": 0.6797,
+      "step": 5136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6896043213828422e-05,
+      "loss": 1.0625,
+      "step": 5137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.689490144024808e-05,
+      "loss": 0.5117,
+      "step": 5138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.689375949530272e-05,
+      "loss": 1.2266,
+      "step": 5139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6892617379020726e-05,
+      "loss": 0.4238,
+      "step": 5140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.689147509143048e-05,
+      "loss": 1.1562,
+      "step": 5141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.689033263256037e-05,
+      "loss": 1.7266,
+      "step": 5142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6889190002438798e-05,
+      "loss": 1.5156,
+      "step": 5143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.688804720109416e-05,
+      "loss": 1.4453,
+      "step": 5144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6886904228554853e-05,
+      "loss": 0.9922,
+      "step": 5145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6885761084849296e-05,
+      "loss": 1.2656,
+      "step": 5146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6884617770005893e-05,
+      "loss": 0.6953,
+      "step": 5147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6883474284053055e-05,
+      "loss": 0.9688,
+      "step": 5148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6882330627019214e-05,
+      "loss": 1.5547,
+      "step": 5149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6881186798932784e-05,
+      "loss": 0.9219,
+      "step": 5150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.68800427998222e-05,
+      "loss": 1.0469,
+      "step": 5151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.687889862971589e-05,
+      "loss": 0.8789,
+      "step": 5152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.68777542886423e-05,
+      "loss": 1.1562,
+      "step": 5153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.687660977662986e-05,
+      "loss": 0.5586,
+      "step": 5154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.687546509370702e-05,
+      "loss": 1.6875,
+      "step": 5155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6874320239902235e-05,
+      "loss": 0.5898,
+      "step": 5156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.687317521524395e-05,
+      "loss": 0.4648,
+      "step": 5157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.687203001976063e-05,
+      "loss": 0.7266,
+      "step": 5158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6870884653480733e-05,
+      "loss": 0.5352,
+      "step": 5159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.686973911643273e-05,
+      "loss": 0.4766,
+      "step": 5160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.686859340864509e-05,
+      "loss": 0.4648,
+      "step": 5161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6867447530146286e-05,
+      "loss": 1.1797,
+      "step": 5162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.68663014809648e-05,
+      "loss": 0.6523,
+      "step": 5163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.686515526112912e-05,
+      "loss": 1.0859,
+      "step": 5164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6864008870667726e-05,
+      "loss": 0.8516,
+      "step": 5165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6862862309609116e-05,
+      "loss": 1.1641,
+      "step": 5166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6861715577981784e-05,
+      "loss": 0.6641,
+      "step": 5167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.686056867581423e-05,
+      "loss": 0.793,
+      "step": 5168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6859421603134957e-05,
+      "loss": 1.0703,
+      "step": 5169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6858274359972485e-05,
+      "loss": 1.2734,
+      "step": 5170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6857126946355312e-05,
+      "loss": 1.1328,
+      "step": 5171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.685597936231197e-05,
+      "loss": 1.5391,
+      "step": 5172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6854831607870967e-05,
+      "loss": 0.793,
+      "step": 5173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.685368368306084e-05,
+      "loss": 1.2734,
+      "step": 5174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6852535587910116e-05,
+      "loss": 0.8164,
+      "step": 5175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6851387322447334e-05,
+      "loss": 0.6562,
+      "step": 5176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.685023888670102e-05,
+      "loss": 0.5312,
+      "step": 5177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.684909028069973e-05,
+      "loss": 1.2188,
+      "step": 5178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6847941504472004e-05,
+      "loss": 0.6094,
+      "step": 5179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6846792558046398e-05,
+      "loss": 1.5391,
+      "step": 5180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6845643441451466e-05,
+      "loss": 0.6289,
+      "step": 5181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.684449415471577e-05,
+      "loss": 0.9922,
+      "step": 5182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6843344697867866e-05,
+      "loss": 0.8477,
+      "step": 5183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6842195070936335e-05,
+      "loss": 0.75,
+      "step": 5184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.684104527394974e-05,
+      "loss": 0.8359,
+      "step": 5185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6839895306936663e-05,
+      "loss": 1.3594,
+      "step": 5186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6838745169925683e-05,
+      "loss": 1.3594,
+      "step": 5187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6837594862945386e-05,
+      "loss": 0.3594,
+      "step": 5188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.683644438602436e-05,
+      "loss": 1.0469,
+      "step": 5189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.68352937391912e-05,
+      "loss": 0.7578,
+      "step": 5190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6834142922474506e-05,
+      "loss": 1.2266,
+      "step": 5191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6832991935902877e-05,
+      "loss": 0.7852,
+      "step": 5192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6831840779504922e-05,
+      "loss": 1.6016,
+      "step": 5193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.683068945330925e-05,
+      "loss": 1.1172,
+      "step": 5194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6829537957344474e-05,
+      "loss": 1.0078,
+      "step": 5195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.682838629163922e-05,
+      "loss": 1.1406,
+      "step": 5196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6827234456222107e-05,
+      "loss": 0.8867,
+      "step": 5197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6826082451121757e-05,
+      "loss": 0.4941,
+      "step": 5198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.682493027636681e-05,
+      "loss": 0.8516,
+      "step": 5199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6823777931985903e-05,
+      "loss": 1.0703,
+      "step": 5200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.682262541800767e-05,
+      "loss": 1.2969,
+      "step": 5201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.682147273446076e-05,
+      "loss": 0.9492,
+      "step": 5202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6820319881373815e-05,
+      "loss": 0.5586,
+      "step": 5203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6819166858775494e-05,
+      "loss": 0.9727,
+      "step": 5204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6818013666694452e-05,
+      "loss": 1.0391,
+      "step": 5205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6816860305159356e-05,
+      "loss": 0.2715,
+      "step": 5206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6815706774198863e-05,
+      "loss": 0.4785,
+      "step": 5207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6814553073841647e-05,
+      "loss": 0.4375,
+      "step": 5208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6813399204116378e-05,
+      "loss": 0.5312,
+      "step": 5209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.681224516505174e-05,
+      "loss": 1.1797,
+      "step": 5210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6811090956676414e-05,
+      "loss": 1.1875,
+      "step": 5211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6809936579019083e-05,
+      "loss": 0.2461,
+      "step": 5212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.680878203210844e-05,
+      "loss": 0.6719,
+      "step": 5213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6807627315973178e-05,
+      "loss": 1.25,
+      "step": 5214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6806472430642e-05,
+      "loss": 1.25,
+      "step": 5215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6805317376143604e-05,
+      "loss": 0.8125,
+      "step": 5216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.680416215250671e-05,
+      "loss": 0.6016,
+      "step": 5217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6803006759760013e-05,
+      "loss": 0.7891,
+      "step": 5218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.680185119793224e-05,
+      "loss": 0.9609,
+      "step": 5219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6800695467052103e-05,
+      "loss": 1.1641,
+      "step": 5220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6799539567148336e-05,
+      "loss": 1.2266,
+      "step": 5221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.679838349824966e-05,
+      "loss": 1.4062,
+      "step": 5222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6797227260384816e-05,
+      "loss": 1.5,
+      "step": 5223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.679607085358253e-05,
+      "loss": 1.3828,
+      "step": 5224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6794914277871553e-05,
+      "loss": 1.1641,
+      "step": 5225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.679375753328062e-05,
+      "loss": 0.7227,
+      "step": 5226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6792600619838496e-05,
+      "loss": 1.3438,
+      "step": 5227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6791443537573916e-05,
+      "loss": 0.7891,
+      "step": 5228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6790286286515653e-05,
+      "loss": 1.4297,
+      "step": 5229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6789128866692465e-05,
+      "loss": 1.1484,
+      "step": 5230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6787971278133114e-05,
+      "loss": 1.0156,
+      "step": 5231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6786813520866376e-05,
+      "loss": 0.4336,
+      "step": 5232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.678565559492102e-05,
+      "loss": 1.3984,
+      "step": 5233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6784497500325828e-05,
+      "loss": 1.6797,
+      "step": 5234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.678333923710959e-05,
+      "loss": 1.4141,
+      "step": 5235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6782180805301083e-05,
+      "loss": 0.9297,
+      "step": 5236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6781022204929102e-05,
+      "loss": 1.2734,
+      "step": 5237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6779863436022445e-05,
+      "loss": 1.4688,
+      "step": 5238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6778704498609905e-05,
+      "loss": 1.4766,
+      "step": 5239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6777545392720297e-05,
+      "loss": 0.832,
+      "step": 5240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6776386118382418e-05,
+      "loss": 0.9492,
+      "step": 5241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.677522667562509e-05,
+      "loss": 0.9297,
+      "step": 5242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6774067064477123e-05,
+      "loss": 1.2266,
+      "step": 5243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6772907284967344e-05,
+      "loss": 1.1641,
+      "step": 5244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.677174733712457e-05,
+      "loss": 1.0781,
+      "step": 5245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6770587220977638e-05,
+      "loss": 1.0703,
+      "step": 5246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6769426936555375e-05,
+      "loss": 1.3359,
+      "step": 5247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6768266483886623e-05,
+      "loss": 0.7852,
+      "step": 5248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.676710586300022e-05,
+      "loss": 1.1016,
+      "step": 5249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6765945073925015e-05,
+      "loss": 0.8164,
+      "step": 5250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.676478411668986e-05,
+      "loss": 1.4297,
+      "step": 5251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6763622991323604e-05,
+      "loss": 1.9609,
+      "step": 5252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6762461697855108e-05,
+      "loss": 0.582,
+      "step": 5253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6761300236313232e-05,
+      "loss": 0.7148,
+      "step": 5254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.676013860672685e-05,
+      "loss": 1.2734,
+      "step": 5255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.675897680912483e-05,
+      "loss": 1.2422,
+      "step": 5256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.675781484353604e-05,
+      "loss": 0.4766,
+      "step": 5257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6756652709989366e-05,
+      "loss": 0.4375,
+      "step": 5258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6755490408513692e-05,
+      "loss": 0.6289,
+      "step": 5259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6754327939137904e-05,
+      "loss": 0.3672,
+      "step": 5260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6753165301890896e-05,
+      "loss": 0.1807,
+      "step": 5261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.675200249680156e-05,
+      "loss": 0.4219,
+      "step": 5262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6750839523898798e-05,
+      "loss": 0.6445,
+      "step": 5263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6749676383211516e-05,
+      "loss": 0.4395,
+      "step": 5264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.674851307476862e-05,
+      "loss": 0.957,
+      "step": 5265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6747349598599023e-05,
+      "loss": 0.6328,
+      "step": 5266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6746185954731643e-05,
+      "loss": 1.6094,
+      "step": 5267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6745022143195406e-05,
+      "loss": 0.3164,
+      "step": 5268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6743858164019225e-05,
+      "loss": 1.25,
+      "step": 5269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.674269401723204e-05,
+      "loss": 0.9766,
+      "step": 5270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.674152970286278e-05,
+      "loss": 1.1562,
+      "step": 5271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6740365220940383e-05,
+      "loss": 1.3828,
+      "step": 5272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6739200571493798e-05,
+      "loss": 1.8828,
+      "step": 5273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.673803575455196e-05,
+      "loss": 0.9492,
+      "step": 5274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6736870770143823e-05,
+      "loss": 0.7852,
+      "step": 5275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.673570561829834e-05,
+      "loss": 0.3223,
+      "step": 5276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6734540299044477e-05,
+      "loss": 0.6641,
+      "step": 5277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.673337481241119e-05,
+      "loss": 1.5938,
+      "step": 5278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6732209158427447e-05,
+      "loss": 0.7656,
+      "step": 5279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.673104333712222e-05,
+      "loss": 0.3457,
+      "step": 5280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6729877348524484e-05,
+      "loss": 0.6602,
+      "step": 5281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6728711192663215e-05,
+      "loss": 0.9883,
+      "step": 5282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6727544869567404e-05,
+      "loss": 0.6992,
+      "step": 5283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.672637837926603e-05,
+      "loss": 1.4062,
+      "step": 5284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.672521172178809e-05,
+      "loss": 0.8359,
+      "step": 5285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6724044897162578e-05,
+      "loss": 0.9844,
+      "step": 5286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6722877905418497e-05,
+      "loss": 0.8359,
+      "step": 5287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.672171074658485e-05,
+      "loss": 1.3594,
+      "step": 5288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6720543420690644e-05,
+      "loss": 1.1328,
+      "step": 5289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.671937592776489e-05,
+      "loss": 0.4902,
+      "step": 5290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.671820826783661e-05,
+      "loss": 1.2891,
+      "step": 5291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.671704044093482e-05,
+      "loss": 1.4297,
+      "step": 5292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6715872447088547e-05,
+      "loss": 1.9531,
+      "step": 5293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6714704286326818e-05,
+      "loss": 0.8555,
+      "step": 5294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.671353595867867e-05,
+      "loss": 1.3203,
+      "step": 5295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6712367464173143e-05,
+      "loss": 1.3516,
+      "step": 5296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.671119880283927e-05,
+      "loss": 1.6641,
+      "step": 5297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.67100299747061e-05,
+      "loss": 1.2422,
+      "step": 5298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6708860979802688e-05,
+      "loss": 1.2422,
+      "step": 5299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6707691818158083e-05,
+      "loss": 0.7461,
+      "step": 5300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.670652248980134e-05,
+      "loss": 0.8555,
+      "step": 5301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.670535299476153e-05,
+      "loss": 1.0,
+      "step": 5302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6704183333067712e-05,
+      "loss": 0.8906,
+      "step": 5303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6703013504748968e-05,
+      "loss": 0.4688,
+      "step": 5304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6701843509834356e-05,
+      "loss": 0.7891,
+      "step": 5305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6700673348352966e-05,
+      "loss": 0.4785,
+      "step": 5306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.669950302033388e-05,
+      "loss": 2.4688,
+      "step": 5307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.669833252580618e-05,
+      "loss": 0.6094,
+      "step": 5308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6697161864798962e-05,
+      "loss": 0.4512,
+      "step": 5309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6695991037341325e-05,
+      "loss": 1.4141,
+      "step": 5310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.669482004346236e-05,
+      "loss": 0.8945,
+      "step": 5311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6693648883191178e-05,
+      "loss": 0.7109,
+      "step": 5312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.669247755655688e-05,
+      "loss": 0.9141,
+      "step": 5313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.498,
+      "step": 5314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6690134404315403e-05,
+      "loss": 1.125,
+      "step": 5315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6688962578766457e-05,
+      "loss": 1.3359,
+      "step": 5316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6687790586970873e-05,
+      "loss": 0.4258,
+      "step": 5317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.668661842895778e-05,
+      "loss": 0.3125,
+      "step": 5318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6685446104756305e-05,
+      "loss": 0.6133,
+      "step": 5319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.668427361439559e-05,
+      "loss": 0.8203,
+      "step": 5320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6683100957904776e-05,
+      "loss": 1.2812,
+      "step": 5321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6681928135313005e-05,
+      "loss": 0.8867,
+      "step": 5322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6680755146649428e-05,
+      "loss": 1.3047,
+      "step": 5323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6679581991943198e-05,
+      "loss": 1.1094,
+      "step": 5324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.667840867122347e-05,
+      "loss": 0.3926,
+      "step": 5325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.667723518451941e-05,
+      "loss": 0.6875,
+      "step": 5326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6676061531860185e-05,
+      "loss": 0.5508,
+      "step": 5327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.667488771327496e-05,
+      "loss": 0.832,
+      "step": 5328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6673713728792902e-05,
+      "loss": 0.5703,
+      "step": 5329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.667253957844321e-05,
+      "loss": 0.6953,
+      "step": 5330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.667136526225505e-05,
+      "loss": 0.6602,
+      "step": 5331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6670190780257607e-05,
+      "loss": 0.8086,
+      "step": 5332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.666901613248008e-05,
+      "loss": 0.4219,
+      "step": 5333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6667841318951658e-05,
+      "loss": 1.1328,
+      "step": 5334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6666666339701545e-05,
+      "loss": 1.2812,
+      "step": 5335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.666549119475894e-05,
+      "loss": 1.2188,
+      "step": 5336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6664315884153048e-05,
+      "loss": 0.793,
+      "step": 5337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6663140407913086e-05,
+      "loss": 1.0781,
+      "step": 5338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6661964766068262e-05,
+      "loss": 0.6055,
+      "step": 5339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66607889586478e-05,
+      "loss": 1.5078,
+      "step": 5340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6659612985680923e-05,
+      "loss": 0.7266,
+      "step": 5341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6658436847196856e-05,
+      "loss": 1.2422,
+      "step": 5342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6657260543224832e-05,
+      "loss": 1.0156,
+      "step": 5343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6656084073794087e-05,
+      "loss": 1.4922,
+      "step": 5344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6654907438933864e-05,
+      "loss": 1.3516,
+      "step": 5345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66537306386734e-05,
+      "loss": 1.3047,
+      "step": 5346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.665255367304195e-05,
+      "loss": 0.8281,
+      "step": 5347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.665137654206876e-05,
+      "loss": 0.9531,
+      "step": 5348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.665019924578309e-05,
+      "loss": 0.8164,
+      "step": 5349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6649021784214202e-05,
+      "loss": 0.875,
+      "step": 5350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6647844157391353e-05,
+      "loss": 1.125,
+      "step": 5351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.664666636534382e-05,
+      "loss": 0.8516,
+      "step": 5352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6645488408100873e-05,
+      "loss": 1.1172,
+      "step": 5353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.664431028569179e-05,
+      "loss": 1.4922,
+      "step": 5354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6643131998145842e-05,
+      "loss": 1.2656,
+      "step": 5355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6641953545492327e-05,
+      "loss": 0.9297,
+      "step": 5356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.664077492776053e-05,
+      "loss": 0.9844,
+      "step": 5357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6639596144979743e-05,
+      "loss": 1.0,
+      "step": 5358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6638417197179265e-05,
+      "loss": 0.918,
+      "step": 5359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6637238084388393e-05,
+      "loss": 1.5938,
+      "step": 5360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6636058806636438e-05,
+      "loss": 0.6094,
+      "step": 5361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6634879363952705e-05,
+      "loss": 0.9844,
+      "step": 5362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.663369975636651e-05,
+      "loss": 0.707,
+      "step": 5363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6632519983907175e-05,
+      "loss": 1.7109,
+      "step": 5364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6631340046604014e-05,
+      "loss": 1.4219,
+      "step": 5365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6630159944486358e-05,
+      "loss": 1.4375,
+      "step": 5366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6628979677583537e-05,
+      "loss": 1.7969,
+      "step": 5367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662779924592488e-05,
+      "loss": 0.4629,
+      "step": 5368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662661864953973e-05,
+      "loss": 0.8906,
+      "step": 5369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6625437888457432e-05,
+      "loss": 1.6953,
+      "step": 5370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662425696270733e-05,
+      "loss": 1.4844,
+      "step": 5371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662307587231877e-05,
+      "loss": 1.3438,
+      "step": 5372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6621894617321113e-05,
+      "loss": 0.625,
+      "step": 5373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662071319774371e-05,
+      "loss": 0.6094,
+      "step": 5374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6619531613615936e-05,
+      "loss": 1.125,
+      "step": 5375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.661834986496715e-05,
+      "loss": 1.2109,
+      "step": 5376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6617167951826722e-05,
+      "loss": 1.4531,
+      "step": 5377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.661598587422403e-05,
+      "loss": 1.3672,
+      "step": 5378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6614803632188455e-05,
+      "loss": 1.6406,
+      "step": 5379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.661362122574937e-05,
+      "loss": 1.1328,
+      "step": 5380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6612438654936176e-05,
+      "loss": 0.9844,
+      "step": 5381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6611255919778256e-05,
+      "loss": 1.4297,
+      "step": 5382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6610073020305014e-05,
+      "loss": 0.6758,
+      "step": 5383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6608889956545836e-05,
+      "loss": 0.9414,
+      "step": 5384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6607706728530138e-05,
+      "loss": 0.9648,
+      "step": 5385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.660652333628732e-05,
+      "loss": 1.0234,
+      "step": 5386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6605339779846804e-05,
+      "loss": 1.5469,
+      "step": 5387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6604156059237994e-05,
+      "loss": 1.1641,
+      "step": 5388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6602972174490312e-05,
+      "loss": 1.8906,
+      "step": 5389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6601788125633192e-05,
+      "loss": 0.7852,
+      "step": 5390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6600603912696052e-05,
+      "loss": 1.7266,
+      "step": 5391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6599419535708326e-05,
+      "loss": 0.7031,
+      "step": 5392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6598234994699456e-05,
+      "loss": 1.125,
+      "step": 5393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.659705028969888e-05,
+      "loss": 1.0625,
+      "step": 5394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6595865420736036e-05,
+      "loss": 0.9336,
+      "step": 5395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.659468038784038e-05,
+      "loss": 1.2578,
+      "step": 5396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.659349519104136e-05,
+      "loss": 0.8906,
+      "step": 5397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6592309830368436e-05,
+      "loss": 1.1406,
+      "step": 5398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6591124305851068e-05,
+      "loss": 1.4766,
+      "step": 5399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.658993861751872e-05,
+      "loss": 0.7422,
+      "step": 5400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6588752765400867e-05,
+      "loss": 1.1797,
+      "step": 5401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.658756674952697e-05,
+      "loss": 0.6914,
+      "step": 5402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6586380569926518e-05,
+      "loss": 0.3301,
+      "step": 5403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6585194226628983e-05,
+      "loss": 1.4453,
+      "step": 5404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6584007719663856e-05,
+      "loss": 0.3555,
+      "step": 5405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6582821049060623e-05,
+      "loss": 0.6289,
+      "step": 5406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.658163421484878e-05,
+      "loss": 0.4883,
+      "step": 5407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6580447217057824e-05,
+      "loss": 0.9844,
+      "step": 5408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6579260055717254e-05,
+      "loss": 0.5781,
+      "step": 5409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6578072730856578e-05,
+      "loss": 0.2832,
+      "step": 5410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6576885242505305e-05,
+      "loss": 0.1885,
+      "step": 5411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.657569759069295e-05,
+      "loss": 0.7734,
+      "step": 5412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6574509775449027e-05,
+      "loss": 1.0156,
+      "step": 5413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.657332179680306e-05,
+      "loss": 0.9492,
+      "step": 5414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.657213365478458e-05,
+      "loss": 1.4609,
+      "step": 5415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6570945349423107e-05,
+      "loss": 0.1836,
+      "step": 5416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6569756880748184e-05,
+      "loss": 0.8047,
+      "step": 5417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.656856824878934e-05,
+      "loss": 0.7383,
+      "step": 5418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6567379453576124e-05,
+      "loss": 1.4141,
+      "step": 5419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6566190495138083e-05,
+      "loss": 1.5391,
+      "step": 5420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6565001373504763e-05,
+      "loss": 1.1484,
+      "step": 5421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6563812088705717e-05,
+      "loss": 1.0547,
+      "step": 5422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.656262264077051e-05,
+      "loss": 1.2422,
+      "step": 5423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.65614330297287e-05,
+      "loss": 1.0078,
+      "step": 5424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6560243255609853e-05,
+      "loss": 0.9531,
+      "step": 5425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6559053318443538e-05,
+      "loss": 1.1172,
+      "step": 5426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6557863218259335e-05,
+      "loss": 1.4922,
+      "step": 5427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.655667295508682e-05,
+      "loss": 1.2031,
+      "step": 5428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6555482528955574e-05,
+      "loss": 1.9141,
+      "step": 5429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6554291939895186e-05,
+      "loss": 1.375,
+      "step": 5430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6553101187935244e-05,
+      "loss": 1.375,
+      "step": 5431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6551910273105347e-05,
+      "loss": 1.3047,
+      "step": 5432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.655071919543509e-05,
+      "loss": 0.7734,
+      "step": 5433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6549527954954078e-05,
+      "loss": 0.377,
+      "step": 5434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6548336551691916e-05,
+      "loss": 1.7344,
+      "step": 5435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6547144985678217e-05,
+      "loss": 0.6523,
+      "step": 5436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6545953256942598e-05,
+      "loss": 0.8516,
+      "step": 5437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.654476136551467e-05,
+      "loss": 1.0,
+      "step": 5438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6543569311424064e-05,
+      "loss": 0.5742,
+      "step": 5439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6542377094700404e-05,
+      "loss": 0.9297,
+      "step": 5440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6541184715373324e-05,
+      "loss": 1.4062,
+      "step": 5441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6539992173472455e-05,
+      "loss": 1.5234,
+      "step": 5442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6538799469027438e-05,
+      "loss": 1.4297,
+      "step": 5443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6537606602067918e-05,
+      "loss": 1.3281,
+      "step": 5444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.653641357262354e-05,
+      "loss": 1.125,
+      "step": 5445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6535220380723958e-05,
+      "loss": 1.2891,
+      "step": 5446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6534027026398824e-05,
+      "loss": 0.875,
+      "step": 5447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.65328335096778e-05,
+      "loss": 0.9219,
+      "step": 5448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.653163983059055e-05,
+      "loss": 0.5508,
+      "step": 5449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6530445989166736e-05,
+      "loss": 0.8398,
+      "step": 5450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.652925198543604e-05,
+      "loss": 1.1016,
+      "step": 5451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6528057819428127e-05,
+      "loss": 0.4961,
+      "step": 5452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6526863491172683e-05,
+      "loss": 0.543,
+      "step": 5453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6525669000699394e-05,
+      "loss": 0.5938,
+      "step": 5454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6524474348037937e-05,
+      "loss": 1.9062,
+      "step": 5455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6523279533218015e-05,
+      "loss": 0.9062,
+      "step": 5456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6522084556269317e-05,
+      "loss": 0.4023,
+      "step": 5457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6520889417221545e-05,
+      "loss": 0.8008,
+      "step": 5458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6519694116104403e-05,
+      "loss": 0.4121,
+      "step": 5459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6518498652947605e-05,
+      "loss": 0.3965,
+      "step": 5460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6517303027780848e-05,
+      "loss": 0.9609,
+      "step": 5461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.651610724063386e-05,
+      "loss": 0.8711,
+      "step": 5462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6514911291536354e-05,
+      "loss": 1.6562,
+      "step": 5463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.651371518051806e-05,
+      "loss": 0.6797,
+      "step": 5464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.651251890760871e-05,
+      "loss": 0.5586,
+      "step": 5465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6511322472838022e-05,
+      "loss": 1.4766,
+      "step": 5466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6510125876235738e-05,
+      "loss": 1.0625,
+      "step": 5467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6508929117831603e-05,
+      "loss": 0.6484,
+      "step": 5468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6507732197655357e-05,
+      "loss": 1.625,
+      "step": 5469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6506535115736747e-05,
+      "loss": 1.4062,
+      "step": 5470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.650533787210553e-05,
+      "loss": 0.9258,
+      "step": 5471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6504140466791458e-05,
+      "loss": 1.0859,
+      "step": 5472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.650294289982429e-05,
+      "loss": 1.5547,
+      "step": 5473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6501745171233797e-05,
+      "loss": 0.8398,
+      "step": 5474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6500547281049737e-05,
+      "loss": 1.0234,
+      "step": 5475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6499349229301892e-05,
+      "loss": 1.2031,
+      "step": 5476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6498151016020036e-05,
+      "loss": 1.0156,
+      "step": 5477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6496952641233943e-05,
+      "loss": 0.6055,
+      "step": 5478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6495754104973403e-05,
+      "loss": 1.0938,
+      "step": 5479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.64945554072682e-05,
+      "loss": 0.7773,
+      "step": 5480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.649335654814813e-05,
+      "loss": 1.6562,
+      "step": 5481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6492157527642993e-05,
+      "loss": 0.9414,
+      "step": 5482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.649095834578258e-05,
+      "loss": 1.0469,
+      "step": 5483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6489759002596702e-05,
+      "loss": 1.6875,
+      "step": 5484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6488559498115165e-05,
+      "loss": 1.0859,
+      "step": 5485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6487359832367776e-05,
+      "loss": 0.3867,
+      "step": 5486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6486160005384367e-05,
+      "loss": 0.8828,
+      "step": 5487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.648496001719474e-05,
+      "loss": 1.375,
+      "step": 5488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6483759867828732e-05,
+      "loss": 1.2031,
+      "step": 5489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6482559557316168e-05,
+      "loss": 0.8242,
+      "step": 5490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6481359085686875e-05,
+      "loss": 1.3672,
+      "step": 5491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.64801584529707e-05,
+      "loss": 0.8633,
+      "step": 5492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.647895765919747e-05,
+      "loss": 0.9102,
+      "step": 5493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.647775670439704e-05,
+      "loss": 0.9844,
+      "step": 5494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6476555588599256e-05,
+      "loss": 0.8242,
+      "step": 5495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.647535431183397e-05,
+      "loss": 1.2266,
+      "step": 5496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6474152874131033e-05,
+      "loss": 0.7539,
+      "step": 5497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6472951275520315e-05,
+      "loss": 1.3906,
+      "step": 5498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6471749516031673e-05,
+      "loss": 0.5469,
+      "step": 5499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6470547595694978e-05,
+      "loss": 1.2266,
+      "step": 5500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6469345514540103e-05,
+      "loss": 1.5234,
+      "step": 5501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6468143272596924e-05,
+      "loss": 0.2139,
+      "step": 5502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6466940869895318e-05,
+      "loss": 1.2109,
+      "step": 5503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6465738306465175e-05,
+      "loss": 0.4863,
+      "step": 5504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6464535582336378e-05,
+      "loss": 0.9648,
+      "step": 5505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6463332697538828e-05,
+      "loss": 1.1484,
+      "step": 5506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.646212965210241e-05,
+      "loss": 1.7812,
+      "step": 5507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.646092644605703e-05,
+      "loss": 0.4727,
+      "step": 5508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6459723079432592e-05,
+      "loss": 1.1797,
+      "step": 5509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6458519552259007e-05,
+      "loss": 0.7812,
+      "step": 5510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.645731586456618e-05,
+      "loss": 0.4941,
+      "step": 5511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.645611201638404e-05,
+      "loss": 0.6836,
+      "step": 5512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6454908007742494e-05,
+      "loss": 0.9023,
+      "step": 5513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6453703838671468e-05,
+      "loss": 1.4922,
+      "step": 5514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.64524995092009e-05,
+      "loss": 0.9648,
+      "step": 5515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6451295019360713e-05,
+      "loss": 0.5078,
+      "step": 5516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.645009036918085e-05,
+      "loss": 0.4531,
+      "step": 5517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6448885558691243e-05,
+      "loss": 0.8125,
+      "step": 5518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6447680587921842e-05,
+      "loss": 1.2891,
+      "step": 5519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6446475456902597e-05,
+      "loss": 1.0547,
+      "step": 5520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6445270165663454e-05,
+      "loss": 0.5977,
+      "step": 5521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6444064714234372e-05,
+      "loss": 0.7695,
+      "step": 5522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6442859102645315e-05,
+      "loss": 1.4844,
+      "step": 5523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.644165333092624e-05,
+      "loss": 1.1953,
+      "step": 5524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.644044739910712e-05,
+      "loss": 0.2734,
+      "step": 5525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6439241307217924e-05,
+      "loss": 1.2422,
+      "step": 5526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.643803505528863e-05,
+      "loss": 1.8047,
+      "step": 5527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.643682864334922e-05,
+      "loss": 0.9414,
+      "step": 5528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6435622071429674e-05,
+      "loss": 0.6953,
+      "step": 5529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6434415339559985e-05,
+      "loss": 0.9141,
+      "step": 5530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.643320844777014e-05,
+      "loss": 1.0391,
+      "step": 5531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.643200139609013e-05,
+      "loss": 0.7188,
+      "step": 5532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6430794184549972e-05,
+      "loss": 1.2344,
+      "step": 5533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6429586813179654e-05,
+      "loss": 0.7812,
+      "step": 5534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.642837928200919e-05,
+      "loss": 0.4766,
+      "step": 5535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.642717159106859e-05,
+      "loss": 0.8047,
+      "step": 5536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6425963740387874e-05,
+      "loss": 0.582,
+      "step": 5537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6424755729997056e-05,
+      "loss": 0.8867,
+      "step": 5538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6423547559926168e-05,
+      "loss": 0.4863,
+      "step": 5539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6422339230205226e-05,
+      "loss": 0.8672,
+      "step": 5540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6421130740864272e-05,
+      "loss": 1.2422,
+      "step": 5541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6419922091933333e-05,
+      "loss": 1.1016,
+      "step": 5542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6418713283442457e-05,
+      "loss": 0.8516,
+      "step": 5543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.641750431542168e-05,
+      "loss": 1.1797,
+      "step": 5544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.641629518790106e-05,
+      "loss": 1.1875,
+      "step": 5545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6415085900910637e-05,
+      "loss": 1.7734,
+      "step": 5546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.641387645448047e-05,
+      "loss": 1.1953,
+      "step": 5547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6412666848640622e-05,
+      "loss": 0.8203,
+      "step": 5548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6411457083421154e-05,
+      "loss": 0.8789,
+      "step": 5549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.641024715885213e-05,
+      "loss": 0.9844,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.640903707496363e-05,
+      "loss": 0.2275,
+      "step": 5551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6407826831785724e-05,
+      "loss": 1.0547,
+      "step": 5552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.640661642934849e-05,
+      "loss": 0.5039,
+      "step": 5553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.640540586768201e-05,
+      "loss": 0.8984,
+      "step": 5554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.640419514681637e-05,
+      "loss": 1.1172,
+      "step": 5555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.640298426678167e-05,
+      "loss": 1.6016,
+      "step": 5556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6401773227608e-05,
+      "loss": 1.1016,
+      "step": 5557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.640056202932546e-05,
+      "loss": 0.3965,
+      "step": 5558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6399350671964147e-05,
+      "loss": 0.5039,
+      "step": 5559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6398139155554177e-05,
+      "loss": 1.1719,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6396927480125657e-05,
+      "loss": 0.2637,
+      "step": 5561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6395715645708697e-05,
+      "loss": 1.5469,
+      "step": 5562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.639450365233342e-05,
+      "loss": 0.7852,
+      "step": 5563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6393291500029952e-05,
+      "loss": 1.4453,
+      "step": 5564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.639207918882842e-05,
+      "loss": 0.8125,
+      "step": 5565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6390866718758945e-05,
+      "loss": 1.1172,
+      "step": 5566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.638965408985167e-05,
+      "loss": 1.3672,
+      "step": 5567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.638844130213673e-05,
+      "loss": 0.2441,
+      "step": 5568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6387228355644267e-05,
+      "loss": 1.0078,
+      "step": 5569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6386015250404433e-05,
+      "loss": 0.6992,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6384801986447373e-05,
+      "loss": 1.1016,
+      "step": 5571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6383588563803243e-05,
+      "loss": 1.0547,
+      "step": 5572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6382374982502203e-05,
+      "loss": 0.4629,
+      "step": 5573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6381161242574412e-05,
+      "loss": 1.3906,
+      "step": 5574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6379947344050035e-05,
+      "loss": 1.4922,
+      "step": 5575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6378733286959248e-05,
+      "loss": 0.918,
+      "step": 5576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.637751907133222e-05,
+      "loss": 1.5547,
+      "step": 5577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.637630469719913e-05,
+      "loss": 0.6445,
+      "step": 5578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.637509016459016e-05,
+      "loss": 0.5625,
+      "step": 5579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6373875473535498e-05,
+      "loss": 0.7305,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6372660624065334e-05,
+      "loss": 1.125,
+      "step": 5581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6371445616209856e-05,
+      "loss": 1.1406,
+      "step": 5582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6370230449999264e-05,
+      "loss": 0.5742,
+      "step": 5583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6369015125463767e-05,
+      "loss": 0.8516,
+      "step": 5584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6367799642633556e-05,
+      "loss": 0.5391,
+      "step": 5585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6366584001538855e-05,
+      "loss": 0.9961,
+      "step": 5586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.636536820220987e-05,
+      "loss": 0.6484,
+      "step": 5587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6364152244676824e-05,
+      "loss": 1.375,
+      "step": 5588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.636293612896993e-05,
+      "loss": 1.0703,
+      "step": 5589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6361719855119417e-05,
+      "loss": 0.8906,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.636050342315551e-05,
+      "loss": 1.4609,
+      "step": 5591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6359286833108453e-05,
+      "loss": 0.4883,
+      "step": 5592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6358070085008474e-05,
+      "loss": 1.1562,
+      "step": 5593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6356853178885817e-05,
+      "loss": 1.3359,
+      "step": 5594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6355636114770726e-05,
+      "loss": 0.9453,
+      "step": 5595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6354418892693447e-05,
+      "loss": 0.9688,
+      "step": 5596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.635320151268424e-05,
+      "loss": 1.2344,
+      "step": 5597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6351983974773355e-05,
+      "loss": 0.7695,
+      "step": 5598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.635076627899105e-05,
+      "loss": 1.2109,
+      "step": 5599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.63495484253676e-05,
+      "loss": 0.6562,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6348330413933268e-05,
+      "loss": 1.3438,
+      "step": 5601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6347112244718323e-05,
+      "loss": 0.3848,
+      "step": 5602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6345893917753046e-05,
+      "loss": 1.1484,
+      "step": 5603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6344675433067713e-05,
+      "loss": 1.0,
+      "step": 5604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6343456790692608e-05,
+      "loss": 0.6875,
+      "step": 5605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6342237990658024e-05,
+      "loss": 0.5,
+      "step": 5606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.634101903299425e-05,
+      "loss": 1.8516,
+      "step": 5607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6339799917731582e-05,
+      "loss": 0.8008,
+      "step": 5608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6338580644900317e-05,
+      "loss": 0.957,
+      "step": 5609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6337361214530766e-05,
+      "loss": 1.0859,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.633614162665323e-05,
+      "loss": 0.8711,
+      "step": 5611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.633492188129802e-05,
+      "loss": 0.7734,
+      "step": 5612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6333701978495454e-05,
+      "loss": 1.2656,
+      "step": 5613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6332481918275856e-05,
+      "loss": 1.0391,
+      "step": 5614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.633126170066954e-05,
+      "loss": 0.2598,
+      "step": 5615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.633004132570684e-05,
+      "loss": 0.582,
+      "step": 5616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6328820793418083e-05,
+      "loss": 0.7695,
+      "step": 5617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6327600103833604e-05,
+      "loss": 1.25,
+      "step": 5618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6326379256983744e-05,
+      "loss": 1.4844,
+      "step": 5619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6325158252898847e-05,
+      "loss": 0.8516,
+      "step": 5620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6323937091609255e-05,
+      "loss": 1.2812,
+      "step": 5621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6322715773145325e-05,
+      "loss": 0.8711,
+      "step": 5622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6321494297537402e-05,
+      "loss": 1.3203,
+      "step": 5623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6320272664815856e-05,
+      "loss": 1.0938,
+      "step": 5624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6319050875011043e-05,
+      "loss": 1.1641,
+      "step": 5625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6317828928153328e-05,
+      "loss": 0.9531,
+      "step": 5626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6316606824273083e-05,
+      "loss": 0.9844,
+      "step": 5627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.631538456340068e-05,
+      "loss": 0.6602,
+      "step": 5628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.63141621455665e-05,
+      "loss": 1.4062,
+      "step": 5629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6312939570800925e-05,
+      "loss": 1.125,
+      "step": 5630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.631171683913434e-05,
+      "loss": 0.668,
+      "step": 5631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.631049395059713e-05,
+      "loss": 0.5391,
+      "step": 5632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6309270905219697e-05,
+      "loss": 0.707,
+      "step": 5633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6308047703032436e-05,
+      "loss": 0.832,
+      "step": 5634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.630682434406574e-05,
+      "loss": 0.8047,
+      "step": 5635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.630560082835002e-05,
+      "loss": 1.0859,
+      "step": 5636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.630437715591569e-05,
+      "loss": 0.4102,
+      "step": 5637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6303153326793156e-05,
+      "loss": 0.1924,
+      "step": 5638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.630192934101284e-05,
+      "loss": 0.6484,
+      "step": 5639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.630070519860516e-05,
+      "loss": 1.4141,
+      "step": 5640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6299480899600534e-05,
+      "loss": 0.8125,
+      "step": 5641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6298256444029405e-05,
+      "loss": 1.1875,
+      "step": 5642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6297031831922193e-05,
+      "loss": 0.9219,
+      "step": 5643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6295807063309346e-05,
+      "loss": 1.0938,
+      "step": 5644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.629458213822129e-05,
+      "loss": 0.7109,
+      "step": 5645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.629335705668848e-05,
+      "loss": 0.6016,
+      "step": 5646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6292131818741363e-05,
+      "loss": 1.0,
+      "step": 5647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6290906424410388e-05,
+      "loss": 0.957,
+      "step": 5648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6289680873726008e-05,
+      "loss": 1.6562,
+      "step": 5649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.628845516671869e-05,
+      "loss": 1.1016,
+      "step": 5650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6287229303418892e-05,
+      "loss": 1.6641,
+      "step": 5651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6286003283857083e-05,
+      "loss": 1.2109,
+      "step": 5652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6284777108063732e-05,
+      "loss": 1.0781,
+      "step": 5653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.628355077606932e-05,
+      "loss": 0.3379,
+      "step": 5654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.628232428790432e-05,
+      "loss": 0.3164,
+      "step": 5655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6281097643599222e-05,
+      "loss": 0.373,
+      "step": 5656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6279870843184505e-05,
+      "loss": 1.0,
+      "step": 5657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6278643886690665e-05,
+      "loss": 0.9336,
+      "step": 5658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6277416774148193e-05,
+      "loss": 0.6484,
+      "step": 5659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.627618950558759e-05,
+      "loss": 0.4688,
+      "step": 5660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.627496208103936e-05,
+      "loss": 0.4336,
+      "step": 5661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6273734500534008e-05,
+      "loss": 0.3145,
+      "step": 5662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6272506764102042e-05,
+      "loss": 0.8359,
+      "step": 5663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6271278871773974e-05,
+      "loss": 0.7148,
+      "step": 5664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6270050823580328e-05,
+      "loss": 0.5508,
+      "step": 5665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6268822619551623e-05,
+      "loss": 1.0625,
+      "step": 5666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6267594259718384e-05,
+      "loss": 1.2578,
+      "step": 5667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6266365744111138e-05,
+      "loss": 1.6875,
+      "step": 5668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6265137072760427e-05,
+      "loss": 1.7109,
+      "step": 5669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.626390824569678e-05,
+      "loss": 1.2188,
+      "step": 5670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.626267926295074e-05,
+      "loss": 0.4941,
+      "step": 5671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6261450124552854e-05,
+      "loss": 0.9922,
+      "step": 5672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.626022083053367e-05,
+      "loss": 0.9258,
+      "step": 5673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.625899138092374e-05,
+      "loss": 0.875,
+      "step": 5674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.625776177575362e-05,
+      "loss": 0.6328,
+      "step": 5675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6256532015053873e-05,
+      "loss": 0.9297,
+      "step": 5676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.625530209885506e-05,
+      "loss": 1.5156,
+      "step": 5677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.625407202718775e-05,
+      "loss": 1.0625,
+      "step": 5678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6252841800082518e-05,
+      "loss": 0.4883,
+      "step": 5679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6251611417569935e-05,
+      "loss": 0.5703,
+      "step": 5680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6250380879680587e-05,
+      "loss": 1.2578,
+      "step": 5681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6249150186445054e-05,
+      "loss": 1.0078,
+      "step": 5682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6247919337893924e-05,
+      "loss": 1.2188,
+      "step": 5683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6246688334057785e-05,
+      "loss": 1.2812,
+      "step": 5684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6245457174967237e-05,
+      "loss": 1.5703,
+      "step": 5685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6244225860652876e-05,
+      "loss": 0.9688,
+      "step": 5686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6242994391145307e-05,
+      "loss": 1.3359,
+      "step": 5687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6241762766475137e-05,
+      "loss": 1.75,
+      "step": 5688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6240530986672976e-05,
+      "loss": 0.9023,
+      "step": 5689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6239299051769437e-05,
+      "loss": 1.2188,
+      "step": 5690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6238066961795144e-05,
+      "loss": 1.2969,
+      "step": 5691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.623683471678071e-05,
+      "loss": 1.1719,
+      "step": 5692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6235602316756766e-05,
+      "loss": 1.0391,
+      "step": 5693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6234369761753945e-05,
+      "loss": 1.2812,
+      "step": 5694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6233137051802875e-05,
+      "loss": 0.5938,
+      "step": 5695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6231904186934192e-05,
+      "loss": 0.8516,
+      "step": 5696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6230671167178548e-05,
+      "loss": 0.9258,
+      "step": 5697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6229437992566583e-05,
+      "loss": 0.8867,
+      "step": 5698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.622820466312894e-05,
+      "loss": 0.8945,
+      "step": 5699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.622697117889628e-05,
+      "loss": 0.8398,
+      "step": 5700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.622573753989926e-05,
+      "loss": 1.0312,
+      "step": 5701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.622450374616853e-05,
+      "loss": 1.8359,
+      "step": 5702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.622326979773477e-05,
+      "loss": 0.7422,
+      "step": 5703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6222035694628634e-05,
+      "loss": 0.5195,
+      "step": 5704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6220801436880805e-05,
+      "loss": 0.8555,
+      "step": 5705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6219567024521952e-05,
+      "loss": 1.2812,
+      "step": 5706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.621833245758276e-05,
+      "loss": 0.875,
+      "step": 5707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.621709773609391e-05,
+      "loss": 1.1328,
+      "step": 5708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6215862860086092e-05,
+      "loss": 0.6367,
+      "step": 5709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6214627829589994e-05,
+      "loss": 1.2109,
+      "step": 5710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6213392644636313e-05,
+      "loss": 1.2812,
+      "step": 5711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6212157305255752e-05,
+      "loss": 0.6172,
+      "step": 5712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.621092181147901e-05,
+      "loss": 0.75,
+      "step": 5713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6209686163336785e-05,
+      "loss": 1.1719,
+      "step": 5714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6208450360859806e-05,
+      "loss": 0.1973,
+      "step": 5715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6207214404078773e-05,
+      "loss": 1.5703,
+      "step": 5716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.620597829302441e-05,
+      "loss": 1.4219,
+      "step": 5717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6204742027727443e-05,
+      "loss": 0.4629,
+      "step": 5718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.620350560821859e-05,
+      "loss": 0.3652,
+      "step": 5719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6202269034528584e-05,
+      "loss": 0.8828,
+      "step": 5720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.620103230668816e-05,
+      "loss": 1.4688,
+      "step": 5721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6199795424728053e-05,
+      "loss": 0.207,
+      "step": 5722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6198558388679007e-05,
+      "loss": 2.0312,
+      "step": 5723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6197321198571766e-05,
+      "loss": 1.1016,
+      "step": 5724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.619608385443708e-05,
+      "loss": 1.4844,
+      "step": 5725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.61948463563057e-05,
+      "loss": 1.3125,
+      "step": 5726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.619360870420838e-05,
+      "loss": 1.5156,
+      "step": 5727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6192370898175888e-05,
+      "loss": 1.3594,
+      "step": 5728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6191132938238984e-05,
+      "loss": 1.1172,
+      "step": 5729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6189894824428436e-05,
+      "loss": 1.625,
+      "step": 5730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6188656556775013e-05,
+      "loss": 0.8164,
+      "step": 5731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6187418135309494e-05,
+      "loss": 1.4922,
+      "step": 5732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6186179560062664e-05,
+      "loss": 1.1094,
+      "step": 5733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6184940831065293e-05,
+      "loss": 1.0469,
+      "step": 5734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6183701948348183e-05,
+      "loss": 1.5156,
+      "step": 5735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6182462911942114e-05,
+      "loss": 0.8164,
+      "step": 5736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6181223721877886e-05,
+      "loss": 0.5586,
+      "step": 5737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6179984378186296e-05,
+      "loss": 0.8516,
+      "step": 5738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.617874488089815e-05,
+      "loss": 1.5938,
+      "step": 5739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6177505230044248e-05,
+      "loss": 1.7422,
+      "step": 5740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6176265425655405e-05,
+      "loss": 1.0234,
+      "step": 5741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6175025467762434e-05,
+      "loss": 1.1328,
+      "step": 5742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.617378535639615e-05,
+      "loss": 1.2578,
+      "step": 5743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.617254509158738e-05,
+      "loss": 0.7188,
+      "step": 5744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6171304673366944e-05,
+      "loss": 0.8281,
+      "step": 5745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6170064101765673e-05,
+      "loss": 0.9336,
+      "step": 5746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.61688233768144e-05,
+      "loss": 0.9648,
+      "step": 5747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6167582498543964e-05,
+      "loss": 0.9531,
+      "step": 5748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6166341466985202e-05,
+      "loss": 1.0547,
+      "step": 5749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.616510028216896e-05,
+      "loss": 0.9648,
+      "step": 5750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.616385894412609e-05,
+      "loss": 1.1406,
+      "step": 5751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.616261745288744e-05,
+      "loss": 1.5469,
+      "step": 5752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.616137580848386e-05,
+      "loss": 1.0,
+      "step": 5753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.616013401094622e-05,
+      "loss": 0.6211,
+      "step": 5754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.615889206030538e-05,
+      "loss": 0.3223,
+      "step": 5755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.615764995659221e-05,
+      "loss": 0.582,
+      "step": 5756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6156407699837577e-05,
+      "loss": 0.4355,
+      "step": 5757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6155165290072354e-05,
+      "loss": 0.8203,
+      "step": 5758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6153922727327424e-05,
+      "loss": 0.4434,
+      "step": 5759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.615268001163367e-05,
+      "loss": 1.0156,
+      "step": 5760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6151437143021974e-05,
+      "loss": 0.1357,
+      "step": 5761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6150194121523228e-05,
+      "loss": 1.6562,
+      "step": 5762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.614895094716833e-05,
+      "loss": 0.3848,
+      "step": 5763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.614770761998817e-05,
+      "loss": 0.2441,
+      "step": 5764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6146464140013652e-05,
+      "loss": 1.1406,
+      "step": 5765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.614522050727569e-05,
+      "loss": 0.5898,
+      "step": 5766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6143976721805178e-05,
+      "loss": 0.6523,
+      "step": 5767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6142732783633044e-05,
+      "loss": 1.5,
+      "step": 5768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6141488692790194e-05,
+      "loss": 1.6641,
+      "step": 5769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.614024444930755e-05,
+      "loss": 1.9219,
+      "step": 5770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.613900005321604e-05,
+      "loss": 1.0391,
+      "step": 5771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6137755504546593e-05,
+      "loss": 1.3594,
+      "step": 5772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6136510803330138e-05,
+      "loss": 1.0547,
+      "step": 5773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6135265949597608e-05,
+      "loss": 0.8555,
+      "step": 5774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6134020943379946e-05,
+      "loss": 1.7422,
+      "step": 5775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.613277578470809e-05,
+      "loss": 1.1328,
+      "step": 5776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6131530473613e-05,
+      "loss": 0.498,
+      "step": 5777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.6130285010125618e-05,
+      "loss": 1.2031,
+      "step": 5778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6129039394276894e-05,
+      "loss": 0.5078,
+      "step": 5779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6127793626097795e-05,
+      "loss": 1.6094,
+      "step": 5780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6126547705619273e-05,
+      "loss": 0.8594,
+      "step": 5781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6125301632872307e-05,
+      "loss": 0.8164,
+      "step": 5782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.612405540788786e-05,
+      "loss": 1.4375,
+      "step": 5783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6122809030696903e-05,
+      "loss": 0.9531,
+      "step": 5784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6121562501330417e-05,
+      "loss": 0.9648,
+      "step": 5785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.612031581981938e-05,
+      "loss": 0.6992,
+      "step": 5786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.611906898619478e-05,
+      "loss": 1.6875,
+      "step": 5787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.611782200048761e-05,
+      "loss": 0.6953,
+      "step": 5788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.611657486272885e-05,
+      "loss": 0.9297,
+      "step": 5789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6115327572949506e-05,
+      "loss": 0.8906,
+      "step": 5790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6114080131180573e-05,
+      "loss": 1.4531,
+      "step": 5791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.611283253745306e-05,
+      "loss": 0.8242,
+      "step": 5792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6111584791797972e-05,
+      "loss": 0.9258,
+      "step": 5793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.611033689424632e-05,
+      "loss": 0.8477,
+      "step": 5794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6109088844829115e-05,
+      "loss": 1.4609,
+      "step": 5795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6107840643577383e-05,
+      "loss": 1.2891,
+      "step": 5796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6106592290522144e-05,
+      "loss": 0.8477,
+      "step": 5797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.610534378569442e-05,
+      "loss": 1.0703,
+      "step": 5798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6104095129125247e-05,
+      "loss": 0.6602,
+      "step": 5799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.610284632084566e-05,
+      "loss": 0.9766,
+      "step": 5800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.610159736088669e-05,
+      "loss": 1.4375,
+      "step": 5801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6100348249279384e-05,
+      "loss": 0.4004,
+      "step": 5802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6099098986054785e-05,
+      "loss": 0.6641,
+      "step": 5803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6097849571243944e-05,
+      "loss": 1.6172,
+      "step": 5804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6096600004877908e-05,
+      "loss": 1.7891,
+      "step": 5805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.609535028698774e-05,
+      "loss": 0.4512,
+      "step": 5806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6094100417604498e-05,
+      "loss": 0.8516,
+      "step": 5807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6092850396759248e-05,
+      "loss": 0.4609,
+      "step": 5808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6091600224483054e-05,
+      "loss": 1.1016,
+      "step": 5809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.609034990080699e-05,
+      "loss": 0.7305,
+      "step": 5810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6089099425762136e-05,
+      "loss": 0.6992,
+      "step": 5811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.608784879937956e-05,
+      "loss": 0.7383,
+      "step": 5812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6086598021690356e-05,
+      "loss": 0.8984,
+      "step": 5813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6085347092725603e-05,
+      "loss": 0.6172,
+      "step": 5814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6084096012516396e-05,
+      "loss": 0.9141,
+      "step": 5815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6082844781093826e-05,
+      "loss": 0.6797,
+      "step": 5816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.608159339848899e-05,
+      "loss": 0.9453,
+      "step": 5817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6080341864732996e-05,
+      "loss": 1.6641,
+      "step": 5818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6079090179856946e-05,
+      "loss": 1.2656,
+      "step": 5819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.607783834389195e-05,
+      "loss": 0.9727,
+      "step": 5820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6076586356869116e-05,
+      "loss": 1.0156,
+      "step": 5821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6075334218819566e-05,
+      "loss": 1.25,
+      "step": 5822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6074081929774416e-05,
+      "loss": 2.0312,
+      "step": 5823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.60728294897648e-05,
+      "loss": 0.5312,
+      "step": 5824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6071576898821834e-05,
+      "loss": 0.957,
+      "step": 5825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6070324156976655e-05,
+      "loss": 1.4219,
+      "step": 5826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.60690712642604e-05,
+      "loss": 1.1328,
+      "step": 5827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6067818220704207e-05,
+      "loss": 0.5859,
+      "step": 5828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6066565026339222e-05,
+      "loss": 1.0078,
+      "step": 5829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.606531168119658e-05,
+      "loss": 1.6328,
+      "step": 5830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6064058185307447e-05,
+      "loss": 0.7578,
+      "step": 5831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.606280453870297e-05,
+      "loss": 0.7773,
+      "step": 5832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6061550741414304e-05,
+      "loss": 0.9609,
+      "step": 5833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6060296793472613e-05,
+      "loss": 1.2734,
+      "step": 5834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6059042694909067e-05,
+      "loss": 1.7031,
+      "step": 5835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.605778844575483e-05,
+      "loss": 1.3516,
+      "step": 5836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6056534046041076e-05,
+      "loss": 0.7656,
+      "step": 5837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6055279495798983e-05,
+      "loss": 0.4668,
+      "step": 5838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6054024795059732e-05,
+      "loss": 0.832,
+      "step": 5839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6052769943854504e-05,
+      "loss": 1.1797,
+      "step": 5840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.605151494221449e-05,
+      "loss": 0.5195,
+      "step": 5841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6050259790170876e-05,
+      "loss": 1.6172,
+      "step": 5842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6049004487754865e-05,
+      "loss": 0.9648,
+      "step": 5843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6047749034997653e-05,
+      "loss": 1.2578,
+      "step": 5844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6046493431930442e-05,
+      "loss": 1.1094,
+      "step": 5845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.604523767858444e-05,
+      "loss": 1.0312,
+      "step": 5846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.604398177499086e-05,
+      "loss": 1.1641,
+      "step": 5847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6042725721180907e-05,
+      "loss": 1.125,
+      "step": 5848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.604146951718581e-05,
+      "loss": 0.6328,
+      "step": 5849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.604021316303678e-05,
+      "loss": 0.7109,
+      "step": 5850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6038956658765048e-05,
+      "loss": 1.3906,
+      "step": 5851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6037700004401845e-05,
+      "loss": 1.1094,
+      "step": 5852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.60364431999784e-05,
+      "loss": 0.3633,
+      "step": 5853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6035186245525948e-05,
+      "loss": 0.9414,
+      "step": 5854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6033929141075738e-05,
+      "loss": 0.6406,
+      "step": 5855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6032671886659002e-05,
+      "loss": 1.1953,
+      "step": 5856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6031414482306996e-05,
+      "loss": 0.5273,
+      "step": 5857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.603015692805097e-05,
+      "loss": 0.9453,
+      "step": 5858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6028899223922174e-05,
+      "loss": 0.5234,
+      "step": 5859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6027641369951873e-05,
+      "loss": 0.209,
+      "step": 5860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6026383366171326e-05,
+      "loss": 0.8398,
+      "step": 5861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.60251252126118e-05,
+      "loss": 0.7031,
+      "step": 5862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6023866909304568e-05,
+      "loss": 1.0234,
+      "step": 5863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.60226084562809e-05,
+      "loss": 0.9805,
+      "step": 5864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6021349853572075e-05,
+      "loss": 0.7852,
+      "step": 5865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6020091101209373e-05,
+      "loss": 0.9609,
+      "step": 5866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6018832199224078e-05,
+      "loss": 1.5234,
+      "step": 5867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6017573147647482e-05,
+      "loss": 0.4648,
+      "step": 5868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.601631394651087e-05,
+      "loss": 1.3359,
+      "step": 5869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6015054595845553e-05,
+      "loss": 0.4062,
+      "step": 5870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6013795095682816e-05,
+      "loss": 0.8398,
+      "step": 5871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6012535446053966e-05,
+      "loss": 1.6094,
+      "step": 5872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6011275646990316e-05,
+      "loss": 1.125,
+      "step": 5873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.601001569852317e-05,
+      "loss": 1.4844,
+      "step": 5874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.600875560068385e-05,
+      "loss": 1.0,
+      "step": 5875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6007495353503663e-05,
+      "loss": 0.7891,
+      "step": 5876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6006234957013944e-05,
+      "loss": 0.4863,
+      "step": 5877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6004974411246008e-05,
+      "loss": 0.7656,
+      "step": 5878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.600371371623119e-05,
+      "loss": 2.0312,
+      "step": 5879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6002452872000822e-05,
+      "loss": 1.5625,
+      "step": 5880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6001191878586243e-05,
+      "loss": 0.957,
+      "step": 5881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.599993073601879e-05,
+      "loss": 1.3125,
+      "step": 5882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.599866944432981e-05,
+      "loss": 0.8242,
+      "step": 5883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5997408003550646e-05,
+      "loss": 1.0703,
+      "step": 5884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.599614641371266e-05,
+      "loss": 0.9727,
+      "step": 5885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5994884674847196e-05,
+      "loss": 0.7031,
+      "step": 5886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5993622786985615e-05,
+      "loss": 1.1406,
+      "step": 5887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5992360750159287e-05,
+      "loss": 1.375,
+      "step": 5888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5991098564399575e-05,
+      "loss": 0.8672,
+      "step": 5889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.598983622973785e-05,
+      "loss": 1.2031,
+      "step": 5890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5988573746205477e-05,
+      "loss": 0.6562,
+      "step": 5891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5987311113833847e-05,
+      "loss": 1.1875,
+      "step": 5892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5986048332654335e-05,
+      "loss": 1.1562,
+      "step": 5893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5984785402698324e-05,
+      "loss": 0.7461,
+      "step": 5894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5983522323997208e-05,
+      "loss": 1.0156,
+      "step": 5895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5982259096582372e-05,
+      "loss": 0.9062,
+      "step": 5896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.598099572048522e-05,
+      "loss": 1.0469,
+      "step": 5897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5979732195737147e-05,
+      "loss": 0.7227,
+      "step": 5898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5978468522369557e-05,
+      "loss": 0.6719,
+      "step": 5899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.597720470041386e-05,
+      "loss": 0.9922,
+      "step": 5900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5975940729901464e-05,
+      "loss": 1.3359,
+      "step": 5901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5974676610863786e-05,
+      "loss": 1.0938,
+      "step": 5902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5973412343332237e-05,
+      "loss": 0.8867,
+      "step": 5903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.597214792733825e-05,
+      "loss": 0.9844,
+      "step": 5904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5970883362913245e-05,
+      "loss": 0.5312,
+      "step": 5905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.596961865008865e-05,
+      "loss": 0.9883,
+      "step": 5906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.59683537888959e-05,
+      "loss": 0.6406,
+      "step": 5907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.596708877936643e-05,
+      "loss": 1.1562,
+      "step": 5908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.596582362153168e-05,
+      "loss": 0.5703,
+      "step": 5909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.59645583154231e-05,
+      "loss": 1.2031,
+      "step": 5910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5963292861072134e-05,
+      "loss": 1.0234,
+      "step": 5911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.596202725851023e-05,
+      "loss": 1.4688,
+      "step": 5912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.596076150776885e-05,
+      "loss": 0.2715,
+      "step": 5913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5959495608879444e-05,
+      "loss": 1.0859,
+      "step": 5914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5958229561873483e-05,
+      "loss": 0.7461,
+      "step": 5915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5956963366782426e-05,
+      "loss": 0.6328,
+      "step": 5916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5955697023637747e-05,
+      "loss": 0.8906,
+      "step": 5917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.595443053247092e-05,
+      "loss": 0.3926,
+      "step": 5918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5953163893313425e-05,
+      "loss": 1.2031,
+      "step": 5919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5951897106196733e-05,
+      "loss": 0.9531,
+      "step": 5920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.595063017115234e-05,
+      "loss": 0.3809,
+      "step": 5921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5949363088211724e-05,
+      "loss": 1.1406,
+      "step": 5922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5948095857406384e-05,
+      "loss": 2.0938,
+      "step": 5923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5946828478767814e-05,
+      "loss": 1.8125,
+      "step": 5924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.594556095232751e-05,
+      "loss": 1.3594,
+      "step": 5925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.594429327811698e-05,
+      "loss": 1.4453,
+      "step": 5926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5943025456167728e-05,
+      "loss": 1.1875,
+      "step": 5927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5941757486511262e-05,
+      "loss": 0.5391,
+      "step": 5928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.59404893691791e-05,
+      "loss": 1.0156,
+      "step": 5929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.593922110420276e-05,
+      "loss": 1.3906,
+      "step": 5930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.593795269161376e-05,
+      "loss": 1.1328,
+      "step": 5931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5936684131443625e-05,
+      "loss": 0.7148,
+      "step": 5932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5935415423723886e-05,
+      "loss": 0.8984,
+      "step": 5933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5934146568486074e-05,
+      "loss": 1.0703,
+      "step": 5934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5932877565761727e-05,
+      "loss": 0.7305,
+      "step": 5935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.593160841558238e-05,
+      "loss": 1.0,
+      "step": 5936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5930339117979583e-05,
+      "loss": 0.6367,
+      "step": 5937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5929069672984876e-05,
+      "loss": 0.5312,
+      "step": 5938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5927800080629815e-05,
+      "loss": 1.2422,
+      "step": 5939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5926530340945953e-05,
+      "loss": 1.1094,
+      "step": 5940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.592526045396484e-05,
+      "loss": 1.0312,
+      "step": 5941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.592399041971805e-05,
+      "loss": 1.2031,
+      "step": 5942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5922720238237145e-05,
+      "loss": 0.3555,
+      "step": 5943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5921449909553686e-05,
+      "loss": 0.5039,
+      "step": 5944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5920179433699256e-05,
+      "loss": 1.0391,
+      "step": 5945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5918908810705428e-05,
+      "loss": 0.9531,
+      "step": 5946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5917638040603773e-05,
+      "loss": 0.8047,
+      "step": 5947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.591636712342589e-05,
+      "loss": 1.3438,
+      "step": 5948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5915096059203353e-05,
+      "loss": 0.8945,
+      "step": 5949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5913824847967762e-05,
+      "loss": 0.7852,
+      "step": 5950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5912553489750705e-05,
+      "loss": 0.5391,
+      "step": 5951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5911281984583786e-05,
+      "loss": 2.125,
+      "step": 5952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.59100103324986e-05,
+      "loss": 0.3418,
+      "step": 5953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.590873853352676e-05,
+      "loss": 0.9492,
+      "step": 5954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.590746658769987e-05,
+      "loss": 1.125,
+      "step": 5955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5906194495049545e-05,
+      "loss": 0.5703,
+      "step": 5956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.59049222556074e-05,
+      "loss": 0.4785,
+      "step": 5957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5903649869405055e-05,
+      "loss": 0.8242,
+      "step": 5958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5902377336474134e-05,
+      "loss": 0.4102,
+      "step": 5959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5901104656846265e-05,
+      "loss": 1.1484,
+      "step": 5960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.589983183055308e-05,
+      "loss": 1.0156,
+      "step": 5961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.589855885762621e-05,
+      "loss": 1.0625,
+      "step": 5962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5897285738097297e-05,
+      "loss": 1.5625,
+      "step": 5963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.589601247199798e-05,
+      "loss": 1.4531,
+      "step": 5964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5894739059359912e-05,
+      "loss": 0.9297,
+      "step": 5965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.589346550021473e-05,
+      "loss": 0.7383,
+      "step": 5966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5892191794594102e-05,
+      "loss": 0.3438,
+      "step": 5967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5890917942529668e-05,
+      "loss": 1.1641,
+      "step": 5968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5889643944053096e-05,
+      "loss": 0.4238,
+      "step": 5969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5888369799196054e-05,
+      "loss": 0.6797,
+      "step": 5970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5887095507990204e-05,
+      "loss": 1.6953,
+      "step": 5971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5885821070467216e-05,
+      "loss": 0.8477,
+      "step": 5972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.588454648665877e-05,
+      "loss": 1.3438,
+      "step": 5973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5883271756596542e-05,
+      "loss": 1.5,
+      "step": 5974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.588199688031221e-05,
+      "loss": 1.0156,
+      "step": 5975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5880721857837465e-05,
+      "loss": 0.7461,
+      "step": 5976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5879446689203995e-05,
+      "loss": 0.3945,
+      "step": 5977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587817137444349e-05,
+      "loss": 0.793,
+      "step": 5978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5876895913587653e-05,
+      "loss": 1.4844,
+      "step": 5979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587562030666818e-05,
+      "loss": 0.2432,
+      "step": 5980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587434455371677e-05,
+      "loss": 1.2266,
+      "step": 5981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5873068654765136e-05,
+      "loss": 1.3359,
+      "step": 5982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587179260984499e-05,
+      "loss": 0.6719,
+      "step": 5983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5870516418988046e-05,
+      "loss": 0.8008,
+      "step": 5984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5869240082226018e-05,
+      "loss": 0.7305,
+      "step": 5985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5867963599590632e-05,
+      "loss": 0.582,
+      "step": 5986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586668697111361e-05,
+      "loss": 0.9375,
+      "step": 5987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586541019682669e-05,
+      "loss": 1.2969,
+      "step": 5988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586413327676159e-05,
+      "loss": 0.9922,
+      "step": 5989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5862856210950066e-05,
+      "loss": 0.9883,
+      "step": 5990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5861578999423838e-05,
+      "loss": 1.4609,
+      "step": 5991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586030164221466e-05,
+      "loss": 1.0312,
+      "step": 5992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5859024139354278e-05,
+      "loss": 0.6094,
+      "step": 5993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5857746490874443e-05,
+      "loss": 1.1719,
+      "step": 5994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.585646869680691e-05,
+      "loss": 1.3125,
+      "step": 5995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5855190757183436e-05,
+      "loss": 0.5391,
+      "step": 5996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.585391267203578e-05,
+      "loss": 1.5156,
+      "step": 5997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5852634441395717e-05,
+      "loss": 1.0469,
+      "step": 5998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5851356065295007e-05,
+      "loss": 0.9258,
+      "step": 5999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5850077543765424e-05,
+      "loss": 0.9844,
+      "step": 6000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5848798876838743e-05,
+      "loss": 0.21,
+      "step": 6001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5847520064546748e-05,
+      "loss": 0.8711,
+      "step": 6002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5846241106921223e-05,
+      "loss": 0.2891,
+      "step": 6003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584496200399395e-05,
+      "loss": 0.8984,
+      "step": 6004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5843682755796724e-05,
+      "loss": 1.0781,
+      "step": 6005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5842403362361336e-05,
+      "loss": 0.5859,
+      "step": 6006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5841123823719585e-05,
+      "loss": 1.0781,
+      "step": 6007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5839844139903274e-05,
+      "loss": 0.6289,
+      "step": 6008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5838564310944205e-05,
+      "loss": 0.4434,
+      "step": 6009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5837284336874193e-05,
+      "loss": 1.1953,
+      "step": 6010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.583600421772504e-05,
+      "loss": 1.3516,
+      "step": 6011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5834723953528576e-05,
+      "loss": 1.7266,
+      "step": 6012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5833443544316607e-05,
+      "loss": 0.4688,
+      "step": 6013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5832162990120963e-05,
+      "loss": 1.0781,
+      "step": 6014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5830882290973468e-05,
+      "loss": 0.957,
+      "step": 6015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5829601446905955e-05,
+      "loss": 0.7227,
+      "step": 6016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5828320457950257e-05,
+      "loss": 0.7031,
+      "step": 6017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.582703932413821e-05,
+      "loss": 0.7852,
+      "step": 6018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5825758045501656e-05,
+      "loss": 0.9219,
+      "step": 6019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5824476622072443e-05,
+      "loss": 0.6172,
+      "step": 6020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5823195053882413e-05,
+      "loss": 0.8555,
+      "step": 6021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.582191334096342e-05,
+      "loss": 0.3555,
+      "step": 6022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5820631483347327e-05,
+      "loss": 0.9883,
+      "step": 6023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5819349481065983e-05,
+      "loss": 1.0234,
+      "step": 6024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5818067334151255e-05,
+      "loss": 1.2734,
+      "step": 6025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.581678504263501e-05,
+      "loss": 0.7109,
+      "step": 6026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5815502606549116e-05,
+      "loss": 1.3438,
+      "step": 6027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5814220025925446e-05,
+      "loss": 0.4609,
+      "step": 6028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.581293730079588e-05,
+      "loss": 0.7266,
+      "step": 6029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5811654431192297e-05,
+      "loss": 0.8281,
+      "step": 6030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.581037141714658e-05,
+      "loss": 1.9219,
+      "step": 6031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5809088258690623e-05,
+      "loss": 1.1953,
+      "step": 6032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.580780495585631e-05,
+      "loss": 1.3359,
+      "step": 6033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.580652150867554e-05,
+      "loss": 1.2266,
+      "step": 6034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5805237917180208e-05,
+      "loss": 0.4414,
+      "step": 6035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5803954181402223e-05,
+      "loss": 1.375,
+      "step": 6036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.580267030137348e-05,
+      "loss": 0.9844,
+      "step": 6037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.58013862771259e-05,
+      "loss": 0.9453,
+      "step": 6038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.580010210869139e-05,
+      "loss": 0.1895,
+      "step": 6039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5798817796101868e-05,
+      "loss": 0.3496,
+      "step": 6040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5797533339389252e-05,
+      "loss": 1.3438,
+      "step": 6041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.579624873858547e-05,
+      "loss": 1.1484,
+      "step": 6042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5794963993722444e-05,
+      "loss": 1.0,
+      "step": 6043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5793679104832108e-05,
+      "loss": 1.1719,
+      "step": 6044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5792394071946398e-05,
+      "loss": 0.8789,
+      "step": 6045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5791108895097246e-05,
+      "loss": 0.832,
+      "step": 6046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5789823574316604e-05,
+      "loss": 1.2109,
+      "step": 6047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5788538109636404e-05,
+      "loss": 1.0703,
+      "step": 6048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5787252501088607e-05,
+      "loss": 0.7539,
+      "step": 6049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5785966748705156e-05,
+      "loss": 0.9258,
+      "step": 6050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5784680852518014e-05,
+      "loss": 1.2109,
+      "step": 6051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5783394812559136e-05,
+      "loss": 0.8047,
+      "step": 6052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.578210862886049e-05,
+      "loss": 0.4941,
+      "step": 6053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5780822301454032e-05,
+      "loss": 1.1875,
+      "step": 6054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5779535830371744e-05,
+      "loss": 0.4883,
+      "step": 6055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5778249215645594e-05,
+      "loss": 1.8984,
+      "step": 6056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5776962457307563e-05,
+      "loss": 1.1094,
+      "step": 6057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5775675555389627e-05,
+      "loss": 1.0859,
+      "step": 6058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5774388509923775e-05,
+      "loss": 1.1562,
+      "step": 6059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.577310132094199e-05,
+      "loss": 0.957,
+      "step": 6060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.577181398847627e-05,
+      "loss": 0.2969,
+      "step": 6061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5770526512558605e-05,
+      "loss": 0.9062,
+      "step": 6062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5769238893220995e-05,
+      "loss": 1.0078,
+      "step": 6063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5767951130495446e-05,
+      "loss": 0.7695,
+      "step": 6064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5766663224413957e-05,
+      "loss": 1.2812,
+      "step": 6065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5765375175008545e-05,
+      "loss": 1.7031,
+      "step": 6066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5764086982311216e-05,
+      "loss": 0.4902,
+      "step": 6067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5762798646353995e-05,
+      "loss": 1.2188,
+      "step": 6068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5761510167168893e-05,
+      "loss": 0.9062,
+      "step": 6069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.576022154478794e-05,
+      "loss": 2.2031,
+      "step": 6070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5758932779243156e-05,
+      "loss": 0.6367,
+      "step": 6071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5757643870566583e-05,
+      "loss": 0.8438,
+      "step": 6072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5756354818790244e-05,
+      "loss": 2.0312,
+      "step": 6073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5755065623946185e-05,
+      "loss": 1.5781,
+      "step": 6074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.575377628606644e-05,
+      "loss": 0.8438,
+      "step": 6075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.575248680518307e-05,
+      "loss": 0.6797,
+      "step": 6076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.57511971813281e-05,
+      "loss": 0.5352,
+      "step": 6077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5749907414533603e-05,
+      "loss": 1.0078,
+      "step": 6078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5748617504831623e-05,
+      "loss": 1.4844,
+      "step": 6079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5747327452254223e-05,
+      "loss": 0.9375,
+      "step": 6080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5746037256833464e-05,
+      "loss": 1.2266,
+      "step": 6081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5744746918601415e-05,
+      "loss": 0.8203,
+      "step": 6082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5743456437590142e-05,
+      "loss": 1.2969,
+      "step": 6083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5742165813831723e-05,
+      "loss": 1.5625,
+      "step": 6084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5740875047358234e-05,
+      "loss": 0.8203,
+      "step": 6085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5739584138201754e-05,
+      "loss": 0.9766,
+      "step": 6086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5738293086394364e-05,
+      "loss": 0.6641,
+      "step": 6087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5737001891968158e-05,
+      "loss": 0.9023,
+      "step": 6088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5735710554955224e-05,
+      "loss": 1.4688,
+      "step": 6089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.573441907538766e-05,
+      "loss": 0.6992,
+      "step": 6090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5733127453297556e-05,
+      "loss": 1.0703,
+      "step": 6091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5731835688717024e-05,
+      "loss": 1.0625,
+      "step": 6092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5730543781678164e-05,
+      "loss": 1.3203,
+      "step": 6093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5729251732213082e-05,
+      "loss": 1.1562,
+      "step": 6094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5727959540353895e-05,
+      "loss": 0.7148,
+      "step": 6095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5726667206132718e-05,
+      "loss": 0.8359,
+      "step": 6096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.572537472958167e-05,
+      "loss": 0.9375,
+      "step": 6097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5724082110732876e-05,
+      "loss": 0.875,
+      "step": 6098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.572278934961846e-05,
+      "loss": 0.8047,
+      "step": 6099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5721496446270554e-05,
+      "loss": 0.6914,
+      "step": 6100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.572020340072129e-05,
+      "loss": 1.4297,
+      "step": 6101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5718910213002806e-05,
+      "loss": 1.0312,
+      "step": 6102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5717616883147244e-05,
+      "loss": 1.875,
+      "step": 6103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.571632341118675e-05,
+      "loss": 1.3828,
+      "step": 6104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.571502979715346e-05,
+      "loss": 1.9062,
+      "step": 6105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5713736041079543e-05,
+      "loss": 0.8516,
+      "step": 6106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5712442142997136e-05,
+      "loss": 0.4395,
+      "step": 6107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5711148102938414e-05,
+      "loss": 1.4844,
+      "step": 6108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.570985392093553e-05,
+      "loss": 0.3984,
+      "step": 6109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5708559597020648e-05,
+      "loss": 1.3516,
+      "step": 6110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.570726513122594e-05,
+      "loss": 1.0391,
+      "step": 6111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.570597052358358e-05,
+      "loss": 0.7812,
+      "step": 6112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5704675774125742e-05,
+      "loss": 0.3262,
+      "step": 6113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.57033808828846e-05,
+      "loss": 0.2754,
+      "step": 6114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.570208584989235e-05,
+      "loss": 0.6914,
+      "step": 6115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.570079067518117e-05,
+      "loss": 0.2754,
+      "step": 6116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5699495358783252e-05,
+      "loss": 1.3906,
+      "step": 6117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5698199900730786e-05,
+      "loss": 0.1875,
+      "step": 6118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5696904301055977e-05,
+      "loss": 0.9883,
+      "step": 6119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5695608559791015e-05,
+      "loss": 0.4961,
+      "step": 6120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5694312676968114e-05,
+      "loss": 1.375,
+      "step": 6121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5693016652619473e-05,
+      "loss": 2.3125,
+      "step": 6122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5691720486777315e-05,
+      "loss": 0.5,
+      "step": 6123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.569042417947384e-05,
+      "loss": 0.8359,
+      "step": 6124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.568912773074128e-05,
+      "loss": 1.4219,
+      "step": 6125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5687831140611847e-05,
+      "loss": 0.8672,
+      "step": 6126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5686534409117773e-05,
+      "loss": 1.2578,
+      "step": 6127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5685237536291285e-05,
+      "loss": 0.543,
+      "step": 6128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.568394052216461e-05,
+      "loss": 0.498,
+      "step": 6129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5682643366769988e-05,
+      "loss": 1.1094,
+      "step": 6130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5681346070139665e-05,
+      "loss": 0.9336,
+      "step": 6131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.568004863230587e-05,
+      "loss": 1.0312,
+      "step": 6132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5678751053300862e-05,
+      "loss": 1.2969,
+      "step": 6133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.567745333315688e-05,
+      "loss": 1.0859,
+      "step": 6134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5676155471906186e-05,
+      "loss": 0.9258,
+      "step": 6135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5674857469581036e-05,
+      "loss": 1.6641,
+      "step": 6136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5673559326213685e-05,
+      "loss": 1.2109,
+      "step": 6137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.56722610418364e-05,
+      "loss": 0.7148,
+      "step": 6138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.567096261648145e-05,
+      "loss": 1.4141,
+      "step": 6139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5669664050181103e-05,
+      "loss": 0.8008,
+      "step": 6140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5668365342967633e-05,
+      "loss": 1.1484,
+      "step": 6141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5667066494873326e-05,
+      "loss": 0.5742,
+      "step": 6142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.566576750593045e-05,
+      "loss": 1.3125,
+      "step": 6143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5664468376171297e-05,
+      "loss": 1.1484,
+      "step": 6144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.566316910562816e-05,
+      "loss": 0.4727,
+      "step": 6145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5661869694333322e-05,
+      "loss": 0.8867,
+      "step": 6146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5660570142319084e-05,
+      "loss": 0.9727,
+      "step": 6147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.565927044961775e-05,
+      "loss": 0.9414,
+      "step": 6148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5657970616261605e-05,
+      "loss": 1.0469,
+      "step": 6149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5656670642282973e-05,
+      "loss": 1.1562,
+      "step": 6150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5655370527714154e-05,
+      "loss": 1.8984,
+      "step": 6151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5654070272587462e-05,
+      "loss": 1.6484,
+      "step": 6152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5652769876935215e-05,
+      "loss": 0.8047,
+      "step": 6153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5651469340789733e-05,
+      "loss": 0.5898,
+      "step": 6154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.565016866418334e-05,
+      "loss": 0.9414,
+      "step": 6155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.564886784714836e-05,
+      "loss": 1.5938,
+      "step": 6156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5647566889717126e-05,
+      "loss": 1.5781,
+      "step": 6157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.564626579192197e-05,
+      "loss": 0.6641,
+      "step": 6158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.564496455379523e-05,
+      "loss": 1.3047,
+      "step": 6159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5643663175369247e-05,
+      "loss": 0.9531,
+      "step": 6160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5642361656676364e-05,
+      "loss": 0.3125,
+      "step": 6161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5641059997748934e-05,
+      "loss": 1.1094,
+      "step": 6162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.56397581986193e-05,
+      "loss": 0.25,
+      "step": 6163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.563845625931982e-05,
+      "loss": 0.9883,
+      "step": 6164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5637154179882856e-05,
+      "loss": 1.3125,
+      "step": 6165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5635851960340768e-05,
+      "loss": 0.8203,
+      "step": 6166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.563454960072592e-05,
+      "loss": 1.125,
+      "step": 6167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5633247101070675e-05,
+      "loss": 1.7266,
+      "step": 6168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.563194446140742e-05,
+      "loss": 1.2969,
+      "step": 6169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5630641681768516e-05,
+      "loss": 1.9609,
+      "step": 6170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.562933876218635e-05,
+      "loss": 1.3125,
+      "step": 6171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.56280357026933e-05,
+      "loss": 0.668,
+      "step": 6172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5626732503321754e-05,
+      "loss": 1.4219,
+      "step": 6173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5625429164104106e-05,
+      "loss": 1.1406,
+      "step": 6174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5624125685072743e-05,
+      "loss": 1.375,
+      "step": 6175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5622822066260063e-05,
+      "loss": 1.2422,
+      "step": 6176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5621518307698466e-05,
+      "loss": 1.4062,
+      "step": 6177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5620214409420356e-05,
+      "loss": 1.5625,
+      "step": 6178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.561891037145814e-05,
+      "loss": 0.4844,
+      "step": 6179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.561760619384423e-05,
+      "loss": 0.7461,
+      "step": 6180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5616301876611037e-05,
+      "loss": 0.9688,
+      "step": 6181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.561499741979098e-05,
+      "loss": 1.4375,
+      "step": 6182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.561369282341648e-05,
+      "loss": 0.3496,
+      "step": 6183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5612388087519957e-05,
+      "loss": 1.4531,
+      "step": 6184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5611083212133846e-05,
+      "loss": 0.8945,
+      "step": 6185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5609778197290573e-05,
+      "loss": 1.4297,
+      "step": 6186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5608473043022573e-05,
+      "loss": 0.7109,
+      "step": 6187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5607167749362288e-05,
+      "loss": 0.4941,
+      "step": 6188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5605862316342156e-05,
+      "loss": 1.1016,
+      "step": 6189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5604556743994624e-05,
+      "loss": 0.9023,
+      "step": 6190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5603251032352135e-05,
+      "loss": 1.4219,
+      "step": 6191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.560194518144715e-05,
+      "loss": 1.0156,
+      "step": 6192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5600639191312116e-05,
+      "loss": 1.5,
+      "step": 6193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5599333061979503e-05,
+      "loss": 0.8203,
+      "step": 6194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.559802679348176e-05,
+      "loss": 1.0703,
+      "step": 6195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5596720385851362e-05,
+      "loss": 1.3906,
+      "step": 6196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.559541383912077e-05,
+      "loss": 0.875,
+      "step": 6197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.559410715332247e-05,
+      "loss": 1.0,
+      "step": 6198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5592800328488926e-05,
+      "loss": 0.9844,
+      "step": 6199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5591493364652623e-05,
+      "loss": 0.9258,
+      "step": 6200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5590186261846043e-05,
+      "loss": 1.5312,
+      "step": 6201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5588879020101676e-05,
+      "loss": 0.7812,
+      "step": 6202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5587571639452003e-05,
+      "loss": 1.3359,
+      "step": 6203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5586264119929526e-05,
+      "loss": 1.25,
+      "step": 6204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.558495646156674e-05,
+      "loss": 0.5039,
+      "step": 6205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.558364866439614e-05,
+      "loss": 1.375,
+      "step": 6206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.558234072845024e-05,
+      "loss": 1.6016,
+      "step": 6207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5581032653761536e-05,
+      "loss": 0.8164,
+      "step": 6208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5579724440362548e-05,
+      "loss": 0.793,
+      "step": 6209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5578416088285787e-05,
+      "loss": 0.4219,
+      "step": 6210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5577107597563766e-05,
+      "loss": 0.8125,
+      "step": 6211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5575798968229015e-05,
+      "loss": 0.6953,
+      "step": 6212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5574490200314053e-05,
+      "loss": 1.7969,
+      "step": 6213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.557318129385141e-05,
+      "loss": 0.8867,
+      "step": 6214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.557187224887361e-05,
+      "loss": 0.6953,
+      "step": 6215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.55705630654132e-05,
+      "loss": 0.6406,
+      "step": 6216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.556925374350271e-05,
+      "loss": 0.8477,
+      "step": 6217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5567944283174682e-05,
+      "loss": 0.6055,
+      "step": 6218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5566634684461667e-05,
+      "loss": 1.6797,
+      "step": 6219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5565324947396208e-05,
+      "loss": 0.8047,
+      "step": 6220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5564015072010855e-05,
+      "loss": 1.5391,
+      "step": 6221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.556270505833817e-05,
+      "loss": 0.9531,
+      "step": 6222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.556139490641071e-05,
+      "loss": 0.7969,
+      "step": 6223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5560084616261036e-05,
+      "loss": 1.4688,
+      "step": 6224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5558774187921716e-05,
+      "loss": 0.5664,
+      "step": 6225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5557463621425315e-05,
+      "loss": 0.7539,
+      "step": 6226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5556152916804407e-05,
+      "loss": 1.3281,
+      "step": 6227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5554842074091575e-05,
+      "loss": 0.7266,
+      "step": 6228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5553531093319386e-05,
+      "loss": 0.9297,
+      "step": 6229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5552219974520434e-05,
+      "loss": 0.7773,
+      "step": 6230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.55509087177273e-05,
+      "loss": 0.9102,
+      "step": 6231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.554959732297257e-05,
+      "loss": 0.9727,
+      "step": 6232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.554828579028885e-05,
+      "loss": 0.9297,
+      "step": 6233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.554697411970872e-05,
+      "loss": 0.8398,
+      "step": 6234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5545662311264796e-05,
+      "loss": 1.0391,
+      "step": 6235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5544350364989675e-05,
+      "loss": 1.4141,
+      "step": 6236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5543038280915957e-05,
+      "loss": 1.1016,
+      "step": 6237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5541726059076264e-05,
+      "loss": 1.0781,
+      "step": 6238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5540413699503206e-05,
+      "loss": 1.4922,
+      "step": 6239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5539101202229395e-05,
+      "loss": 0.543,
+      "step": 6240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5537788567287454e-05,
+      "loss": 0.8555,
+      "step": 6241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5536475794710012e-05,
+      "loss": 1.0547,
+      "step": 6242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5535162884529692e-05,
+      "loss": 1.0312,
+      "step": 6243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5533849836779123e-05,
+      "loss": 1.0547,
+      "step": 6244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5532536651490945e-05,
+      "loss": 1.1406,
+      "step": 6245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5531223328697794e-05,
+      "loss": 1.3672,
+      "step": 6246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5529909868432306e-05,
+      "loss": 0.6953,
+      "step": 6247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5528596270727134e-05,
+      "loss": 0.9922,
+      "step": 6248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.552728253561492e-05,
+      "loss": 0.9375,
+      "step": 6249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5525968663128317e-05,
+      "loss": 0.9375,
+      "step": 6250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.552465465329998e-05,
+      "loss": 0.8047,
+      "step": 6251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5523340506162566e-05,
+      "loss": 1.9609,
+      "step": 6252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.552202622174874e-05,
+      "loss": 1.25,
+      "step": 6253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5520711800091163e-05,
+      "loss": 0.9766,
+      "step": 6254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5519397241222506e-05,
+      "loss": 1.5156,
+      "step": 6255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.551808254517544e-05,
+      "loss": 0.5703,
+      "step": 6256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5516767711982638e-05,
+      "loss": 0.6055,
+      "step": 6257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5515452741676783e-05,
+      "loss": 0.9375,
+      "step": 6258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.551413763429056e-05,
+      "loss": 0.5977,
+      "step": 6259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5512822389856643e-05,
+      "loss": 0.582,
+      "step": 6260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.551150700840773e-05,
+      "loss": 1.1562,
+      "step": 6261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5510191489976513e-05,
+      "loss": 0.9805,
+      "step": 6262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5508875834595683e-05,
+      "loss": 0.1982,
+      "step": 6263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5507560042297943e-05,
+      "loss": 0.7383,
+      "step": 6264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5506244113115996e-05,
+      "loss": 0.5,
+      "step": 6265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.550492804708254e-05,
+      "loss": 0.5742,
+      "step": 6266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.55036118442303e-05,
+      "loss": 1.2969,
+      "step": 6267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5502295504591972e-05,
+      "loss": 0.7383,
+      "step": 6268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5500979028200284e-05,
+      "loss": 1.1719,
+      "step": 6269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.549966241508795e-05,
+      "loss": 1.6328,
+      "step": 6270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5498345665287697e-05,
+      "loss": 0.9688,
+      "step": 6271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.549702877883224e-05,
+      "loss": 1.0625,
+      "step": 6272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5495711755754324e-05,
+      "loss": 1.0078,
+      "step": 6273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5494394596086672e-05,
+      "loss": 0.9297,
+      "step": 6274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5493077299862026e-05,
+      "loss": 0.7891,
+      "step": 6275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5491759867113124e-05,
+      "loss": 1.1406,
+      "step": 6276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5490442297872707e-05,
+      "loss": 0.8789,
+      "step": 6277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5489124592173523e-05,
+      "loss": 0.5039,
+      "step": 6278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5487806750048326e-05,
+      "loss": 1.0234,
+      "step": 6279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5486488771529866e-05,
+      "loss": 0.6016,
+      "step": 6280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5485170656650898e-05,
+      "loss": 0.8594,
+      "step": 6281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5483852405444183e-05,
+      "loss": 1.6719,
+      "step": 6282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.548253401794249e-05,
+      "loss": 1.1797,
+      "step": 6283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5481215494178582e-05,
+      "loss": 1.3672,
+      "step": 6284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.547989683418523e-05,
+      "loss": 1.2734,
+      "step": 6285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5478578037995205e-05,
+      "loss": 1.2656,
+      "step": 6286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5477259105641285e-05,
+      "loss": 1.6953,
+      "step": 6287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5475940037156254e-05,
+      "loss": 0.373,
+      "step": 6288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5474620832572896e-05,
+      "loss": 0.543,
+      "step": 6289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5473301491923992e-05,
+      "loss": 0.9844,
+      "step": 6290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5471982015242342e-05,
+      "loss": 1.4141,
+      "step": 6291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.547066240256073e-05,
+      "loss": 0.9961,
+      "step": 6292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5469342653911966e-05,
+      "loss": 1.0625,
+      "step": 6293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5468022769328837e-05,
+      "loss": 0.8984,
+      "step": 6294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5466702748844157e-05,
+      "loss": 1.1562,
+      "step": 6295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.546538259249073e-05,
+      "loss": 0.7734,
+      "step": 6296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5464062300301365e-05,
+      "loss": 0.9336,
+      "step": 6297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.546274187230888e-05,
+      "loss": 0.8008,
+      "step": 6298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5461421308546098e-05,
+      "loss": 0.5078,
+      "step": 6299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5460100609045828e-05,
+      "loss": 0.9062,
+      "step": 6300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5458779773840896e-05,
+      "loss": 1.7969,
+      "step": 6301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.545745880296414e-05,
+      "loss": 1.25,
+      "step": 6302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5456137696448386e-05,
+      "loss": 1.1172,
+      "step": 6303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5454816454326467e-05,
+      "loss": 1.3047,
+      "step": 6304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5453495076631223e-05,
+      "loss": 0.9102,
+      "step": 6305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5452173563395488e-05,
+      "loss": 1.0234,
+      "step": 6306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.545085191465212e-05,
+      "loss": 1.3828,
+      "step": 6307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5449530130433957e-05,
+      "loss": 0.7031,
+      "step": 6308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5448208210773853e-05,
+      "loss": 0.2891,
+      "step": 6309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5446886155704667e-05,
+      "loss": 1.2188,
+      "step": 6310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.544556396525925e-05,
+      "loss": 0.9961,
+      "step": 6311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.544424163947047e-05,
+      "loss": 0.4121,
+      "step": 6312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5442919178371186e-05,
+      "loss": 0.416,
+      "step": 6313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5441596581994272e-05,
+      "loss": 1.0,
+      "step": 6314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5440273850372596e-05,
+      "loss": 0.5547,
+      "step": 6315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5438950983539034e-05,
+      "loss": 0.4277,
+      "step": 6316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5437627981526465e-05,
+      "loss": 0.2148,
+      "step": 6317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.543630484436777e-05,
+      "loss": 1.5938,
+      "step": 6318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5434981572095833e-05,
+      "loss": 0.2695,
+      "step": 6319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5433658164743547e-05,
+      "loss": 2.0469,
+      "step": 6320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5432334622343798e-05,
+      "loss": 1.2344,
+      "step": 6321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.543101094492948e-05,
+      "loss": 1.9062,
+      "step": 6322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.54296871325335e-05,
+      "loss": 1.4219,
+      "step": 6323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5428363185188753e-05,
+      "loss": 1.5391,
+      "step": 6324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.542703910292815e-05,
+      "loss": 0.4609,
+      "step": 6325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5425714885784588e-05,
+      "loss": 0.498,
+      "step": 6326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5424390533790995e-05,
+      "loss": 1.0547,
+      "step": 6327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5423066046980274e-05,
+      "loss": 1.0625,
+      "step": 6328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5421741425385347e-05,
+      "loss": 1.3203,
+      "step": 6329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5420416669039137e-05,
+      "loss": 1.8906,
+      "step": 6330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.541909177797457e-05,
+      "loss": 1.0781,
+      "step": 6331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5417766752224574e-05,
+      "loss": 0.7109,
+      "step": 6332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.541644159182208e-05,
+      "loss": 1.4453,
+      "step": 6333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5415116296800024e-05,
+      "loss": 0.7266,
+      "step": 6334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5413790867191346e-05,
+      "loss": 1.2031,
+      "step": 6335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5412465303028984e-05,
+      "loss": 0.4453,
+      "step": 6336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5411139604345887e-05,
+      "loss": 0.918,
+      "step": 6337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5409813771175002e-05,
+      "loss": 1.3203,
+      "step": 6338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5408487803549282e-05,
+      "loss": 1.1875,
+      "step": 6339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.540716170150169e-05,
+      "loss": 0.6523,
+      "step": 6340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5405835465065167e-05,
+      "loss": 1.6719,
+      "step": 6341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.540450909427269e-05,
+      "loss": 0.8711,
+      "step": 6342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5403182589157218e-05,
+      "loss": 1.5234,
+      "step": 6343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5401855949751724e-05,
+      "loss": 1.4531,
+      "step": 6344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.540052917608918e-05,
+      "loss": 1.0859,
+      "step": 6345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5399202268202556e-05,
+      "loss": 0.9844,
+      "step": 6346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5397875226124832e-05,
+      "loss": 0.6133,
+      "step": 6347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5396548049888996e-05,
+      "loss": 0.8008,
+      "step": 6348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5395220739528026e-05,
+      "loss": 0.6992,
+      "step": 6349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5393893295074914e-05,
+      "loss": 0.8672,
+      "step": 6350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5392565716562656e-05,
+      "loss": 1.6484,
+      "step": 6351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5391238004024243e-05,
+      "loss": 1.8359,
+      "step": 6352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5389910157492675e-05,
+      "loss": 0.5898,
+      "step": 6353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5388582177000952e-05,
+      "loss": 0.582,
+      "step": 6354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5387254062582086e-05,
+      "loss": 0.7539,
+      "step": 6355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5385925814269075e-05,
+      "loss": 1.4219,
+      "step": 6356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5384597432094942e-05,
+      "loss": 0.7891,
+      "step": 6357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5383268916092697e-05,
+      "loss": 0.6523,
+      "step": 6358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5381940266295356e-05,
+      "loss": 0.9609,
+      "step": 6359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.538061148273595e-05,
+      "loss": 1.0625,
+      "step": 6360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5379282565447494e-05,
+      "loss": 1.4766,
+      "step": 6361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5377953514463023e-05,
+      "loss": 0.4453,
+      "step": 6362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.537662432981557e-05,
+      "loss": 1.9375,
+      "step": 6363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.537529501153817e-05,
+      "loss": 0.8555,
+      "step": 6364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5373965559663853e-05,
+      "loss": 0.8555,
+      "step": 6365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5372635974225673e-05,
+      "loss": 0.8672,
+      "step": 6366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.537130625525667e-05,
+      "loss": 1.1719,
+      "step": 6367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5369976402789894e-05,
+      "loss": 1.0312,
+      "step": 6368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.536864641685839e-05,
+      "loss": 0.832,
+      "step": 6369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.536731629749523e-05,
+      "loss": 1.125,
+      "step": 6370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5365986044733453e-05,
+      "loss": 0.9414,
+      "step": 6371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5364655658606134e-05,
+      "loss": 1.7422,
+      "step": 6372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.536332513914633e-05,
+      "loss": 1.3203,
+      "step": 6373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5361994486387118e-05,
+      "loss": 0.9922,
+      "step": 6374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5360663700361565e-05,
+      "loss": 1.2969,
+      "step": 6375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5359332781102743e-05,
+      "loss": 1.0391,
+      "step": 6376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.535800172864374e-05,
+      "loss": 0.6172,
+      "step": 6377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.535667054301763e-05,
+      "loss": 1.4219,
+      "step": 6378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5355339224257497e-05,
+      "loss": 1.3516,
+      "step": 6379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5354007772396434e-05,
+      "loss": 1.1016,
+      "step": 6380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.535267618746753e-05,
+      "loss": 1.1094,
+      "step": 6381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5351344469503885e-05,
+      "loss": 0.5586,
+      "step": 6382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5350012618538593e-05,
+      "loss": 1.5859,
+      "step": 6383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5348680634604758e-05,
+      "loss": 1.0625,
+      "step": 6384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.534734851773548e-05,
+      "loss": 0.9609,
+      "step": 6385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5346016267963873e-05,
+      "loss": 0.6016,
+      "step": 6386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5344683885323047e-05,
+      "loss": 0.7812,
+      "step": 6387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5343351369846114e-05,
+      "loss": 0.707,
+      "step": 6388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.534201872156619e-05,
+      "loss": 0.9883,
+      "step": 6389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.534068594051641e-05,
+      "loss": 0.7891,
+      "step": 6390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.533935302672988e-05,
+      "loss": 0.8164,
+      "step": 6391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5338019980239744e-05,
+      "loss": 0.7656,
+      "step": 6392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5336686801079123e-05,
+      "loss": 1.1328,
+      "step": 6393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.533535348928116e-05,
+      "loss": 0.5273,
+      "step": 6394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5334020044878984e-05,
+      "loss": 1.0938,
+      "step": 6395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.533268646790574e-05,
+      "loss": 0.6875,
+      "step": 6396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.533135275839458e-05,
+      "loss": 1.0469,
+      "step": 6397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.533001891637864e-05,
+      "loss": 0.7188,
+      "step": 6398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.532868494189108e-05,
+      "loss": 1.0234,
+      "step": 6399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5327350834965047e-05,
+      "loss": 1.0078,
+      "step": 6400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5326016595633706e-05,
+      "loss": 1.2109,
+      "step": 6401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5324682223930212e-05,
+      "loss": 0.4844,
+      "step": 6402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5323347719887733e-05,
+      "loss": 0.6992,
+      "step": 6403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5322013083539436e-05,
+      "loss": 0.8594,
+      "step": 6404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5320678314918488e-05,
+      "loss": 1.4609,
+      "step": 6405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5319343414058072e-05,
+      "loss": 0.3887,
+      "step": 6406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5318008380991357e-05,
+      "loss": 0.4707,
+      "step": 6407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5316673215751524e-05,
+      "loss": 1.1641,
+      "step": 6408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5315337918371763e-05,
+      "loss": 0.8672,
+      "step": 6409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5314002488885257e-05,
+      "loss": 1.0547,
+      "step": 6410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5312666927325203e-05,
+      "loss": 0.2734,
+      "step": 6411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.531133123372478e-05,
+      "loss": 0.4961,
+      "step": 6412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.53099954081172e-05,
+      "loss": 0.2344,
+      "step": 6413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5308659450535658e-05,
+      "loss": 0.9922,
+      "step": 6414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5307323361013358e-05,
+      "loss": 0.957,
+      "step": 6415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5305987139583508e-05,
+      "loss": 1.8594,
+      "step": 6416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5304650786279315e-05,
+      "loss": 1.0859,
+      "step": 6417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5303314301133997e-05,
+      "loss": 1.8281,
+      "step": 6418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5301977684180764e-05,
+      "loss": 1.5156,
+      "step": 6419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5300640935452843e-05,
+      "loss": 1.1094,
+      "step": 6420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5299304054983455e-05,
+      "loss": 1.0547,
+      "step": 6421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5297967042805828e-05,
+      "loss": 1.1719,
+      "step": 6422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.529662989895319e-05,
+      "loss": 0.7695,
+      "step": 6423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5295292623458774e-05,
+      "loss": 1.3047,
+      "step": 6424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.529395521635582e-05,
+      "loss": 0.6875,
+      "step": 6425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5292617677677562e-05,
+      "loss": 0.4043,
+      "step": 6426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5291280007457247e-05,
+      "loss": 0.7266,
+      "step": 6427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.528994220572812e-05,
+      "loss": 0.3965,
+      "step": 6428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5288604272523427e-05,
+      "loss": 1.2578,
+      "step": 6429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5287266207876433e-05,
+      "loss": 0.5469,
+      "step": 6430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.528592801182038e-05,
+      "loss": 1.1328,
+      "step": 6431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5284589684388536e-05,
+      "loss": 0.8477,
+      "step": 6432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.528325122561416e-05,
+      "loss": 0.5898,
+      "step": 6433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.528191263553052e-05,
+      "loss": 1.3125,
+      "step": 6434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5280573914170883e-05,
+      "loss": 1.2969,
+      "step": 6435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.527923506156852e-05,
+      "loss": 1.2031,
+      "step": 6436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.527789607775671e-05,
+      "loss": 1.5547,
+      "step": 6437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5276556962768732e-05,
+      "loss": 1.6562,
+      "step": 6438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.527521771663787e-05,
+      "loss": 0.5234,
+      "step": 6439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.52738783393974e-05,
+      "loss": 1.5234,
+      "step": 6440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5272538831080624e-05,
+      "loss": 1.3828,
+      "step": 6441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.527119919172082e-05,
+      "loss": 1.0391,
+      "step": 6442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5269859421351294e-05,
+      "loss": 1.2812,
+      "step": 6443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5268519520005344e-05,
+      "loss": 0.8281,
+      "step": 6444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5267179487716263e-05,
+      "loss": 1.0156,
+      "step": 6445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5265839324517366e-05,
+      "loss": 0.625,
+      "step": 6446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5264499030441954e-05,
+      "loss": 0.9062,
+      "step": 6447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5263158605523344e-05,
+      "loss": 0.7578,
+      "step": 6448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5261818049794847e-05,
+      "loss": 0.707,
+      "step": 6449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.526047736328978e-05,
+      "loss": 0.7617,
+      "step": 6450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.525913654604147e-05,
+      "loss": 1.4375,
+      "step": 6451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5257795598083236e-05,
+      "loss": 1.1562,
+      "step": 6452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5256454519448411e-05,
+      "loss": 1.1406,
+      "step": 6453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5255113310170317e-05,
+      "loss": 1.0547,
+      "step": 6454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5253771970282301e-05,
+      "loss": 0.5352,
+      "step": 6455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5252430499817685e-05,
+      "loss": 0.9062,
+      "step": 6456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5251088898809823e-05,
+      "loss": 1.1484,
+      "step": 6457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5249747167292052e-05,
+      "loss": 1.0781,
+      "step": 6458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5248405305297725e-05,
+      "loss": 1.4688,
+      "step": 6459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5247063312860188e-05,
+      "loss": 0.3984,
+      "step": 6460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5245721190012791e-05,
+      "loss": 0.3281,
+      "step": 6461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5244378936788901e-05,
+      "loss": 0.9766,
+      "step": 6462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5243036553221871e-05,
+      "loss": 0.2139,
+      "step": 6463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5241694039345065e-05,
+      "loss": 1.1562,
+      "step": 6464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5240351395191853e-05,
+      "loss": 0.4414,
+      "step": 6465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5239008620795602e-05,
+      "loss": 0.3203,
+      "step": 6466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5237665716189685e-05,
+      "loss": 0.9141,
+      "step": 6467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5236322681407478e-05,
+      "loss": 0.7188,
+      "step": 6468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5234979516482366e-05,
+      "loss": 0.4414,
+      "step": 6469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5233636221447722e-05,
+      "loss": 0.7109,
+      "step": 6470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5232292796336942e-05,
+      "loss": 1.2031,
+      "step": 6471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.523094924118341e-05,
+      "loss": 0.6406,
+      "step": 6472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5229605556020519e-05,
+      "loss": 1.2031,
+      "step": 6473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.522826174088167e-05,
+      "loss": 1.25,
+      "step": 6474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.522691779580025e-05,
+      "loss": 1.5391,
+      "step": 6475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5225573720809674e-05,
+      "loss": 1.2812,
+      "step": 6476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5224229515943343e-05,
+      "loss": 0.7539,
+      "step": 6477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5222885181234661e-05,
+      "loss": 0.7422,
+      "step": 6478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5221540716717046e-05,
+      "loss": 0.498,
+      "step": 6479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.522019612242391e-05,
+      "loss": 1.3203,
+      "step": 6480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5218851398388674e-05,
+      "loss": 1.1328,
+      "step": 6481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5217506544644756e-05,
+      "loss": 1.9062,
+      "step": 6482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5216161561225584e-05,
+      "loss": 1.3516,
+      "step": 6483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5214816448164584e-05,
+      "loss": 2.2188,
+      "step": 6484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5213471205495187e-05,
+      "loss": 0.6094,
+      "step": 6485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5212125833250833e-05,
+      "loss": 1.0938,
+      "step": 6486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5210780331464952e-05,
+      "loss": 1.2031,
+      "step": 6487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5209434700170989e-05,
+      "loss": 1.0703,
+      "step": 6488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5208088939402386e-05,
+      "loss": 0.8516,
+      "step": 6489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5206743049192594e-05,
+      "loss": 0.5195,
+      "step": 6490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.520539702957506e-05,
+      "loss": 0.707,
+      "step": 6491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5204050880583237e-05,
+      "loss": 1.3359,
+      "step": 6492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5202704602250587e-05,
+      "loss": 0.7031,
+      "step": 6493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5201358194610566e-05,
+      "loss": 0.6992,
+      "step": 6494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.520001165769664e-05,
+      "loss": 1.3828,
+      "step": 6495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5198664991542278e-05,
+      "loss": 1.375,
+      "step": 6496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5197318196180942e-05,
+      "loss": 0.8906,
+      "step": 6497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5195971271646111e-05,
+      "loss": 1.1406,
+      "step": 6498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5194624217971257e-05,
+      "loss": 0.8398,
+      "step": 6499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5193277035189865e-05,
+      "loss": 0.8203,
+      "step": 6500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5191929723335412e-05,
+      "loss": 1.0156,
+      "step": 6501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.519058228244139e-05,
+      "loss": 1.6797,
+      "step": 6502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5189234712541283e-05,
+      "loss": 0.8828,
+      "step": 6503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5187887013668588e-05,
+      "loss": 1.125,
+      "step": 6504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5186539185856797e-05,
+      "loss": 1.1172,
+      "step": 6505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5185191229139409e-05,
+      "loss": 0.5039,
+      "step": 6506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5183843143549926e-05,
+      "loss": 0.3398,
+      "step": 6507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5182494929121855e-05,
+      "loss": 0.4316,
+      "step": 6508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5181146585888704e-05,
+      "loss": 0.875,
+      "step": 6509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5179798113883984e-05,
+      "loss": 0.9922,
+      "step": 6510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5178449513141207e-05,
+      "loss": 1.0625,
+      "step": 6511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5177100783693894e-05,
+      "loss": 2.0312,
+      "step": 6512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5175751925575565e-05,
+      "loss": 1.0391,
+      "step": 6513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5174402938819745e-05,
+      "loss": 1.0938,
+      "step": 6514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.517305382345996e-05,
+      "loss": 0.4512,
+      "step": 6515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5171704579529748e-05,
+      "loss": 0.582,
+      "step": 6516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.517035520706263e-05,
+      "loss": 1.1016,
+      "step": 6517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5169005706092152e-05,
+      "loss": 1.3203,
+      "step": 6518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5167656076651855e-05,
+      "loss": 1.7109,
+      "step": 6519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5166306318775278e-05,
+      "loss": 1.7031,
+      "step": 6520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5164956432495973e-05,
+      "loss": 1.1484,
+      "step": 6521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5163606417847483e-05,
+      "loss": 0.8047,
+      "step": 6522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5162256274863367e-05,
+      "loss": 1.2109,
+      "step": 6523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.516090600357718e-05,
+      "loss": 0.9922,
+      "step": 6524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5159555604022478e-05,
+      "loss": 1.4844,
+      "step": 6525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5158205076232828e-05,
+      "loss": 0.375,
+      "step": 6526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5156854420241793e-05,
+      "loss": 0.5078,
+      "step": 6527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5155503636082945e-05,
+      "loss": 1.0391,
+      "step": 6528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.515415272378985e-05,
+      "loss": 1.4453,
+      "step": 6529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5152801683396092e-05,
+      "loss": 1.0781,
+      "step": 6530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5151450514935242e-05,
+      "loss": 1.1953,
+      "step": 6531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5150099218440888e-05,
+      "loss": 0.6445,
+      "step": 6532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5148747793946612e-05,
+      "loss": 1.4141,
+      "step": 6533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5147396241486e-05,
+      "loss": 0.6211,
+      "step": 6534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5146044561092645e-05,
+      "loss": 1.0078,
+      "step": 6535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5144692752800145e-05,
+      "loss": 0.8477,
+      "step": 6536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5143340816642094e-05,
+      "loss": 0.2988,
+      "step": 6537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5141988752652093e-05,
+      "loss": 1.0391,
+      "step": 6538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5140636560863746e-05,
+      "loss": 1.3672,
+      "step": 6539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5139284241310661e-05,
+      "loss": 1.0,
+      "step": 6540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.513793179402645e-05,
+      "loss": 1.3438,
+      "step": 6541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5136579219044722e-05,
+      "loss": 1.1328,
+      "step": 6542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5135226516399094e-05,
+      "loss": 1.2812,
+      "step": 6543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5133873686123195e-05,
+      "loss": 1.0078,
+      "step": 6544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5132520728250635e-05,
+      "loss": 0.4746,
+      "step": 6545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5131167642815055e-05,
+      "loss": 1.0234,
+      "step": 6546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5129814429850066e-05,
+      "loss": 0.8594,
+      "step": 6547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5128461089389317e-05,
+      "loss": 0.9883,
+      "step": 6548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5127107621466434e-05,
+      "loss": 0.6602,
+      "step": 6549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5125754026115062e-05,
+      "loss": 1.2578,
+      "step": 6550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.512440030336884e-05,
+      "loss": 0.7617,
+      "step": 6551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5123046453261414e-05,
+      "loss": 1.5625,
+      "step": 6552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5121692475826431e-05,
+      "loss": 0.8008,
+      "step": 6553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5120338371097545e-05,
+      "loss": 1.7578,
+      "step": 6554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5118984139108407e-05,
+      "loss": 0.9883,
+      "step": 6555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.511762977989268e-05,
+      "loss": 0.9062,
+      "step": 6556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5116275293484022e-05,
+      "loss": 0.8242,
+      "step": 6557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5114920679916096e-05,
+      "loss": 1.0312,
+      "step": 6558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5113565939222571e-05,
+      "loss": 0.8828,
+      "step": 6559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.511221107143712e-05,
+      "loss": 1.0234,
+      "step": 6560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5110856076593413e-05,
+      "loss": 0.8555,
+      "step": 6561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5109500954725124e-05,
+      "loss": 1.2344,
+      "step": 6562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5108145705865945e-05,
+      "loss": 0.8672,
+      "step": 6563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5106790330049546e-05,
+      "loss": 0.5664,
+      "step": 6564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5105434827309622e-05,
+      "loss": 0.3555,
+      "step": 6565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5104079197679857e-05,
+      "loss": 0.6523,
+      "step": 6566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5102723441193944e-05,
+      "loss": 1.0078,
+      "step": 6567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5101367557885585e-05,
+      "loss": 1.3516,
+      "step": 6568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5100011547788472e-05,
+      "loss": 0.9453,
+      "step": 6569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5098655410936312e-05,
+      "loss": 1.3203,
+      "step": 6570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5097299147362807e-05,
+      "loss": 1.2422,
+      "step": 6571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5095942757101664e-05,
+      "loss": 1.1953,
+      "step": 6572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.50945862401866e-05,
+      "loss": 0.9023,
+      "step": 6573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5093229596651327e-05,
+      "loss": 0.8242,
+      "step": 6574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5091872826529561e-05,
+      "loss": 1.1484,
+      "step": 6575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.509051592985502e-05,
+      "loss": 0.8711,
+      "step": 6576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508915890666144e-05,
+      "loss": 1.2266,
+      "step": 6577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5087801756982536e-05,
+      "loss": 2.0312,
+      "step": 6578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508644448085205e-05,
+      "loss": 1.0547,
+      "step": 6579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5085087078303702e-05,
+      "loss": 0.4492,
+      "step": 6580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5083729549371238e-05,
+      "loss": 0.9258,
+      "step": 6581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5082371894088396e-05,
+      "loss": 1.2734,
+      "step": 6582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5081014112488918e-05,
+      "loss": 0.6797,
+      "step": 6583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507965620460655e-05,
+      "loss": 1.1094,
+      "step": 6584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5078298170475046e-05,
+      "loss": 0.4102,
+      "step": 6585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507694001012815e-05,
+      "loss": 0.7734,
+      "step": 6586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5075581723599623e-05,
+      "loss": 1.0703,
+      "step": 6587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5074223310923222e-05,
+      "loss": 1.2812,
+      "step": 6588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507286477213271e-05,
+      "loss": 0.7539,
+      "step": 6589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5071506107261848e-05,
+      "loss": 0.6992,
+      "step": 6590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5070147316344414e-05,
+      "loss": 1.5703,
+      "step": 6591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5068788399414167e-05,
+      "loss": 1.3984,
+      "step": 6592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5067429356504892e-05,
+      "loss": 0.9141,
+      "step": 6593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5066070187650357e-05,
+      "loss": 0.9766,
+      "step": 6594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5064710892884349e-05,
+      "loss": 1.1641,
+      "step": 6595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5063351472240648e-05,
+      "loss": 1.3125,
+      "step": 6596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5061991925753043e-05,
+      "loss": 1.1016,
+      "step": 6597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5060632253455328e-05,
+      "loss": 1.1094,
+      "step": 6598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5059272455381289e-05,
+      "loss": 0.8477,
+      "step": 6599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5057912531564723e-05,
+      "loss": 0.7109,
+      "step": 6600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5056552482039433e-05,
+      "loss": 1.2891,
+      "step": 6601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.505519230683922e-05,
+      "loss": 0.8125,
+      "step": 6602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.505383200599789e-05,
+      "loss": 0.6445,
+      "step": 6603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.505247157954925e-05,
+      "loss": 0.6953,
+      "step": 6604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5051111027527116e-05,
+      "loss": 1.2031,
+      "step": 6605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5049750349965295e-05,
+      "loss": 0.375,
+      "step": 6606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5048389546897615e-05,
+      "loss": 1.3203,
+      "step": 6607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.504702861835789e-05,
+      "loss": 1.1094,
+      "step": 6608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5045667564379949e-05,
+      "loss": 0.8594,
+      "step": 6609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5044306384997615e-05,
+      "loss": 0.793,
+      "step": 6610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5042945080244723e-05,
+      "loss": 0.4082,
+      "step": 6611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5041583650155103e-05,
+      "loss": 0.3633,
+      "step": 6612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5040222094762596e-05,
+      "loss": 0.2812,
+      "step": 6613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5038860414101035e-05,
+      "loss": 1.1641,
+      "step": 6614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503749860820427e-05,
+      "loss": 0.543,
+      "step": 6615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5036136677106146e-05,
+      "loss": 0.9648,
+      "step": 6616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503477462084051e-05,
+      "loss": 1.1406,
+      "step": 6617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5033412439441213e-05,
+      "loss": 1.5,
+      "step": 6618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5032050132942115e-05,
+      "loss": 1.0234,
+      "step": 6619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503068770137707e-05,
+      "loss": 1.2812,
+      "step": 6620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5029325144779945e-05,
+      "loss": 1.3359,
+      "step": 6621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5027962463184597e-05,
+      "loss": 0.6602,
+      "step": 6622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5026599656624903e-05,
+      "loss": 1.1719,
+      "step": 6623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5025236725134726e-05,
+      "loss": 0.918,
+      "step": 6624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5023873668747948e-05,
+      "loss": 1.5547,
+      "step": 6625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502251048749844e-05,
+      "loss": 1.0391,
+      "step": 6626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5021147181420086e-05,
+      "loss": 1.4297,
+      "step": 6627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5019783750546765e-05,
+      "loss": 0.4199,
+      "step": 6628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5018420194912369e-05,
+      "loss": 1.1562,
+      "step": 6629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5017056514550783e-05,
+      "loss": 1.2109,
+      "step": 6630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5015692709495904e-05,
+      "loss": 1.0312,
+      "step": 6631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5014328779781626e-05,
+      "loss": 1.4219,
+      "step": 6632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5012964725441844e-05,
+      "loss": 0.5391,
+      "step": 6633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5011600546510466e-05,
+      "loss": 1.8594,
+      "step": 6634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5010236243021394e-05,
+      "loss": 1.1172,
+      "step": 6635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5008871815008536e-05,
+      "loss": 0.2988,
+      "step": 6636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5007507262505808e-05,
+      "loss": 0.8555,
+      "step": 6637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5006142585547117e-05,
+      "loss": 1.2266,
+      "step": 6638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5004777784166385e-05,
+      "loss": 0.2451,
+      "step": 6639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5003412858397535e-05,
+      "loss": 1.0391,
+      "step": 6640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5002047808274483e-05,
+      "loss": 1.3125,
+      "step": 6641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.500068263383116e-05,
+      "loss": 0.5625,
+      "step": 6642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4999317335101498e-05,
+      "loss": 0.9258,
+      "step": 6643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4997951912119429e-05,
+      "loss": 0.5742,
+      "step": 6644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4996586364918888e-05,
+      "loss": 2.0781,
+      "step": 6645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4995220693533812e-05,
+      "loss": 1.4766,
+      "step": 6646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4993854897998149e-05,
+      "loss": 1.3047,
+      "step": 6647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4992488978345839e-05,
+      "loss": 0.7305,
+      "step": 6648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4991122934610833e-05,
+      "loss": 0.793,
+      "step": 6649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4989756766827079e-05,
+      "loss": 1.0078,
+      "step": 6650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4988390475028538e-05,
+      "loss": 0.8047,
+      "step": 6651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4987024059249161e-05,
+      "loss": 0.4668,
+      "step": 6652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4985657519522912e-05,
+      "loss": 0.8789,
+      "step": 6653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4984290855883758e-05,
+      "loss": 0.7383,
+      "step": 6654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4982924068365656e-05,
+      "loss": 0.9531,
+      "step": 6655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4981557157002588e-05,
+      "loss": 0.7227,
+      "step": 6656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4980190121828519e-05,
+      "loss": 0.9961,
+      "step": 6657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4978822962877428e-05,
+      "loss": 0.2578,
+      "step": 6658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4977455680183295e-05,
+      "loss": 1.0703,
+      "step": 6659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4976088273780098e-05,
+      "loss": 1.0312,
+      "step": 6660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4974720743701826e-05,
+      "loss": 1.4141,
+      "step": 6661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.497335308998247e-05,
+      "loss": 1.1172,
+      "step": 6662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4971985312656014e-05,
+      "loss": 0.9219,
+      "step": 6663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4970617411756461e-05,
+      "loss": 0.3691,
+      "step": 6664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4969249387317802e-05,
+      "loss": 0.9297,
+      "step": 6665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4967881239374041e-05,
+      "loss": 0.7188,
+      "step": 6666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4966512967959177e-05,
+      "loss": 1.1797,
+      "step": 6667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4965144573107226e-05,
+      "loss": 0.6289,
+      "step": 6668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4963776054852188e-05,
+      "loss": 0.4434,
+      "step": 6669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4962407413228086e-05,
+      "loss": 0.6641,
+      "step": 6670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4961038648268926e-05,
+      "loss": 1.6953,
+      "step": 6671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4959669760008733e-05,
+      "loss": 0.3789,
+      "step": 6672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4958300748481528e-05,
+      "loss": 1.6016,
+      "step": 6673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4956931613721338e-05,
+      "loss": 0.6797,
+      "step": 6674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4955562355762187e-05,
+      "loss": 0.9375,
+      "step": 6675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4954192974638111e-05,
+      "loss": 1.5469,
+      "step": 6676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4952823470383142e-05,
+      "loss": 1.2734,
+      "step": 6677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4951453843031318e-05,
+      "loss": 0.9961,
+      "step": 6678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4950084092616679e-05,
+      "loss": 1.0547,
+      "step": 6679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4948714219173268e-05,
+      "loss": 0.9492,
+      "step": 6680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4947344222735134e-05,
+      "loss": 0.6641,
+      "step": 6681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4945974103336327e-05,
+      "loss": 0.3203,
+      "step": 6682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4944603861010894e-05,
+      "loss": 0.7031,
+      "step": 6683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4943233495792897e-05,
+      "loss": 0.4766,
+      "step": 6684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4941863007716393e-05,
+      "loss": 1.1484,
+      "step": 6685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4940492396815442e-05,
+      "loss": 1.2344,
+      "step": 6686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4939121663124115e-05,
+      "loss": 1.2656,
+      "step": 6687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4937750806676472e-05,
+      "loss": 1.0938,
+      "step": 6688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4936379827506589e-05,
+      "loss": 0.7266,
+      "step": 6689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4935008725648539e-05,
+      "loss": 0.7695,
+      "step": 6690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.49336375011364e-05,
+      "loss": 1.625,
+      "step": 6691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4932266154004252e-05,
+      "loss": 1.2734,
+      "step": 6692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4930894684286175e-05,
+      "loss": 0.3926,
+      "step": 6693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.492952309201626e-05,
+      "loss": 1.2812,
+      "step": 6694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4928151377228593e-05,
+      "loss": 1.25,
+      "step": 6695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.492677953995727e-05,
+      "loss": 0.9727,
+      "step": 6696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4925407580236382e-05,
+      "loss": 1.0469,
+      "step": 6697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4924035498100034e-05,
+      "loss": 0.8164,
+      "step": 6698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.492266329358232e-05,
+      "loss": 0.9062,
+      "step": 6699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.492129096671735e-05,
+      "loss": 0.8906,
+      "step": 6700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4919918517539226e-05,
+      "loss": 0.957,
+      "step": 6701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4918545946082067e-05,
+      "loss": 0.5391,
+      "step": 6702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4917173252379982e-05,
+      "loss": 0.5508,
+      "step": 6703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4915800436467084e-05,
+      "loss": 1.3359,
+      "step": 6704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.49144274983775e-05,
+      "loss": 1.7031,
+      "step": 6705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.491305443814535e-05,
+      "loss": 1.875,
+      "step": 6706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4911681255804758e-05,
+      "loss": 0.2539,
+      "step": 6707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4910307951389854e-05,
+      "loss": 1.1719,
+      "step": 6708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4908934524934771e-05,
+      "loss": 0.4805,
+      "step": 6709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4907560976473642e-05,
+      "loss": 0.4395,
+      "step": 6710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4906187306040608e-05,
+      "loss": 1.0078,
+      "step": 6711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4904813513669804e-05,
+      "loss": 1.0078,
+      "step": 6712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4903439599395384e-05,
+      "loss": 1.1328,
+      "step": 6713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4902065563251484e-05,
+      "loss": 0.9609,
+      "step": 6714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4900691405272265e-05,
+      "loss": 0.5586,
+      "step": 6715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4899317125491868e-05,
+      "loss": 0.5742,
+      "step": 6716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4897942723944462e-05,
+      "loss": 0.6562,
+      "step": 6717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4896568200664197e-05,
+      "loss": 0.1865,
+      "step": 6718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4895193555685237e-05,
+      "loss": 0.2109,
+      "step": 6719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4893818789041752e-05,
+      "loss": 0.7266,
+      "step": 6720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4892443900767905e-05,
+      "loss": 0.7109,
+      "step": 6721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4891068890897868e-05,
+      "loss": 0.6875,
+      "step": 6722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.488969375946582e-05,
+      "loss": 1.3906,
+      "step": 6723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4888318506505931e-05,
+      "loss": 2.0156,
+      "step": 6724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4886943132052386e-05,
+      "loss": 1.0938,
+      "step": 6725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4885567636139368e-05,
+      "loss": 1.3516,
+      "step": 6726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4884192018801066e-05,
+      "loss": 0.8047,
+      "step": 6727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4882816280071659e-05,
+      "loss": 0.9336,
+      "step": 6728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4881440419985353e-05,
+      "loss": 1.1719,
+      "step": 6729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4880064438576333e-05,
+      "loss": 1.1484,
+      "step": 6730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4878688335878809e-05,
+      "loss": 1.3672,
+      "step": 6731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4877312111926968e-05,
+      "loss": 1.3047,
+      "step": 6732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4875935766755024e-05,
+      "loss": 0.7031,
+      "step": 6733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.487455930039718e-05,
+      "loss": 1.3438,
+      "step": 6734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4873182712887653e-05,
+      "loss": 0.9531,
+      "step": 6735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4871806004260649e-05,
+      "loss": 0.6914,
+      "step": 6736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4870429174550389e-05,
+      "loss": 0.4316,
+      "step": 6737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4869052223791092e-05,
+      "loss": 0.4727,
+      "step": 6738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4867675152016978e-05,
+      "loss": 0.9609,
+      "step": 6739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4866297959262275e-05,
+      "loss": 0.6641,
+      "step": 6740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4864920645561211e-05,
+      "loss": 1.3516,
+      "step": 6741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.486354321094802e-05,
+      "loss": 0.8477,
+      "step": 6742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4862165655456933e-05,
+      "loss": 1.0781,
+      "step": 6743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4860787979122187e-05,
+      "loss": 1.2031,
+      "step": 6744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4859410181978027e-05,
+      "loss": 1.1875,
+      "step": 6745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4858032264058691e-05,
+      "loss": 0.9023,
+      "step": 6746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.485665422539843e-05,
+      "loss": 0.8281,
+      "step": 6747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4855276066031493e-05,
+      "loss": 0.5195,
+      "step": 6748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4853897785992131e-05,
+      "loss": 0.7695,
+      "step": 6749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.48525193853146e-05,
+      "loss": 0.9805,
+      "step": 6750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.485114086403316e-05,
+      "loss": 0.8008,
+      "step": 6751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.484976222218207e-05,
+      "loss": 1.0,
+      "step": 6752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4848383459795601e-05,
+      "loss": 0.6328,
+      "step": 6753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.484700457690801e-05,
+      "loss": 0.7617,
+      "step": 6754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4845625573553577e-05,
+      "loss": 1.9141,
+      "step": 6755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.484424644976657e-05,
+      "loss": 1.2109,
+      "step": 6756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4842867205581269e-05,
+      "loss": 0.707,
+      "step": 6757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.484148784103195e-05,
+      "loss": 1.1641,
+      "step": 6758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.48401083561529e-05,
+      "loss": 0.4004,
+      "step": 6759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4838728750978399e-05,
+      "loss": 0.5078,
+      "step": 6760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4837349025542743e-05,
+      "loss": 1.8438,
+      "step": 6761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4835969179880216e-05,
+      "loss": 1.7422,
+      "step": 6762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4834589214025117e-05,
+      "loss": 0.5391,
+      "step": 6763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.483320912801174e-05,
+      "loss": 0.293,
+      "step": 6764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4831828921874392e-05,
+      "loss": 1.2422,
+      "step": 6765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.483044859564737e-05,
+      "loss": 0.75,
+      "step": 6766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4829068149364983e-05,
+      "loss": 1.4453,
+      "step": 6767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.482768758306154e-05,
+      "loss": 1.2188,
+      "step": 6768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4826306896771353e-05,
+      "loss": 1.2188,
+      "step": 6769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4824926090528737e-05,
+      "loss": 0.9102,
+      "step": 6770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4823545164368012e-05,
+      "loss": 1.0234,
+      "step": 6771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.48221641183235e-05,
+      "loss": 1.3984,
+      "step": 6772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4820782952429522e-05,
+      "loss": 1.5625,
+      "step": 6773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4819401666720407e-05,
+      "loss": 0.8164,
+      "step": 6774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4818020261230488e-05,
+      "loss": 0.2578,
+      "step": 6775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4816638735994096e-05,
+      "loss": 0.8516,
+      "step": 6776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4815257091045565e-05,
+      "loss": 0.582,
+      "step": 6777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4813875326419237e-05,
+      "loss": 0.8906,
+      "step": 6778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4812493442149454e-05,
+      "loss": 0.4531,
+      "step": 6779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.481111143827056e-05,
+      "loss": 0.8008,
+      "step": 6780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4809729314816906e-05,
+      "loss": 0.7812,
+      "step": 6781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4808347071822841e-05,
+      "loss": 0.6992,
+      "step": 6782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4806964709322717e-05,
+      "loss": 1.1016,
+      "step": 6783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4805582227350894e-05,
+      "loss": 1.3125,
+      "step": 6784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4804199625941732e-05,
+      "loss": 0.6055,
+      "step": 6785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4802816905129594e-05,
+      "loss": 0.5625,
+      "step": 6786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4801434064948845e-05,
+      "loss": 0.832,
+      "step": 6787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4800051105433852e-05,
+      "loss": 0.9492,
+      "step": 6788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4798668026618991e-05,
+      "loss": 0.7812,
+      "step": 6789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4797284828538634e-05,
+      "loss": 1.6953,
+      "step": 6790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4795901511227158e-05,
+      "loss": 1.0859,
+      "step": 6791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4794518074718951e-05,
+      "loss": 1.0625,
+      "step": 6792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4793134519048384e-05,
+      "loss": 1.2031,
+      "step": 6793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4791750844249857e-05,
+      "loss": 1.2969,
+      "step": 6794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.479036705035775e-05,
+      "loss": 1.1562,
+      "step": 6795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4788983137406463e-05,
+      "loss": 0.9766,
+      "step": 6796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4787599105430385e-05,
+      "loss": 0.7891,
+      "step": 6797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.478621495446392e-05,
+      "loss": 1.1797,
+      "step": 6798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4784830684541465e-05,
+      "loss": 1.1328,
+      "step": 6799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4783446295697427e-05,
+      "loss": 0.9062,
+      "step": 6800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4782061787966213e-05,
+      "loss": 1.1641,
+      "step": 6801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4780677161382232e-05,
+      "loss": 1.1641,
+      "step": 6802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4779292415979902e-05,
+      "loss": 0.3379,
+      "step": 6803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4777907551793634e-05,
+      "loss": 0.5547,
+      "step": 6804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4776522568857847e-05,
+      "loss": 1.1797,
+      "step": 6805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.477513746720697e-05,
+      "loss": 1.0312,
+      "step": 6806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.477375224687542e-05,
+      "loss": 0.165,
+      "step": 6807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4772366907897628e-05,
+      "loss": 1.4922,
+      "step": 6808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4770981450308029e-05,
+      "loss": 1.6172,
+      "step": 6809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4769595874141052e-05,
+      "loss": 1.1406,
+      "step": 6810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4768210179431136e-05,
+      "loss": 0.4961,
+      "step": 6811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4766824366212719e-05,
+      "loss": 1.1641,
+      "step": 6812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4765438434520248e-05,
+      "loss": 0.3574,
+      "step": 6813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4764052384388163e-05,
+      "loss": 0.4082,
+      "step": 6814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4762666215850918e-05,
+      "loss": 0.9805,
+      "step": 6815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4761279928942962e-05,
+      "loss": 1.3281,
+      "step": 6816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.475989352369875e-05,
+      "loss": 0.8125,
+      "step": 6817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4758507000152739e-05,
+      "loss": 0.9219,
+      "step": 6818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.475712035833939e-05,
+      "loss": 0.7773,
+      "step": 6819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.475573359829317e-05,
+      "loss": 1.6094,
+      "step": 6820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4754346720048535e-05,
+      "loss": 1.125,
+      "step": 6821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4752959723639966e-05,
+      "loss": 1.0078,
+      "step": 6822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4751572609101929e-05,
+      "loss": 0.6523,
+      "step": 6823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.47501853764689e-05,
+      "loss": 1.5938,
+      "step": 6824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.474879802577536e-05,
+      "loss": 1.0156,
+      "step": 6825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4747410557055786e-05,
+      "loss": 0.7773,
+      "step": 6826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4746022970344662e-05,
+      "loss": 1.2109,
+      "step": 6827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4744635265676478e-05,
+      "loss": 1.2031,
+      "step": 6828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4743247443085725e-05,
+      "loss": 1.2109,
+      "step": 6829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.474185950260689e-05,
+      "loss": 1.8516,
+      "step": 6830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4740471444274474e-05,
+      "loss": 0.6211,
+      "step": 6831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4739083268122971e-05,
+      "loss": 1.1484,
+      "step": 6832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4737694974186887e-05,
+      "loss": 0.5547,
+      "step": 6833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4736306562500726e-05,
+      "loss": 0.9609,
+      "step": 6834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4734918033098991e-05,
+      "loss": 1.5,
+      "step": 6835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4733529386016197e-05,
+      "loss": 1.2109,
+      "step": 6836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4732140621286854e-05,
+      "loss": 0.8477,
+      "step": 6837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4730751738945479e-05,
+      "loss": 1.3594,
+      "step": 6838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4729362739026596e-05,
+      "loss": 1.2188,
+      "step": 6839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.472797362156472e-05,
+      "loss": 0.9258,
+      "step": 6840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4726584386594379e-05,
+      "loss": 0.8438,
+      "step": 6841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4725195034150102e-05,
+      "loss": 0.6602,
+      "step": 6842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4723805564266415e-05,
+      "loss": 0.7773,
+      "step": 6843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.472241597697786e-05,
+      "loss": 0.5312,
+      "step": 6844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4721026272318968e-05,
+      "loss": 0.7109,
+      "step": 6845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4719636450324278e-05,
+      "loss": 0.8711,
+      "step": 6846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4718246511028331e-05,
+      "loss": 1.2734,
+      "step": 6847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.471685645446568e-05,
+      "loss": 0.9414,
+      "step": 6848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4715466280670864e-05,
+      "loss": 0.7656,
+      "step": 6849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4714075989678444e-05,
+      "loss": 0.6641,
+      "step": 6850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4712685581522965e-05,
+      "loss": 1.4844,
+      "step": 6851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4711295056238989e-05,
+      "loss": 0.3789,
+      "step": 6852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4709904413861074e-05,
+      "loss": 0.4453,
+      "step": 6853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4708513654423786e-05,
+      "loss": 1.5078,
+      "step": 6854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4707122777961688e-05,
+      "loss": 1.2031,
+      "step": 6855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4705731784509348e-05,
+      "loss": 0.7383,
+      "step": 6856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4704340674101339e-05,
+      "loss": 1.0547,
+      "step": 6857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4702949446772234e-05,
+      "loss": 1.0312,
+      "step": 6858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4701558102556613e-05,
+      "loss": 1.7578,
+      "step": 6859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4700166641489055e-05,
+      "loss": 0.2021,
+      "step": 6860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4698775063604143e-05,
+      "loss": 0.5586,
+      "step": 6861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4697383368936462e-05,
+      "loss": 1.6406,
+      "step": 6862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4695991557520604e-05,
+      "loss": 0.7695,
+      "step": 6863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4694599629391157e-05,
+      "loss": 1.5391,
+      "step": 6864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4693207584582719e-05,
+      "loss": 1.0312,
+      "step": 6865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4691815423129883e-05,
+      "loss": 1.3516,
+      "step": 6866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4690423145067259e-05,
+      "loss": 1.1875,
+      "step": 6867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4689030750429441e-05,
+      "loss": 0.9961,
+      "step": 6868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4687638239251043e-05,
+      "loss": 1.9922,
+      "step": 6869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4686245611566665e-05,
+      "loss": 0.5625,
+      "step": 6870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4684852867410926e-05,
+      "loss": 1.3281,
+      "step": 6871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4683460006818442e-05,
+      "loss": 1.1172,
+      "step": 6872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4682067029823828e-05,
+      "loss": 1.1172,
+      "step": 6873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4680673936461704e-05,
+      "loss": 1.2891,
+      "step": 6874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4679280726766699e-05,
+      "loss": 1.4375,
+      "step": 6875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4677887400773432e-05,
+      "loss": 1.1094,
+      "step": 6876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4676493958516537e-05,
+      "loss": 0.9102,
+      "step": 6877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4675100400030648e-05,
+      "loss": 0.6797,
+      "step": 6878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4673706725350396e-05,
+      "loss": 0.7695,
+      "step": 6879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4672312934510422e-05,
+      "loss": 1.0781,
+      "step": 6880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.467091902754537e-05,
+      "loss": 0.8164,
+      "step": 6881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4669525004489875e-05,
+      "loss": 0.7344,
+      "step": 6882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4668130865378593e-05,
+      "loss": 0.9062,
+      "step": 6883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.466673661024617e-05,
+      "loss": 0.5898,
+      "step": 6884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4665342239127259e-05,
+      "loss": 1.0547,
+      "step": 6885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4663947752056514e-05,
+      "loss": 0.9922,
+      "step": 6886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4662553149068594e-05,
+      "loss": 0.6836,
+      "step": 6887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4661158430198164e-05,
+      "loss": 1.1172,
+      "step": 6888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4659763595479883e-05,
+      "loss": 0.7734,
+      "step": 6889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4658368644948422e-05,
+      "loss": 1.4375,
+      "step": 6890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4656973578638449e-05,
+      "loss": 1.3125,
+      "step": 6891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4655578396584638e-05,
+      "loss": 1.3516,
+      "step": 6892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.465418309882166e-05,
+      "loss": 0.8281,
+      "step": 6893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4652787685384198e-05,
+      "loss": 1.0781,
+      "step": 6894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4651392156306937e-05,
+      "loss": 1.3828,
+      "step": 6895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.464999651162455e-05,
+      "loss": 1.0391,
+      "step": 6896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.464860075137174e-05,
+      "loss": 1.2734,
+      "step": 6897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.464720487558318e-05,
+      "loss": 1.1328,
+      "step": 6898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4645808884293576e-05,
+      "loss": 1.1875,
+      "step": 6899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4644412777537617e-05,
+      "loss": 0.8633,
+      "step": 6900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4643016555350005e-05,
+      "loss": 1.5625,
+      "step": 6901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.464162021776544e-05,
+      "loss": 0.3164,
+      "step": 6902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4640223764818627e-05,
+      "loss": 2.0156,
+      "step": 6903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4638827196544272e-05,
+      "loss": 0.4648,
+      "step": 6904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4637430512977085e-05,
+      "loss": 0.375,
+      "step": 6905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4636033714151781e-05,
+      "loss": 0.5938,
+      "step": 6906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4634636800103072e-05,
+      "loss": 0.2266,
+      "step": 6907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4633239770865684e-05,
+      "loss": 0.8047,
+      "step": 6908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4631842626474332e-05,
+      "loss": 1.0938,
+      "step": 6909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.463044536696374e-05,
+      "loss": 0.6484,
+      "step": 6910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4629047992368643e-05,
+      "loss": 1.1328,
+      "step": 6911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4627650502723763e-05,
+      "loss": 0.4766,
+      "step": 6912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4626252898063836e-05,
+      "loss": 0.5,
+      "step": 6913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4624855178423599e-05,
+      "loss": 1.2969,
+      "step": 6914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.462345734383779e-05,
+      "loss": 0.8438,
+      "step": 6915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.462205939434115e-05,
+      "loss": 1.7031,
+      "step": 6916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4620661329968424e-05,
+      "loss": 0.8086,
+      "step": 6917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.461926315075436e-05,
+      "loss": 0.4727,
+      "step": 6918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4617864856733705e-05,
+      "loss": 1.6406,
+      "step": 6919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4616466447941216e-05,
+      "loss": 0.4883,
+      "step": 6920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4615067924411646e-05,
+      "loss": 1.5391,
+      "step": 6921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4613669286179755e-05,
+      "loss": 1.0,
+      "step": 6922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4612270533280306e-05,
+      "loss": 0.6289,
+      "step": 6923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4610871665748059e-05,
+      "loss": 1.6641,
+      "step": 6924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4609472683617787e-05,
+      "loss": 1.4219,
+      "step": 6925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4608073586924257e-05,
+      "loss": 1.7812,
+      "step": 6926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.460667437570224e-05,
+      "loss": 1.1094,
+      "step": 6927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4605275049986517e-05,
+      "loss": 0.6328,
+      "step": 6928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4603875609811862e-05,
+      "loss": 0.9961,
+      "step": 6929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4602476055213062e-05,
+      "loss": 1.4453,
+      "step": 6930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4601076386224894e-05,
+      "loss": 1.1094,
+      "step": 6931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4599676602882147e-05,
+      "loss": 1.5859,
+      "step": 6932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4598276705219616e-05,
+      "loss": 0.8047,
+      "step": 6933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.459687669327209e-05,
+      "loss": 0.5312,
+      "step": 6934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4595476567074367e-05,
+      "loss": 1.5547,
+      "step": 6935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4594076326661242e-05,
+      "loss": 0.9258,
+      "step": 6936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4592675972067521e-05,
+      "loss": 0.707,
+      "step": 6937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4591275503328004e-05,
+      "loss": 0.8516,
+      "step": 6938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4589874920477498e-05,
+      "loss": 0.9766,
+      "step": 6939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4588474223550817e-05,
+      "loss": 0.8789,
+      "step": 6940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.458707341258277e-05,
+      "loss": 0.3828,
+      "step": 6941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4585672487608175e-05,
+      "loss": 0.8828,
+      "step": 6942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4584271448661849e-05,
+      "loss": 1.1875,
+      "step": 6943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4582870295778615e-05,
+      "loss": 0.9648,
+      "step": 6944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4581469028993293e-05,
+      "loss": 1.1406,
+      "step": 6945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4580067648340715e-05,
+      "loss": 1.4219,
+      "step": 6946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4578666153855707e-05,
+      "loss": 1.2656,
+      "step": 6947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4577264545573107e-05,
+      "loss": 1.0156,
+      "step": 6948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4575862823527742e-05,
+      "loss": 1.1094,
+      "step": 6949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4574460987754456e-05,
+      "loss": 0.8945,
+      "step": 6950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4573059038288089e-05,
+      "loss": 1.7188,
+      "step": 6951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4571656975163483e-05,
+      "loss": 1.3047,
+      "step": 6952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4570254798415488e-05,
+      "loss": 0.8008,
+      "step": 6953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4568852508078952e-05,
+      "loss": 1.8828,
+      "step": 6954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4567450104188726e-05,
+      "loss": 0.4531,
+      "step": 6955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4566047586779666e-05,
+      "loss": 0.6094,
+      "step": 6956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4564644955886628e-05,
+      "loss": 1.625,
+      "step": 6957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.456324221154448e-05,
+      "loss": 0.3965,
+      "step": 6958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4561839353788076e-05,
+      "loss": 0.7773,
+      "step": 6959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4560436382652293e-05,
+      "loss": 0.9844,
+      "step": 6960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4559033298171987e-05,
+      "loss": 0.459,
+      "step": 6961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4557630100382044e-05,
+      "loss": 0.7695,
+      "step": 6962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.455622678931733e-05,
+      "loss": 0.3203,
+      "step": 6963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4554823365012724e-05,
+      "loss": 1.9062,
+      "step": 6964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4553419827503107e-05,
+      "loss": 1.25,
+      "step": 6965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4552016176823367e-05,
+      "loss": 0.4395,
+      "step": 6966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4550612413008383e-05,
+      "loss": 1.0938,
+      "step": 6967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4549208536093049e-05,
+      "loss": 1.8672,
+      "step": 6968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4547804546112254e-05,
+      "loss": 1.7891,
+      "step": 6969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4546400443100894e-05,
+      "loss": 0.6133,
+      "step": 6970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4544996227093864e-05,
+      "loss": 0.5469,
+      "step": 6971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4543591898126069e-05,
+      "loss": 1.4609,
+      "step": 6972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4542187456232407e-05,
+      "loss": 1.3359,
+      "step": 6973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4540782901447788e-05,
+      "loss": 1.2578,
+      "step": 6974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4539378233807116e-05,
+      "loss": 1.1094,
+      "step": 6975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4537973453345306e-05,
+      "loss": 0.8711,
+      "step": 6976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.453656856009727e-05,
+      "loss": 1.0938,
+      "step": 6977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4535163554097929e-05,
+      "loss": 0.832,
+      "step": 6978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4533758435382199e-05,
+      "loss": 1.3516,
+      "step": 6979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4532353203985003e-05,
+      "loss": 0.9531,
+      "step": 6980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4530947859941269e-05,
+      "loss": 0.5312,
+      "step": 6981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4529542403285924e-05,
+      "loss": 0.9453,
+      "step": 6982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4528136834053896e-05,
+      "loss": 0.8164,
+      "step": 6983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4526731152280123e-05,
+      "loss": 1.3203,
+      "step": 6984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4525325357999535e-05,
+      "loss": 0.4023,
+      "step": 6985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4523919451247084e-05,
+      "loss": 1.5156,
+      "step": 6986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.45225134320577e-05,
+      "loss": 0.375,
+      "step": 6987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4521107300466337e-05,
+      "loss": 0.5469,
+      "step": 6988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4519701056507933e-05,
+      "loss": 0.8125,
+      "step": 6989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4518294700217448e-05,
+      "loss": 1.2656,
+      "step": 6990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4516888231629832e-05,
+      "loss": 1.0703,
+      "step": 6991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4515481650780036e-05,
+      "loss": 0.3574,
+      "step": 6992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4514074957703027e-05,
+      "loss": 1.5078,
+      "step": 6993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4512668152433764e-05,
+      "loss": 1.1641,
+      "step": 6994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4511261235007212e-05,
+      "loss": 1.4062,
+      "step": 6995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4509854205458335e-05,
+      "loss": 1.2891,
+      "step": 6996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4508447063822106e-05,
+      "loss": 0.9727,
+      "step": 6997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4507039810133499e-05,
+      "loss": 0.7891,
+      "step": 6998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4505632444427486e-05,
+      "loss": 0.7539,
+      "step": 6999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4504224966739048e-05,
+      "loss": 0.6328,
+      "step": 7000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4502817377103167e-05,
+      "loss": 1.6016,
+      "step": 7001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4501409675554828e-05,
+      "loss": 1.4062,
+      "step": 7002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4500001862129014e-05,
+      "loss": 1.6797,
+      "step": 7003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4498593936860711e-05,
+      "loss": 0.9219,
+      "step": 7004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4497185899784927e-05,
+      "loss": 1.2344,
+      "step": 7005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4495777750936637e-05,
+      "loss": 1.1797,
+      "step": 7006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4494369490350858e-05,
+      "loss": 0.2422,
+      "step": 7007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4492961118062574e-05,
+      "loss": 0.7031,
+      "step": 7008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.44915526341068e-05,
+      "loss": 1.2656,
+      "step": 7009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.449014403851854e-05,
+      "loss": 0.9023,
+      "step": 7010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4488735331332797e-05,
+      "loss": 1.1484,
+      "step": 7011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4487326512584592e-05,
+      "loss": 0.6641,
+      "step": 7012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4485917582308933e-05,
+      "loss": 1.1328,
+      "step": 7013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4484508540540838e-05,
+      "loss": 1.5156,
+      "step": 7014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4483099387315332e-05,
+      "loss": 0.418,
+      "step": 7015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4481690122667432e-05,
+      "loss": 0.9297,
+      "step": 7016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4480280746632166e-05,
+      "loss": 0.9414,
+      "step": 7017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4478871259244563e-05,
+      "loss": 0.9961,
+      "step": 7018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4477461660539655e-05,
+      "loss": 0.8203,
+      "step": 7019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4476051950552472e-05,
+      "loss": 1.1094,
+      "step": 7020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4474642129318057e-05,
+      "loss": 0.6953,
+      "step": 7021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4473232196871443e-05,
+      "loss": 0.9492,
+      "step": 7022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4471822153247677e-05,
+      "loss": 1.4453,
+      "step": 7023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4470411998481801e-05,
+      "loss": 0.9844,
+      "step": 7024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4469001732608865e-05,
+      "loss": 0.4863,
+      "step": 7025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.446759135566392e-05,
+      "loss": 1.2578,
+      "step": 7026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4466180867682019e-05,
+      "loss": 0.6211,
+      "step": 7027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4464770268698214e-05,
+      "loss": 1.1797,
+      "step": 7028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.446335955874757e-05,
+      "loss": 1.3594,
+      "step": 7029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4461948737865143e-05,
+      "loss": 0.9102,
+      "step": 7030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4460537806086002e-05,
+      "loss": 0.8086,
+      "step": 7031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4459126763445208e-05,
+      "loss": 0.6445,
+      "step": 7032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4457715609977841e-05,
+      "loss": 1.0312,
+      "step": 7033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4456304345718964e-05,
+      "loss": 0.5508,
+      "step": 7034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.445489297070366e-05,
+      "loss": 0.8516,
+      "step": 7035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4453481484966997e-05,
+      "loss": 0.6367,
+      "step": 7036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4452069888544067e-05,
+      "loss": 1.1328,
+      "step": 7037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4450658181469947e-05,
+      "loss": 0.6719,
+      "step": 7038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4449246363779725e-05,
+      "loss": 1.3125,
+      "step": 7039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.444783443550849e-05,
+      "loss": 1.4453,
+      "step": 7040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4446422396691335e-05,
+      "loss": 1.1016,
+      "step": 7041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4445010247363354e-05,
+      "loss": 0.7266,
+      "step": 7042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4443597987559644e-05,
+      "loss": 1.5234,
+      "step": 7043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4442185617315306e-05,
+      "loss": 1.0859,
+      "step": 7044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.444077313666544e-05,
+      "loss": 1.2109,
+      "step": 7045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4439360545645155e-05,
+      "loss": 0.957,
+      "step": 7046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4437947844289556e-05,
+      "loss": 0.9844,
+      "step": 7047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.443653503263376e-05,
+      "loss": 1.4688,
+      "step": 7048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4435122110712877e-05,
+      "loss": 0.75,
+      "step": 7049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4433709078562018e-05,
+      "loss": 1.1172,
+      "step": 7050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4432295936216311e-05,
+      "loss": 1.2578,
+      "step": 7051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4430882683710878e-05,
+      "loss": 0.9727,
+      "step": 7052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4429469321080837e-05,
+      "loss": 1.5781,
+      "step": 7053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4428055848361324e-05,
+      "loss": 1.375,
+      "step": 7054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.442664226558746e-05,
+      "loss": 0.2676,
+      "step": 7055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4425228572794386e-05,
+      "loss": 0.8242,
+      "step": 7056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4423814770017233e-05,
+      "loss": 0.3301,
+      "step": 7057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.442240085729114e-05,
+      "loss": 1.0391,
+      "step": 7058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4420986834651251e-05,
+      "loss": 1.0547,
+      "step": 7059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.441957270213271e-05,
+      "loss": 1.1797,
+      "step": 7060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4418158459770656e-05,
+      "loss": 0.6016,
+      "step": 7061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4416744107600246e-05,
+      "loss": 1.1406,
+      "step": 7062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4415329645656636e-05,
+      "loss": 0.6133,
+      "step": 7063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4413915073974967e-05,
+      "loss": 0.6289,
+      "step": 7064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4412500392590408e-05,
+      "loss": 0.9688,
+      "step": 7065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4411085601538116e-05,
+      "loss": 1.2812,
+      "step": 7066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4409670700853253e-05,
+      "loss": 0.5234,
+      "step": 7067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4408255690570987e-05,
+      "loss": 1.5547,
+      "step": 7068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4406840570726485e-05,
+      "loss": 0.543,
+      "step": 7069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4405425341354921e-05,
+      "loss": 0.8086,
+      "step": 7070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4404010002491465e-05,
+      "loss": 1.0469,
+      "step": 7071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4402594554171294e-05,
+      "loss": 0.418,
+      "step": 7072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4401178996429588e-05,
+      "loss": 1.6016,
+      "step": 7073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4399763329301532e-05,
+      "loss": 1.9688,
+      "step": 7074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4398347552822307e-05,
+      "loss": 1.4453,
+      "step": 7075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4396931667027102e-05,
+      "loss": 0.8867,
+      "step": 7076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4395515671951107e-05,
+      "loss": 1.3594,
+      "step": 7077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4394099567629516e-05,
+      "loss": 0.6992,
+      "step": 7078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4392683354097523e-05,
+      "loss": 0.9102,
+      "step": 7079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4391267031390328e-05,
+      "loss": 1.2812,
+      "step": 7080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.438985059954313e-05,
+      "loss": 0.5938,
+      "step": 7081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4388434058591135e-05,
+      "loss": 0.8906,
+      "step": 7082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4387017408569545e-05,
+      "loss": 0.8164,
+      "step": 7083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4385600649513578e-05,
+      "loss": 1.2109,
+      "step": 7084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.438418378145844e-05,
+      "loss": 0.9375,
+      "step": 7085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4382766804439346e-05,
+      "loss": 0.7852,
+      "step": 7086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4381349718491513e-05,
+      "loss": 1.4609,
+      "step": 7087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4379932523650164e-05,
+      "loss": 1.5312,
+      "step": 7088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4378515219950517e-05,
+      "loss": 0.9414,
+      "step": 7089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4377097807427803e-05,
+      "loss": 1.1484,
+      "step": 7090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4375680286117246e-05,
+      "loss": 1.4297,
+      "step": 7091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4374262656054079e-05,
+      "loss": 1.0859,
+      "step": 7092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4372844917273533e-05,
+      "loss": 1.2812,
+      "step": 7093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4371427069810851e-05,
+      "loss": 1.6016,
+      "step": 7094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.437000911370126e-05,
+      "loss": 1.1797,
+      "step": 7095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4368591048980016e-05,
+      "loss": 0.9727,
+      "step": 7096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4367172875682351e-05,
+      "loss": 1.3281,
+      "step": 7097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4365754593843522e-05,
+      "loss": 1.3203,
+      "step": 7098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.436433620349877e-05,
+      "loss": 0.8672,
+      "step": 7099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4362917704683354e-05,
+      "loss": 1.1953,
+      "step": 7100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4361499097432528e-05,
+      "loss": 1.9062,
+      "step": 7101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4360080381781548e-05,
+      "loss": 0.1855,
+      "step": 7102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4358661557765673e-05,
+      "loss": 1.25,
+      "step": 7103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4357242625420172e-05,
+      "loss": 0.6719,
+      "step": 7104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4355823584780307e-05,
+      "loss": 0.7734,
+      "step": 7105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4354404435881346e-05,
+      "loss": 1.125,
+      "step": 7106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4352985178758564e-05,
+      "loss": 0.8867,
+      "step": 7107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.435156581344723e-05,
+      "loss": 0.4219,
+      "step": 7108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4350146339982622e-05,
+      "loss": 0.7695,
+      "step": 7109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4348726758400026e-05,
+      "loss": 0.5742,
+      "step": 7110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4347307068734714e-05,
+      "loss": 0.9883,
+      "step": 7111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4345887271021982e-05,
+      "loss": 1.0938,
+      "step": 7112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4344467365297105e-05,
+      "loss": 1.0938,
+      "step": 7113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4343047351595382e-05,
+      "loss": 0.3867,
+      "step": 7114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4341627229952102e-05,
+      "loss": 1.2969,
+      "step": 7115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4340207000402564e-05,
+      "loss": 0.6953,
+      "step": 7116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4338786662982063e-05,
+      "loss": 0.3066,
+      "step": 7117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.43373662177259e-05,
+      "loss": 0.5898,
+      "step": 7118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.433594566466938e-05,
+      "loss": 0.3691,
+      "step": 7119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4334525003847808e-05,
+      "loss": 1.1016,
+      "step": 7120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4333104235296496e-05,
+      "loss": 1.0859,
+      "step": 7121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.433168335905075e-05,
+      "loss": 0.8594,
+      "step": 7122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4330262375145886e-05,
+      "loss": 1.0312,
+      "step": 7123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4328841283617227e-05,
+      "loss": 1.1328,
+      "step": 7124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4327420084500086e-05,
+      "loss": 1.2188,
+      "step": 7125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.432599877782979e-05,
+      "loss": 1.3125,
+      "step": 7126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4324577363641656e-05,
+      "loss": 1.0547,
+      "step": 7127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4323155841971021e-05,
+      "loss": 1.2031,
+      "step": 7128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4321734212853212e-05,
+      "loss": 0.9766,
+      "step": 7129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4320312476323559e-05,
+      "loss": 1.4375,
+      "step": 7130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.43188906324174e-05,
+      "loss": 1.75,
+      "step": 7131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4317468681170075e-05,
+      "loss": 1.0156,
+      "step": 7132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4316046622616922e-05,
+      "loss": 1.2891,
+      "step": 7133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4314624456793285e-05,
+      "loss": 0.5742,
+      "step": 7134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4313202183734514e-05,
+      "loss": 0.9688,
+      "step": 7135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4311779803475953e-05,
+      "loss": 0.4219,
+      "step": 7136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4310357316052956e-05,
+      "loss": 0.5039,
+      "step": 7137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.430893472150088e-05,
+      "loss": 0.7383,
+      "step": 7138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4307512019855075e-05,
+      "loss": 1.0938,
+      "step": 7139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4306089211150909e-05,
+      "loss": 2.0312,
+      "step": 7140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4304666295423737e-05,
+      "loss": 1.5781,
+      "step": 7141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4303243272708928e-05,
+      "loss": 1.4609,
+      "step": 7142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.430182014304185e-05,
+      "loss": 0.75,
+      "step": 7143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4300396906457872e-05,
+      "loss": 1.0859,
+      "step": 7144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4298973562992367e-05,
+      "loss": 1.0625,
+      "step": 7145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.429755011268071e-05,
+      "loss": 0.7578,
+      "step": 7146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4296126555558281e-05,
+      "loss": 0.6172,
+      "step": 7147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4294702891660456e-05,
+      "loss": 0.8398,
+      "step": 7148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4293279121022628e-05,
+      "loss": 0.5469,
+      "step": 7149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4291855243680176e-05,
+      "loss": 0.9023,
+      "step": 7150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4290431259668486e-05,
+      "loss": 1.7031,
+      "step": 7151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4289007169022959e-05,
+      "loss": 1.0391,
+      "step": 7152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4287582971778984e-05,
+      "loss": 1.1562,
+      "step": 7153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4286158667971957e-05,
+      "loss": 1.2734,
+      "step": 7154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4284734257637279e-05,
+      "loss": 0.9805,
+      "step": 7155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4283309740810347e-05,
+      "loss": 0.4883,
+      "step": 7156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4281885117526577e-05,
+      "loss": 1.0312,
+      "step": 7157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4280460387821364e-05,
+      "loss": 0.498,
+      "step": 7158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4279035551730127e-05,
+      "loss": 1.1406,
+      "step": 7159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4277610609288274e-05,
+      "loss": 0.2461,
+      "step": 7160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4276185560531223e-05,
+      "loss": 0.5625,
+      "step": 7161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.427476040549439e-05,
+      "loss": 1.0547,
+      "step": 7162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4273335144213199e-05,
+      "loss": 0.4824,
+      "step": 7163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4271909776723068e-05,
+      "loss": 0.4355,
+      "step": 7164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4270484303059426e-05,
+      "loss": 1.25,
+      "step": 7165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4269058723257702e-05,
+      "loss": 0.4922,
+      "step": 7166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4267633037353325e-05,
+      "loss": 0.8594,
+      "step": 7167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4266207245381733e-05,
+      "loss": 0.3105,
+      "step": 7168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4264781347378356e-05,
+      "loss": 1.8203,
+      "step": 7169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4263355343378637e-05,
+      "loss": 1.0469,
+      "step": 7170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4261929233418022e-05,
+      "loss": 1.0781,
+      "step": 7171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4260503017531946e-05,
+      "loss": 0.4395,
+      "step": 7172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4259076695755863e-05,
+      "loss": 1.4375,
+      "step": 7173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4257650268125219e-05,
+      "loss": 0.918,
+      "step": 7174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4256223734675467e-05,
+      "loss": 0.543,
+      "step": 7175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4254797095442066e-05,
+      "loss": 0.707,
+      "step": 7176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4253370350460467e-05,
+      "loss": 1.125,
+      "step": 7177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4251943499766135e-05,
+      "loss": 0.8945,
+      "step": 7178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.425051654339453e-05,
+      "loss": 0.7344,
+      "step": 7179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4249089481381118e-05,
+      "loss": 1.2344,
+      "step": 7180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4247662313761368e-05,
+      "loss": 0.6445,
+      "step": 7181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.424623504057075e-05,
+      "loss": 1.5469,
+      "step": 7182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4244807661844737e-05,
+      "loss": 0.9102,
+      "step": 7183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4243380177618806e-05,
+      "loss": 0.9844,
+      "step": 7184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4241952587928434e-05,
+      "loss": 0.8242,
+      "step": 7185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4240524892809102e-05,
+      "loss": 1.0859,
+      "step": 7186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4239097092296297e-05,
+      "loss": 1.0938,
+      "step": 7187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.42376691864255e-05,
+      "loss": 1.0078,
+      "step": 7188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4236241175232204e-05,
+      "loss": 0.7617,
+      "step": 7189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.42348130587519e-05,
+      "loss": 1.1641,
+      "step": 7190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.423338483702008e-05,
+      "loss": 0.918,
+      "step": 7191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4231956510072246e-05,
+      "loss": 0.8906,
+      "step": 7192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4230528077943894e-05,
+      "loss": 0.9375,
+      "step": 7193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4229099540670523e-05,
+      "loss": 1.375,
+      "step": 7194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4227670898287645e-05,
+      "loss": 0.875,
+      "step": 7195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.422624215083076e-05,
+      "loss": 1.1719,
+      "step": 7196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4224813298335382e-05,
+      "loss": 1.3828,
+      "step": 7197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4223384340837023e-05,
+      "loss": 1.5781,
+      "step": 7198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4221955278371197e-05,
+      "loss": 0.5586,
+      "step": 7199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.422052611097342e-05,
+      "loss": 0.9844,
+      "step": 7200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.421909683867922e-05,
+      "loss": 1.9844,
+      "step": 7201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4217667461524112e-05,
+      "loss": 0.4785,
+      "step": 7202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4216237979543624e-05,
+      "loss": 0.4824,
+      "step": 7203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4214808392773284e-05,
+      "loss": 1.0156,
+      "step": 7204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4213378701248626e-05,
+      "loss": 0.9375,
+      "step": 7205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.421194890500518e-05,
+      "loss": 1.1016,
+      "step": 7206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.421051900407848e-05,
+      "loss": 0.8828,
+      "step": 7207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4209088998504071e-05,
+      "loss": 1.0,
+      "step": 7208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4207658888317485e-05,
+      "loss": 0.9414,
+      "step": 7209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4206228673554277e-05,
+      "loss": 1.2188,
+      "step": 7210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4204798354249985e-05,
+      "loss": 1.3906,
+      "step": 7211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.420336793044016e-05,
+      "loss": 1.4062,
+      "step": 7212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4201937402160355e-05,
+      "loss": 0.7617,
+      "step": 7213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.420050676944612e-05,
+      "loss": 0.5977,
+      "step": 7214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4199076032333022e-05,
+      "loss": 1.7344,
+      "step": 7215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4197645190856605e-05,
+      "loss": 0.8281,
+      "step": 7216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4196214245052447e-05,
+      "loss": 0.9766,
+      "step": 7217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4194783194956098e-05,
+      "loss": 1.0078,
+      "step": 7218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4193352040603137e-05,
+      "loss": 1.3281,
+      "step": 7219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4191920782029127e-05,
+      "loss": 0.1797,
+      "step": 7220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4190489419269643e-05,
+      "loss": 1.625,
+      "step": 7221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4189057952360256e-05,
+      "loss": 0.3926,
+      "step": 7222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.418762638133655e-05,
+      "loss": 0.6367,
+      "step": 7223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4186194706234103e-05,
+      "loss": 1.2656,
+      "step": 7224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4184762927088494e-05,
+      "loss": 0.6797,
+      "step": 7225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.418333104393531e-05,
+      "loss": 0.3105,
+      "step": 7226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4181899056810138e-05,
+      "loss": 0.8242,
+      "step": 7227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4180466965748572e-05,
+      "loss": 1.5234,
+      "step": 7228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4179034770786201e-05,
+      "loss": 2.25,
+      "step": 7229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4177602471958625e-05,
+      "loss": 0.668,
+      "step": 7230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4176170069301438e-05,
+      "loss": 0.8828,
+      "step": 7231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4174737562850242e-05,
+      "loss": 0.918,
+      "step": 7232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4173304952640639e-05,
+      "loss": 0.4883,
+      "step": 7233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4171872238708242e-05,
+      "loss": 1.1562,
+      "step": 7234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.417043942108865e-05,
+      "loss": 0.8672,
+      "step": 7235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4169006499817477e-05,
+      "loss": 1.3359,
+      "step": 7236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4167573474930341e-05,
+      "loss": 1.2031,
+      "step": 7237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4166140346462853e-05,
+      "loss": 0.8945,
+      "step": 7238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4164707114450634e-05,
+      "loss": 1.0547,
+      "step": 7239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4163273778929308e-05,
+      "loss": 0.6992,
+      "step": 7240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4161840339934491e-05,
+      "loss": 1.0078,
+      "step": 7241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4160406797501817e-05,
+      "loss": 1.5703,
+      "step": 7242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4158973151666915e-05,
+      "loss": 0.8164,
+      "step": 7243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4157539402465412e-05,
+      "loss": 1.5156,
+      "step": 7244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4156105549932946e-05,
+      "loss": 0.5156,
+      "step": 7245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4154671594105151e-05,
+      "loss": 0.9414,
+      "step": 7246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4153237535017668e-05,
+      "loss": 0.8008,
+      "step": 7247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4151803372706139e-05,
+      "loss": 0.5352,
+      "step": 7248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4150369107206205e-05,
+      "loss": 0.8867,
+      "step": 7249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4148934738553523e-05,
+      "loss": 0.8281,
+      "step": 7250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4147500266783731e-05,
+      "loss": 1.3516,
+      "step": 7251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4146065691932486e-05,
+      "loss": 1.0391,
+      "step": 7252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4144631014035444e-05,
+      "loss": 0.4277,
+      "step": 7253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4143196233128263e-05,
+      "loss": 0.4766,
+      "step": 7254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.41417613492466e-05,
+      "loss": 0.9805,
+      "step": 7255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4140326362426115e-05,
+      "loss": 0.5508,
+      "step": 7256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4138891272702479e-05,
+      "loss": 0.7773,
+      "step": 7257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4137456080111356e-05,
+      "loss": 0.5469,
+      "step": 7258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4136020784688418e-05,
+      "loss": 1.2422,
+      "step": 7259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4134585386469337e-05,
+      "loss": 0.5039,
+      "step": 7260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4133149885489782e-05,
+      "loss": 0.2402,
+      "step": 7261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4131714281785445e-05,
+      "loss": 0.625,
+      "step": 7262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4130278575391994e-05,
+      "loss": 0.459,
+      "step": 7263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4128842766345116e-05,
+      "loss": 0.75,
+      "step": 7264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4127406854680497e-05,
+      "loss": 0.8867,
+      "step": 7265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4125970840433823e-05,
+      "loss": 0.8203,
+      "step": 7266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4124534723640785e-05,
+      "loss": 0.8438,
+      "step": 7267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4123098504337078e-05,
+      "loss": 0.5859,
+      "step": 7268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4121662182558397e-05,
+      "loss": 0.6133,
+      "step": 7269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.412022575834044e-05,
+      "loss": 1.9219,
+      "step": 7270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4118789231718906e-05,
+      "loss": 0.7656,
+      "step": 7271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4117352602729499e-05,
+      "loss": 0.7148,
+      "step": 7272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4115915871407925e-05,
+      "loss": 2.1719,
+      "step": 7273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4114479037789895e-05,
+      "loss": 0.498,
+      "step": 7274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4113042101911115e-05,
+      "loss": 0.5742,
+      "step": 7275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4111605063807303e-05,
+      "loss": 0.4316,
+      "step": 7276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4110167923514172e-05,
+      "loss": 1.6016,
+      "step": 7277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4108730681067442e-05,
+      "loss": 1.4688,
+      "step": 7278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.410729333650283e-05,
+      "loss": 1.8594,
+      "step": 7279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4105855889856067e-05,
+      "loss": 0.6211,
+      "step": 7280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4104418341162873e-05,
+      "loss": 1.2266,
+      "step": 7281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.410298069045898e-05,
+      "loss": 0.2871,
+      "step": 7282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4101542937780117e-05,
+      "loss": 0.8242,
+      "step": 7283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4100105083162021e-05,
+      "loss": 1.3281,
+      "step": 7284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4098667126640424e-05,
+      "loss": 0.8945,
+      "step": 7285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4097229068251068e-05,
+      "loss": 0.4414,
+      "step": 7286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4095790908029691e-05,
+      "loss": 0.7852,
+      "step": 7287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4094352646012041e-05,
+      "loss": 1.3203,
+      "step": 7288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.409291428223386e-05,
+      "loss": 1.625,
+      "step": 7289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4091475816730901e-05,
+      "loss": 1.7656,
+      "step": 7290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4090037249538915e-05,
+      "loss": 1.4219,
+      "step": 7291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4088598580693656e-05,
+      "loss": 1.1328,
+      "step": 7292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4087159810230872e-05,
+      "loss": 0.7617,
+      "step": 7293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4085720938186333e-05,
+      "loss": 1.0469,
+      "step": 7294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4084281964595799e-05,
+      "loss": 1.0312,
+      "step": 7295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4082842889495029e-05,
+      "loss": 1.125,
+      "step": 7296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4081403712919795e-05,
+      "loss": 0.9648,
+      "step": 7297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4079964434905862e-05,
+      "loss": 0.9648,
+      "step": 7298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4078525055489002e-05,
+      "loss": 0.8906,
+      "step": 7299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4077085574704992e-05,
+      "loss": 1.2109,
+      "step": 7300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4075645992589606e-05,
+      "loss": 1.0,
+      "step": 7301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4074206309178626e-05,
+      "loss": 0.8477,
+      "step": 7302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4072766524507831e-05,
+      "loss": 0.7344,
+      "step": 7303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4071326638613006e-05,
+      "loss": 1.7812,
+      "step": 7304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4069886651529936e-05,
+      "loss": 1.5156,
+      "step": 7305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4068446563294413e-05,
+      "loss": 1.1094,
+      "step": 7306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.406700637394223e-05,
+      "loss": 0.5273,
+      "step": 7307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4065566083509175e-05,
+      "loss": 0.3438,
+      "step": 7308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4064125692031052e-05,
+      "loss": 0.5312,
+      "step": 7309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4062685199543655e-05,
+      "loss": 0.8047,
+      "step": 7310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4061244606082793e-05,
+      "loss": 0.8047,
+      "step": 7311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.405980391168426e-05,
+      "loss": 0.5039,
+      "step": 7312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4058363116383871e-05,
+      "loss": 0.9336,
+      "step": 7313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.405692222021743e-05,
+      "loss": 0.7188,
+      "step": 7314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4055481223220755e-05,
+      "loss": 1.8203,
+      "step": 7315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4054040125429656e-05,
+      "loss": 0.8125,
+      "step": 7316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.405259892687995e-05,
+      "loss": 0.5312,
+      "step": 7317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4051157627607457e-05,
+      "loss": 1.3516,
+      "step": 7318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4049716227648e-05,
+      "loss": 1.3906,
+      "step": 7319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4048274727037401e-05,
+      "loss": 1.9219,
+      "step": 7320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4046833125811487e-05,
+      "loss": 0.3906,
+      "step": 7321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4045391424006089e-05,
+      "loss": 0.3789,
+      "step": 7322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4043949621657043e-05,
+      "loss": 0.6133,
+      "step": 7323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4042507718800171e-05,
+      "loss": 0.8789,
+      "step": 7324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4041065715471325e-05,
+      "loss": 1.1797,
+      "step": 7325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4039623611706332e-05,
+      "loss": 0.8516,
+      "step": 7326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4038181407541043e-05,
+      "loss": 1.75,
+      "step": 7327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4036739103011294e-05,
+      "loss": 1.4688,
+      "step": 7328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4035296698152938e-05,
+      "loss": 0.5547,
+      "step": 7329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4033854193001824e-05,
+      "loss": 1.6641,
+      "step": 7330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4032411587593801e-05,
+      "loss": 1.4531,
+      "step": 7331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4030968881964726e-05,
+      "loss": 1.4219,
+      "step": 7332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4029526076150452e-05,
+      "loss": 1.7422,
+      "step": 7333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4028083170186841e-05,
+      "loss": 1.3984,
+      "step": 7334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4026640164109757e-05,
+      "loss": 0.9766,
+      "step": 7335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.402519705795506e-05,
+      "loss": 0.3301,
+      "step": 7336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4023753851758621e-05,
+      "loss": 1.7578,
+      "step": 7337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4022310545556303e-05,
+      "loss": 1.1094,
+      "step": 7338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4020867139383984e-05,
+      "loss": 0.6133,
+      "step": 7339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4019423633277535e-05,
+      "loss": 1.0547,
+      "step": 7340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4017980027272835e-05,
+      "loss": 0.6445,
+      "step": 7341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.401653632140576e-05,
+      "loss": 0.6562,
+      "step": 7342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4015092515712197e-05,
+      "loss": 0.793,
+      "step": 7343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4013648610228023e-05,
+      "loss": 0.9609,
+      "step": 7344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.401220460498913e-05,
+      "loss": 0.8828,
+      "step": 7345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4010760500031406e-05,
+      "loss": 0.8281,
+      "step": 7346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4009316295390743e-05,
+      "loss": 1.1562,
+      "step": 7347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4007871991103032e-05,
+      "loss": 0.3965,
+      "step": 7348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4006427587204172e-05,
+      "loss": 1.3359,
+      "step": 7349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4004983083730061e-05,
+      "loss": 0.8203,
+      "step": 7350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4003538480716603e-05,
+      "loss": 0.6367,
+      "step": 7351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.40020937781997e-05,
+      "loss": 0.3926,
+      "step": 7352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4000648976215259e-05,
+      "loss": 1.0391,
+      "step": 7353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3999204074799186e-05,
+      "loss": 0.9688,
+      "step": 7354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3997759073987397e-05,
+      "loss": 0.4473,
+      "step": 7355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3996313973815803e-05,
+      "loss": 0.6953,
+      "step": 7356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3994868774320323e-05,
+      "loss": 0.25,
+      "step": 7357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3993423475536872e-05,
+      "loss": 1.2734,
+      "step": 7358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3991978077501376e-05,
+      "loss": 0.9844,
+      "step": 7359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3990532580249755e-05,
+      "loss": 0.2773,
+      "step": 7360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3989086983817934e-05,
+      "loss": 0.8516,
+      "step": 7361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3987641288241846e-05,
+      "loss": 1.0391,
+      "step": 7362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.398619549355742e-05,
+      "loss": 1.8594,
+      "step": 7363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3984749599800589e-05,
+      "loss": 1.6797,
+      "step": 7364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.398330360700729e-05,
+      "loss": 1.9297,
+      "step": 7365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.398185751521346e-05,
+      "loss": 1.625,
+      "step": 7366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3980411324455045e-05,
+      "loss": 0.1924,
+      "step": 7367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3978965034767979e-05,
+      "loss": 0.7227,
+      "step": 7368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3977518646188222e-05,
+      "loss": 1.0703,
+      "step": 7369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3976072158751706e-05,
+      "loss": 1.4766,
+      "step": 7370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3974625572494394e-05,
+      "loss": 0.8906,
+      "step": 7371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3973178887452233e-05,
+      "loss": 1.1328,
+      "step": 7372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3971732103661183e-05,
+      "loss": 1.8516,
+      "step": 7373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3970285221157199e-05,
+      "loss": 1.2344,
+      "step": 7374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.396883823997624e-05,
+      "loss": 1.1875,
+      "step": 7375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3967391160154274e-05,
+      "loss": 1.3359,
+      "step": 7376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3965943981727263e-05,
+      "loss": 0.9062,
+      "step": 7377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3964496704731179e-05,
+      "loss": 0.8516,
+      "step": 7378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3963049329201988e-05,
+      "loss": 1.5,
+      "step": 7379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3961601855175664e-05,
+      "loss": 0.8945,
+      "step": 7380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3960154282688183e-05,
+      "loss": 1.2812,
+      "step": 7381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3958706611775523e-05,
+      "loss": 1.2969,
+      "step": 7382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3957258842473665e-05,
+      "loss": 0.9688,
+      "step": 7383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.395581097481859e-05,
+      "loss": 1.2031,
+      "step": 7384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3954363008846282e-05,
+      "loss": 1.5,
+      "step": 7385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3952914944592733e-05,
+      "loss": 0.9102,
+      "step": 7386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3951466782093927e-05,
+      "loss": 0.9531,
+      "step": 7387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3950018521385862e-05,
+      "loss": 0.625,
+      "step": 7388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3948570162504528e-05,
+      "loss": 0.3594,
+      "step": 7389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3947121705485929e-05,
+      "loss": 1.1875,
+      "step": 7390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3945673150366057e-05,
+      "loss": 1.75,
+      "step": 7391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.394422449718092e-05,
+      "loss": 0.4082,
+      "step": 7392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.394277574596652e-05,
+      "loss": 0.9961,
+      "step": 7393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3941326896758866e-05,
+      "loss": 1.2422,
+      "step": 7394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3939877949593965e-05,
+      "loss": 1.1875,
+      "step": 7395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3938428904507833e-05,
+      "loss": 0.8555,
+      "step": 7396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3936979761536477e-05,
+      "loss": 0.9062,
+      "step": 7397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3935530520715921e-05,
+      "loss": 0.7461,
+      "step": 7398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3934081182082177e-05,
+      "loss": 1.1406,
+      "step": 7399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3932631745671278e-05,
+      "loss": 1.2188,
+      "step": 7400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3931182211519237e-05,
+      "loss": 0.625,
+      "step": 7401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3929732579662087e-05,
+      "loss": 0.3457,
+      "step": 7402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3928282850135852e-05,
+      "loss": 0.3203,
+      "step": 7403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3926833022976569e-05,
+      "loss": 0.5625,
+      "step": 7404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.392538309822027e-05,
+      "loss": 0.4961,
+      "step": 7405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.392393307590299e-05,
+      "loss": 0.3477,
+      "step": 7406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3922482956060765e-05,
+      "loss": 0.5469,
+      "step": 7407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3921032738729641e-05,
+      "loss": 0.8086,
+      "step": 7408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.391958242394566e-05,
+      "loss": 1.4609,
+      "step": 7409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3918132011744866e-05,
+      "loss": 0.3008,
+      "step": 7410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3916681502163311e-05,
+      "loss": 1.0156,
+      "step": 7411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3915230895237044e-05,
+      "loss": 0.9258,
+      "step": 7412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3913780191002114e-05,
+      "loss": 1.0156,
+      "step": 7413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3912329389494585e-05,
+      "loss": 1.0781,
+      "step": 7414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3910878490750507e-05,
+      "loss": 1.3516,
+      "step": 7415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.390942749480595e-05,
+      "loss": 0.4277,
+      "step": 7416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3907976401696963e-05,
+      "loss": 0.4512,
+      "step": 7417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3906525211459622e-05,
+      "loss": 1.6484,
+      "step": 7418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3905073924129994e-05,
+      "loss": 1.4141,
+      "step": 7419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3903622539744146e-05,
+      "loss": 1.3281,
+      "step": 7420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3902171058338154e-05,
+      "loss": 0.8086,
+      "step": 7421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3900719479948088e-05,
+      "loss": 1.6484,
+      "step": 7422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3899267804610026e-05,
+      "loss": 0.8125,
+      "step": 7423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3897816032360052e-05,
+      "loss": 1.0078,
+      "step": 7424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3896364163234246e-05,
+      "loss": 0.7383,
+      "step": 7425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3894912197268692e-05,
+      "loss": 0.832,
+      "step": 7426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3893460134499475e-05,
+      "loss": 1.4375,
+      "step": 7427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3892007974962689e-05,
+      "loss": 0.6914,
+      "step": 7428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3890555718694422e-05,
+      "loss": 0.9297,
+      "step": 7429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3889103365730773e-05,
+      "loss": 1.1094,
+      "step": 7430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3887650916107831e-05,
+      "loss": 1.6797,
+      "step": 7431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.38861983698617e-05,
+      "loss": 1.2031,
+      "step": 7432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3884745727028483e-05,
+      "loss": 0.6875,
+      "step": 7433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3883292987644282e-05,
+      "loss": 0.6445,
+      "step": 7434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3881840151745201e-05,
+      "loss": 0.9805,
+      "step": 7435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.388038721936735e-05,
+      "loss": 1.1484,
+      "step": 7436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3878934190546843e-05,
+      "loss": 0.7773,
+      "step": 7437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3877481065319786e-05,
+      "loss": 0.5195,
+      "step": 7438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3876027843722302e-05,
+      "loss": 1.0312,
+      "step": 7439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3874574525790508e-05,
+      "loss": 1.1875,
+      "step": 7440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.387312111156052e-05,
+      "loss": 1.0547,
+      "step": 7441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3871667601068468e-05,
+      "loss": 0.957,
+      "step": 7442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.387021399435047e-05,
+      "loss": 1.1562,
+      "step": 7443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3868760291442662e-05,
+      "loss": 1.3047,
+      "step": 7444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3867306492381164e-05,
+      "loss": 1.0312,
+      "step": 7445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3865852597202117e-05,
+      "loss": 1.0781,
+      "step": 7446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3864398605941651e-05,
+      "loss": 1.0703,
+      "step": 7447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3862944518635908e-05,
+      "loss": 1.0391,
+      "step": 7448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3861490335321023e-05,
+      "loss": 0.8359,
+      "step": 7449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3860036056033142e-05,
+      "loss": 1.1641,
+      "step": 7450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3858581680808408e-05,
+      "loss": 0.9258,
+      "step": 7451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3857127209682966e-05,
+      "loss": 1.3203,
+      "step": 7452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3855672642692966e-05,
+      "loss": 1.0312,
+      "step": 7453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.385421797987456e-05,
+      "loss": 0.707,
+      "step": 7454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3852763221263902e-05,
+      "loss": 0.3926,
+      "step": 7455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3851308366897152e-05,
+      "loss": 0.8477,
+      "step": 7456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3849853416810464e-05,
+      "loss": 1.25,
+      "step": 7457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.384839837104e-05,
+      "loss": 0.7852,
+      "step": 7458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3846943229621923e-05,
+      "loss": 0.9453,
+      "step": 7459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3845487992592399e-05,
+      "loss": 0.543,
+      "step": 7460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3844032659987602e-05,
+      "loss": 0.6836,
+      "step": 7461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3842577231843693e-05,
+      "loss": 0.8789,
+      "step": 7462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3841121708196857e-05,
+      "loss": 1.7422,
+      "step": 7463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3839666089083254e-05,
+      "loss": 0.5039,
+      "step": 7464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3838210374539073e-05,
+      "loss": 1.1016,
+      "step": 7465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3836754564600494e-05,
+      "loss": 0.8242,
+      "step": 7466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3835298659303696e-05,
+      "loss": 0.4961,
+      "step": 7467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3833842658684864e-05,
+      "loss": 1.5234,
+      "step": 7468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3832386562780184e-05,
+      "loss": 1.2109,
+      "step": 7469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.383093037162585e-05,
+      "loss": 0.7656,
+      "step": 7470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.382947408525805e-05,
+      "loss": 0.8828,
+      "step": 7471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3828017703712982e-05,
+      "loss": 1.7969,
+      "step": 7472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.382656122702684e-05,
+      "loss": 1.5,
+      "step": 7473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3825104655235823e-05,
+      "loss": 1.9062,
+      "step": 7474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3823647988376134e-05,
+      "loss": 1.2891,
+      "step": 7475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3822191226483974e-05,
+      "loss": 0.4648,
+      "step": 7476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3820734369595555e-05,
+      "loss": 0.7031,
+      "step": 7477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3819277417747079e-05,
+      "loss": 0.9883,
+      "step": 7478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3817820370974759e-05,
+      "loss": 1.25,
+      "step": 7479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3816363229314812e-05,
+      "loss": 1.1016,
+      "step": 7480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3814905992803449e-05,
+      "loss": 0.6836,
+      "step": 7481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3813448661476889e-05,
+      "loss": 0.6172,
+      "step": 7482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3811991235371353e-05,
+      "loss": 0.5547,
+      "step": 7483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3810533714523065e-05,
+      "loss": 0.8867,
+      "step": 7484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3809076098968247e-05,
+      "loss": 1.4609,
+      "step": 7485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3807618388743127e-05,
+      "loss": 1.3047,
+      "step": 7486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3806160583883936e-05,
+      "loss": 1.2266,
+      "step": 7487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3804702684426906e-05,
+      "loss": 1.5078,
+      "step": 7488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3803244690408275e-05,
+      "loss": 1.3672,
+      "step": 7489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.380178660186427e-05,
+      "loss": 1.1328,
+      "step": 7490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3800328418831144e-05,
+      "loss": 0.5547,
+      "step": 7491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3798870141345122e-05,
+      "loss": 1.1953,
+      "step": 7492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3797411769442462e-05,
+      "loss": 0.8516,
+      "step": 7493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3795953303159402e-05,
+      "loss": 0.9219,
+      "step": 7494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3794494742532195e-05,
+      "loss": 1.25,
+      "step": 7495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3793036087597091e-05,
+      "loss": 0.7227,
+      "step": 7496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.379157733839034e-05,
+      "loss": 0.9258,
+      "step": 7497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3790118494948202e-05,
+      "loss": 0.9023,
+      "step": 7498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3788659557306932e-05,
+      "loss": 0.6992,
+      "step": 7499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.378720052550279e-05,
+      "loss": 0.9141,
+      "step": 7500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3785741399572041e-05,
+      "loss": 1.7812,
+      "step": 7501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.378428217955095e-05,
+      "loss": 1.0078,
+      "step": 7502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3782822865475782e-05,
+      "loss": 0.4062,
+      "step": 7503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3781363457382804e-05,
+      "loss": 1.7578,
+      "step": 7504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3779903955308298e-05,
+      "loss": 0.8125,
+      "step": 7505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3778444359288525e-05,
+      "loss": 1.1562,
+      "step": 7506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3776984669359774e-05,
+      "loss": 1.2578,
+      "step": 7507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3775524885558316e-05,
+      "loss": 0.498,
+      "step": 7508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3774065007920434e-05,
+      "loss": 2.1875,
+      "step": 7509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3772605036482412e-05,
+      "loss": 0.2344,
+      "step": 7510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3771144971280537e-05,
+      "loss": 0.8125,
+      "step": 7511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3769684812351098e-05,
+      "loss": 0.6484,
+      "step": 7512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3768224559730383e-05,
+      "loss": 0.7422,
+      "step": 7513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3766764213454686e-05,
+      "loss": 0.5391,
+      "step": 7514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3765303773560301e-05,
+      "loss": 0.9375,
+      "step": 7515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3763843240083528e-05,
+      "loss": 1.125,
+      "step": 7516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3762382613060665e-05,
+      "loss": 1.0859,
+      "step": 7517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3760921892528015e-05,
+      "loss": 0.7188,
+      "step": 7518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3759461078521885e-05,
+      "loss": 1.2812,
+      "step": 7519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3758000171078576e-05,
+      "loss": 1.1875,
+      "step": 7520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3756539170234404e-05,
+      "loss": 1.0156,
+      "step": 7521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3755078076025674e-05,
+      "loss": 1.0859,
+      "step": 7522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3753616888488703e-05,
+      "loss": 1.2578,
+      "step": 7523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3752155607659812e-05,
+      "loss": 1.0234,
+      "step": 7524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.375069423357531e-05,
+      "loss": 1.1875,
+      "step": 7525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3749232766271527e-05,
+      "loss": 1.2812,
+      "step": 7526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3747771205784777e-05,
+      "loss": 0.6133,
+      "step": 7527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3746309552151392e-05,
+      "loss": 0.4219,
+      "step": 7528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3744847805407698e-05,
+      "loss": 0.6055,
+      "step": 7529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3743385965590023e-05,
+      "loss": 0.1797,
+      "step": 7530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3741924032734707e-05,
+      "loss": 1.0,
+      "step": 7531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3740462006878073e-05,
+      "loss": 1.1562,
+      "step": 7532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3738999888056466e-05,
+      "loss": 1.0156,
+      "step": 7533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3737537676306223e-05,
+      "loss": 1.3125,
+      "step": 7534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3736075371663687e-05,
+      "loss": 0.875,
+      "step": 7535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.37346129741652e-05,
+      "loss": 0.5078,
+      "step": 7536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.373315048384711e-05,
+      "loss": 1.0938,
+      "step": 7537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3731687900745765e-05,
+      "loss": 1.125,
+      "step": 7538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3730225224897513e-05,
+      "loss": 1.1641,
+      "step": 7539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3728762456338713e-05,
+      "loss": 1.3047,
+      "step": 7540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3727299595105715e-05,
+      "loss": 0.6875,
+      "step": 7541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.372583664123488e-05,
+      "loss": 1.3438,
+      "step": 7542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3724373594762566e-05,
+      "loss": 0.9336,
+      "step": 7543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3722910455725136e-05,
+      "loss": 0.8164,
+      "step": 7544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3721447224158957e-05,
+      "loss": 0.7969,
+      "step": 7545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3719983900100392e-05,
+      "loss": 1.5938,
+      "step": 7546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3718520483585815e-05,
+      "loss": 0.8359,
+      "step": 7547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3717056974651594e-05,
+      "loss": 0.8633,
+      "step": 7548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3715593373334104e-05,
+      "loss": 0.8281,
+      "step": 7549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3714129679669719e-05,
+      "loss": 0.6367,
+      "step": 7550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.371266589369482e-05,
+      "loss": 1.5547,
+      "step": 7551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3711202015445789e-05,
+      "loss": 1.0938,
+      "step": 7552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3709738044959005e-05,
+      "loss": 1.3438,
+      "step": 7553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.370827398227086e-05,
+      "loss": 0.793,
+      "step": 7554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3706809827417731e-05,
+      "loss": 0.4316,
+      "step": 7555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3705345580436016e-05,
+      "loss": 0.1309,
+      "step": 7556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3703881241362105e-05,
+      "loss": 1.4297,
+      "step": 7557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3702416810232395e-05,
+      "loss": 0.9961,
+      "step": 7558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.370095228708328e-05,
+      "loss": 0.5195,
+      "step": 7559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.369948767195116e-05,
+      "loss": 1.0469,
+      "step": 7560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3698022964872434e-05,
+      "loss": 0.9062,
+      "step": 7561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3696558165883509e-05,
+      "loss": 0.8711,
+      "step": 7562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3695093275020788e-05,
+      "loss": 0.2383,
+      "step": 7563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3693628292320683e-05,
+      "loss": 1.7734,
+      "step": 7564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.36921632178196e-05,
+      "loss": 0.9102,
+      "step": 7565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.369069805155396e-05,
+      "loss": 1.7188,
+      "step": 7566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3689232793560162e-05,
+      "loss": 1.0078,
+      "step": 7567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3687767443874642e-05,
+      "loss": 0.6094,
+      "step": 7568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3686302002533806e-05,
+      "loss": 0.9531,
+      "step": 7569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3684836469574085e-05,
+      "loss": 1.8906,
+      "step": 7570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3683370845031898e-05,
+      "loss": 0.6172,
+      "step": 7571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.368190512894367e-05,
+      "loss": 1.2031,
+      "step": 7572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3680439321345835e-05,
+      "loss": 0.8945,
+      "step": 7573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3678973422274822e-05,
+      "loss": 1.3594,
+      "step": 7574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3677507431767062e-05,
+      "loss": 1.4141,
+      "step": 7575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3676041349858993e-05,
+      "loss": 1.5156,
+      "step": 7576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3674575176587053e-05,
+      "loss": 1.3359,
+      "step": 7577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.367310891198768e-05,
+      "loss": 0.5273,
+      "step": 7578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3671642556097316e-05,
+      "loss": 1.4219,
+      "step": 7579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3670176108952409e-05,
+      "loss": 0.3633,
+      "step": 7580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3668709570589401e-05,
+      "loss": 0.832,
+      "step": 7581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3667242941044748e-05,
+      "loss": 0.8984,
+      "step": 7582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3665776220354894e-05,
+      "loss": 0.8398,
+      "step": 7583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3664309408556298e-05,
+      "loss": 1.0781,
+      "step": 7584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3662842505685414e-05,
+      "loss": 0.4805,
+      "step": 7585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.36613755117787e-05,
+      "loss": 0.377,
+      "step": 7586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365990842687262e-05,
+      "loss": 0.6328,
+      "step": 7587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365844125100363e-05,
+      "loss": 0.4297,
+      "step": 7588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.36569739842082e-05,
+      "loss": 1.0703,
+      "step": 7589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3655506626522796e-05,
+      "loss": 0.9961,
+      "step": 7590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3654039177983887e-05,
+      "loss": 0.7344,
+      "step": 7591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3652571638627946e-05,
+      "loss": 1.4375,
+      "step": 7592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3651104008491444e-05,
+      "loss": 0.8711,
+      "step": 7593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3649636287610861e-05,
+      "loss": 1.8906,
+      "step": 7594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3648168476022674e-05,
+      "loss": 0.875,
+      "step": 7595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3646700573763365e-05,
+      "loss": 0.9102,
+      "step": 7596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3645232580869414e-05,
+      "loss": 0.5586,
+      "step": 7597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3643764497377307e-05,
+      "loss": 0.9648,
+      "step": 7598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3642296323323539e-05,
+      "loss": 0.9609,
+      "step": 7599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3640828058744586e-05,
+      "loss": 0.6875,
+      "step": 7600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3639359703676956e-05,
+      "loss": 0.7812,
+      "step": 7601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3637891258157127e-05,
+      "loss": 0.7773,
+      "step": 7602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3636422722221608e-05,
+      "loss": 1.2422,
+      "step": 7603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3634954095906894e-05,
+      "loss": 1.4297,
+      "step": 7604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3633485379249483e-05,
+      "loss": 1.1797,
+      "step": 7605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3632016572285883e-05,
+      "loss": 0.6875,
+      "step": 7606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3630547675052595e-05,
+      "loss": 0.4707,
+      "step": 7607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.362907868758613e-05,
+      "loss": 2.0156,
+      "step": 7608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3627609609922998e-05,
+      "loss": 1.1016,
+      "step": 7609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3626140442099708e-05,
+      "loss": 1.0391,
+      "step": 7610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.362467118415278e-05,
+      "loss": 0.7695,
+      "step": 7611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3623201836118724e-05,
+      "loss": 1.8906,
+      "step": 7612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3621732398034066e-05,
+      "loss": 0.8438,
+      "step": 7613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.362026286993532e-05,
+      "loss": 1.6562,
+      "step": 7614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.361879325185902e-05,
+      "loss": 1.1484,
+      "step": 7615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3617323543841678e-05,
+      "loss": 1.4062,
+      "step": 7616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3615853745919832e-05,
+      "loss": 1.1484,
+      "step": 7617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3614383858130009e-05,
+      "loss": 1.0625,
+      "step": 7618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3612913880508742e-05,
+      "loss": 1.2344,
+      "step": 7619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3611443813092567e-05,
+      "loss": 1.6719,
+      "step": 7620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3609973655918016e-05,
+      "loss": 1.6172,
+      "step": 7621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3608503409021631e-05,
+      "loss": 1.3125,
+      "step": 7622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3607033072439957e-05,
+      "loss": 1.0859,
+      "step": 7623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3605562646209531e-05,
+      "loss": 0.418,
+      "step": 7624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3604092130366904e-05,
+      "loss": 0.252,
+      "step": 7625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.360262152494862e-05,
+      "loss": 0.5078,
+      "step": 7626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3601150829991232e-05,
+      "loss": 0.3809,
+      "step": 7627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3599680045531289e-05,
+      "loss": 1.6641,
+      "step": 7628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3598209171605352e-05,
+      "loss": 1.1875,
+      "step": 7629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3596738208249972e-05,
+      "loss": 0.8008,
+      "step": 7630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.359526715550171e-05,
+      "loss": 0.6562,
+      "step": 7631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.359379601339713e-05,
+      "loss": 0.5469,
+      "step": 7632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3592324781972791e-05,
+      "loss": 0.6797,
+      "step": 7633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3590853461265265e-05,
+      "loss": 1.3203,
+      "step": 7634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3589382051311113e-05,
+      "loss": 1.0703,
+      "step": 7635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3587910552146909e-05,
+      "loss": 0.5078,
+      "step": 7636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3586438963809222e-05,
+      "loss": 1.3203,
+      "step": 7637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3584967286334634e-05,
+      "loss": 1.2734,
+      "step": 7638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3583495519759716e-05,
+      "loss": 1.125,
+      "step": 7639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3582023664121047e-05,
+      "loss": 0.7539,
+      "step": 7640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3580551719455212e-05,
+      "loss": 1.2812,
+      "step": 7641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3579079685798789e-05,
+      "loss": 1.0781,
+      "step": 7642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.357760756318837e-05,
+      "loss": 0.7031,
+      "step": 7643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3576135351660537e-05,
+      "loss": 1.3906,
+      "step": 7644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3574663051251888e-05,
+      "loss": 1.0078,
+      "step": 7645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3573190661999007e-05,
+      "loss": 1.2109,
+      "step": 7646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3571718183938492e-05,
+      "loss": 0.9531,
+      "step": 7647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.357024561710694e-05,
+      "loss": 1.0312,
+      "step": 7648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3568772961540953e-05,
+      "loss": 0.7383,
+      "step": 7649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3567300217277125e-05,
+      "loss": 1.1953,
+      "step": 7650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3565827384352066e-05,
+      "loss": 1.0547,
+      "step": 7651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3564354462802378e-05,
+      "loss": 1.3125,
+      "step": 7652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3562881452664671e-05,
+      "loss": 0.8984,
+      "step": 7653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3561408353975554e-05,
+      "loss": 1.5938,
+      "step": 7654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3559935166771635e-05,
+      "loss": 0.6445,
+      "step": 7655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3558461891089535e-05,
+      "loss": 0.7422,
+      "step": 7656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3556988526965873e-05,
+      "loss": 0.8047,
+      "step": 7657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3555515074437256e-05,
+      "loss": 0.1963,
+      "step": 7658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3554041533540319e-05,
+      "loss": 0.375,
+      "step": 7659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3552567904311671e-05,
+      "loss": 1.5547,
+      "step": 7660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3551094186787948e-05,
+      "loss": 1.8828,
+      "step": 7661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3549620381005772e-05,
+      "loss": 0.5312,
+      "step": 7662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3548146487001778e-05,
+      "loss": 0.2988,
+      "step": 7663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3546672504812594e-05,
+      "loss": 0.6484,
+      "step": 7664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3545198434474854e-05,
+      "loss": 0.5586,
+      "step": 7665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3543724276025196e-05,
+      "loss": 0.5195,
+      "step": 7666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.354225002950026e-05,
+      "loss": 0.7109,
+      "step": 7667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3540775694936681e-05,
+      "loss": 0.7227,
+      "step": 7668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3539301272371109e-05,
+      "loss": 0.3457,
+      "step": 7669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.353782676184018e-05,
+      "loss": 1.0156,
+      "step": 7670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3536352163380555e-05,
+      "loss": 1.2734,
+      "step": 7671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.353487747702887e-05,
+      "loss": 0.3809,
+      "step": 7672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3533402702821785e-05,
+      "loss": 1.1562,
+      "step": 7673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3531927840795951e-05,
+      "loss": 1.4766,
+      "step": 7674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3530452890988022e-05,
+      "loss": 1.4219,
+      "step": 7675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3528977853434662e-05,
+      "loss": 1.6641,
+      "step": 7676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3527502728172523e-05,
+      "loss": 0.8867,
+      "step": 7677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3526027515238276e-05,
+      "loss": 0.6719,
+      "step": 7678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3524552214668579e-05,
+      "loss": 0.8008,
+      "step": 7679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3523076826500103e-05,
+      "loss": 0.8945,
+      "step": 7680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3521601350769517e-05,
+      "loss": 1.1172,
+      "step": 7681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3520125787513492e-05,
+      "loss": 1.1875,
+      "step": 7682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.35186501367687e-05,
+      "loss": 1.5234,
+      "step": 7683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3517174398571818e-05,
+      "loss": 0.9766,
+      "step": 7684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3515698572959524e-05,
+      "loss": 0.5039,
+      "step": 7685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3514222659968494e-05,
+      "loss": 0.2139,
+      "step": 7686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3512746659635418e-05,
+      "loss": 1.0859,
+      "step": 7687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3511270571996972e-05,
+      "loss": 1.0312,
+      "step": 7688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3509794397089846e-05,
+      "loss": 0.8398,
+      "step": 7689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3508318134950732e-05,
+      "loss": 0.5234,
+      "step": 7690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3506841785616316e-05,
+      "loss": 1.4609,
+      "step": 7691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3505365349123297e-05,
+      "loss": 0.6992,
+      "step": 7692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3503888825508358e-05,
+      "loss": 1.7266,
+      "step": 7693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3502412214808208e-05,
+      "loss": 0.8477,
+      "step": 7694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3500935517059545e-05,
+      "loss": 1.0312,
+      "step": 7695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3499458732299067e-05,
+      "loss": 0.668,
+      "step": 7696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.349798186056348e-05,
+      "loss": 1.3828,
+      "step": 7697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3496504901889487e-05,
+      "loss": 1.1094,
+      "step": 7698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.34950278563138e-05,
+      "loss": 0.832,
+      "step": 7699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3493550723873127e-05,
+      "loss": 0.793,
+      "step": 7700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3492073504604185e-05,
+      "loss": 1.7344,
+      "step": 7701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3490596198543679e-05,
+      "loss": 1.5391,
+      "step": 7702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3489118805728331e-05,
+      "loss": 1.3281,
+      "step": 7703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3487641326194866e-05,
+      "loss": 1.2578,
+      "step": 7704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3486163759979997e-05,
+      "loss": 0.6016,
+      "step": 7705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3484686107120452e-05,
+      "loss": 1.6328,
+      "step": 7706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3483208367652952e-05,
+      "loss": 1.1016,
+      "step": 7707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3481730541614226e-05,
+      "loss": 0.3203,
+      "step": 7708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3480252629041006e-05,
+      "loss": 0.6641,
+      "step": 7709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3478774629970022e-05,
+      "loss": 0.4922,
+      "step": 7710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3477296544438007e-05,
+      "loss": 0.4375,
+      "step": 7711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.34758183724817e-05,
+      "loss": 0.2256,
+      "step": 7712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3474340114137839e-05,
+      "loss": 0.2139,
+      "step": 7713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3472861769443162e-05,
+      "loss": 1.1719,
+      "step": 7714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3471383338434413e-05,
+      "loss": 1.0156,
+      "step": 7715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3469904821148336e-05,
+      "loss": 0.8164,
+      "step": 7716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3468426217621678e-05,
+      "loss": 1.2109,
+      "step": 7717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.346694752789119e-05,
+      "loss": 0.6289,
+      "step": 7718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.346546875199362e-05,
+      "loss": 1.3047,
+      "step": 7719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3463989889965725e-05,
+      "loss": 1.0078,
+      "step": 7720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3462510941844256e-05,
+      "loss": 0.7422,
+      "step": 7721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3461031907665976e-05,
+      "loss": 1.5156,
+      "step": 7722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.345955278746764e-05,
+      "loss": 0.6914,
+      "step": 7723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3458073581286013e-05,
+      "loss": 1.3203,
+      "step": 7724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3456594289157856e-05,
+      "loss": 0.4512,
+      "step": 7725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3455114911119937e-05,
+      "loss": 0.6367,
+      "step": 7726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3453635447209025e-05,
+      "loss": 0.6094,
+      "step": 7727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3452155897461889e-05,
+      "loss": 0.8125,
+      "step": 7728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.34506762619153e-05,
+      "loss": 0.2617,
+      "step": 7729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.344919654060604e-05,
+      "loss": 1.1094,
+      "step": 7730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3447716733570877e-05,
+      "loss": 1.2969,
+      "step": 7731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3446236840846592e-05,
+      "loss": 1.0312,
+      "step": 7732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3444756862469967e-05,
+      "loss": 1.2969,
+      "step": 7733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3443276798477791e-05,
+      "loss": 0.7305,
+      "step": 7734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.344179664890684e-05,
+      "loss": 0.6133,
+      "step": 7735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3440316413793905e-05,
+      "loss": 1.0703,
+      "step": 7736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3438836093175778e-05,
+      "loss": 1.2109,
+      "step": 7737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3437355687089249e-05,
+      "loss": 0.2285,
+      "step": 7738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3435875195571111e-05,
+      "loss": 0.457,
+      "step": 7739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.343439461865816e-05,
+      "loss": 0.8633,
+      "step": 7740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3432913956387195e-05,
+      "loss": 1.0469,
+      "step": 7741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3431433208795016e-05,
+      "loss": 0.5586,
+      "step": 7742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3429952375918425e-05,
+      "loss": 0.4609,
+      "step": 7743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3428471457794226e-05,
+      "loss": 1.2344,
+      "step": 7744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3426990454459227e-05,
+      "loss": 1.6094,
+      "step": 7745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3425509365950236e-05,
+      "loss": 1.2344,
+      "step": 7746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3424028192304062e-05,
+      "loss": 0.9141,
+      "step": 7747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.342254693355752e-05,
+      "loss": 0.5625,
+      "step": 7748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3421065589747427e-05,
+      "loss": 0.625,
+      "step": 7749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3419584160910592e-05,
+      "loss": 1.0859,
+      "step": 7750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3418102647083846e-05,
+      "loss": 1.3906,
+      "step": 7751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3416621048303999e-05,
+      "loss": 0.8828,
+      "step": 7752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3415139364607885e-05,
+      "loss": 0.3594,
+      "step": 7753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3413657596032322e-05,
+      "loss": 0.5586,
+      "step": 7754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3412175742614137e-05,
+      "loss": 0.5547,
+      "step": 7755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3410693804390167e-05,
+      "loss": 0.3223,
+      "step": 7756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3409211781397237e-05,
+      "loss": 0.4082,
+      "step": 7757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3407729673672187e-05,
+      "loss": 0.793,
+      "step": 7758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3406247481251846e-05,
+      "loss": 0.625,
+      "step": 7759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.340476520417306e-05,
+      "loss": 0.4629,
+      "step": 7760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3403282842472662e-05,
+      "loss": 1.5547,
+      "step": 7761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3401800396187499e-05,
+      "loss": 1.1641,
+      "step": 7762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3400317865354415e-05,
+      "loss": 2.125,
+      "step": 7763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3398835250010255e-05,
+      "loss": 1.5312,
+      "step": 7764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3397352550191869e-05,
+      "loss": 0.4473,
+      "step": 7765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3395869765936107e-05,
+      "loss": 0.7422,
+      "step": 7766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3394386897279826e-05,
+      "loss": 1.0078,
+      "step": 7767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3392903944259871e-05,
+      "loss": 0.75,
+      "step": 7768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3391420906913108e-05,
+      "loss": 0.9258,
+      "step": 7769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3389937785276397e-05,
+      "loss": 0.8633,
+      "step": 7770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3388454579386594e-05,
+      "loss": 0.4199,
+      "step": 7771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3386971289280566e-05,
+      "loss": 0.457,
+      "step": 7772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3385487914995175e-05,
+      "loss": 0.6523,
+      "step": 7773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3384004456567293e-05,
+      "loss": 0.4863,
+      "step": 7774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3382520914033786e-05,
+      "loss": 1.2266,
+      "step": 7775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3381037287431526e-05,
+      "loss": 1.6875,
+      "step": 7776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.337955357679739e-05,
+      "loss": 0.7578,
+      "step": 7777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3378069782168252e-05,
+      "loss": 1.1484,
+      "step": 7778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.337658590358099e-05,
+      "loss": 1.2969,
+      "step": 7779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.337510194107248e-05,
+      "loss": 1.3984,
+      "step": 7780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3373617894679616e-05,
+      "loss": 1.0469,
+      "step": 7781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.337213376443927e-05,
+      "loss": 0.5898,
+      "step": 7782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3370649550388332e-05,
+      "loss": 0.7109,
+      "step": 7783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3369165252563697e-05,
+      "loss": 0.8398,
+      "step": 7784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3367680871002244e-05,
+      "loss": 0.9727,
+      "step": 7785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3366196405740874e-05,
+      "loss": 1.4844,
+      "step": 7786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3364711856816482e-05,
+      "loss": 0.8086,
+      "step": 7787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.336322722426596e-05,
+      "loss": 1.4766,
+      "step": 7788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3361742508126208e-05,
+      "loss": 0.9102,
+      "step": 7789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.336025770843413e-05,
+      "loss": 1.6406,
+      "step": 7790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3358772825226626e-05,
+      "loss": 1.5469,
+      "step": 7791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3357287858540604e-05,
+      "loss": 0.7656,
+      "step": 7792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3355802808412967e-05,
+      "loss": 0.9492,
+      "step": 7793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3354317674880628e-05,
+      "loss": 0.8359,
+      "step": 7794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3352832457980497e-05,
+      "loss": 1.3281,
+      "step": 7795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3351347157749485e-05,
+      "loss": 0.6445,
+      "step": 7796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3349861774224515e-05,
+      "loss": 1.3203,
+      "step": 7797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3348376307442494e-05,
+      "loss": 1.1797,
+      "step": 7798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.334689075744035e-05,
+      "loss": 1.1406,
+      "step": 7799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3345405124255e-05,
+      "loss": 1.2656,
+      "step": 7800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3343919407923371e-05,
+      "loss": 1.8516,
+      "step": 7801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3342433608482387e-05,
+      "loss": 0.8086,
+      "step": 7802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3340947725968976e-05,
+      "loss": 0.8008,
+      "step": 7803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3339461760420067e-05,
+      "loss": 0.5469,
+      "step": 7804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3337975711872594e-05,
+      "loss": 1.1016,
+      "step": 7805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3336489580363489e-05,
+      "loss": 0.4082,
+      "step": 7806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.333500336592969e-05,
+      "loss": 0.3496,
+      "step": 7807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3333517068608132e-05,
+      "loss": 0.5938,
+      "step": 7808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3332030688435765e-05,
+      "loss": 0.7812,
+      "step": 7809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3330544225449516e-05,
+      "loss": 0.9219,
+      "step": 7810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3329057679686342e-05,
+      "loss": 0.9766,
+      "step": 7811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3327571051183182e-05,
+      "loss": 1.1406,
+      "step": 7812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3326084339976988e-05,
+      "loss": 0.8633,
+      "step": 7813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3324597546104707e-05,
+      "loss": 1.2266,
+      "step": 7814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.33231106696033e-05,
+      "loss": 0.5273,
+      "step": 7815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3321623710509713e-05,
+      "loss": 1.4922,
+      "step": 7816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3320136668860903e-05,
+      "loss": 1.1094,
+      "step": 7817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3318649544693836e-05,
+      "loss": 0.3965,
+      "step": 7818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3317162338045467e-05,
+      "loss": 1.1562,
+      "step": 7819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3315675048952758e-05,
+      "loss": 0.4922,
+      "step": 7820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3314187677452679e-05,
+      "loss": 0.3535,
+      "step": 7821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3312700223582189e-05,
+      "loss": 1.375,
+      "step": 7822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3311212687378268e-05,
+      "loss": 1.9609,
+      "step": 7823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3309725068877879e-05,
+      "loss": 1.4609,
+      "step": 7824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3308237368117995e-05,
+      "loss": 1.1016,
+      "step": 7825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3306749585135596e-05,
+      "loss": 1.0312,
+      "step": 7826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3305261719967652e-05,
+      "loss": 0.6953,
+      "step": 7827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3303773772651151e-05,
+      "loss": 0.8555,
+      "step": 7828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3302285743223066e-05,
+      "loss": 0.9609,
+      "step": 7829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.330079763172039e-05,
+      "loss": 0.5312,
+      "step": 7830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3299309438180097e-05,
+      "loss": 0.8945,
+      "step": 7831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.329782116263918e-05,
+      "loss": 1.4688,
+      "step": 7832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3296332805134631e-05,
+      "loss": 1.1172,
+      "step": 7833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3294844365703436e-05,
+      "loss": 0.7422,
+      "step": 7834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3293355844382592e-05,
+      "loss": 1.3828,
+      "step": 7835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3291867241209095e-05,
+      "loss": 1.3516,
+      "step": 7836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.329037855621994e-05,
+      "loss": 0.9805,
+      "step": 7837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.328888978945213e-05,
+      "loss": 1.1953,
+      "step": 7838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3287400940942661e-05,
+      "loss": 0.875,
+      "step": 7839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3285912010728542e-05,
+      "loss": 0.4746,
+      "step": 7840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3284422998846775e-05,
+      "loss": 0.3438,
+      "step": 7841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3282933905334372e-05,
+      "loss": 0.9023,
+      "step": 7842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3281444730228339e-05,
+      "loss": 1.8828,
+      "step": 7843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.327995547356569e-05,
+      "loss": 1.7812,
+      "step": 7844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3278466135383435e-05,
+      "loss": 0.8711,
+      "step": 7845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3276976715718594e-05,
+      "loss": 1.4375,
+      "step": 7846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3275487214608186e-05,
+      "loss": 1.1094,
+      "step": 7847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3273997632089227e-05,
+      "loss": 0.8945,
+      "step": 7848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.327250796819874e-05,
+      "loss": 1.0938,
+      "step": 7849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3271018222973749e-05,
+      "loss": 0.6016,
+      "step": 7850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.326952839645128e-05,
+      "loss": 1.25,
+      "step": 7851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.326803848866836e-05,
+      "loss": 0.8672,
+      "step": 7852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3266548499662022e-05,
+      "loss": 0.4297,
+      "step": 7853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3265058429469295e-05,
+      "loss": 1.6484,
+      "step": 7854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3263568278127212e-05,
+      "loss": 1.0156,
+      "step": 7855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3262078045672817e-05,
+      "loss": 1.0,
+      "step": 7856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3260587732143137e-05,
+      "loss": 0.2412,
+      "step": 7857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3259097337575222e-05,
+      "loss": 1.0938,
+      "step": 7858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3257606862006102e-05,
+      "loss": 0.9492,
+      "step": 7859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3256116305472834e-05,
+      "loss": 0.4531,
+      "step": 7860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3254625668012458e-05,
+      "loss": 0.9805,
+      "step": 7861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.325313494966202e-05,
+      "loss": 0.3203,
+      "step": 7862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3251644150458577e-05,
+      "loss": 0.2617,
+      "step": 7863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3250153270439174e-05,
+      "loss": 0.2578,
+      "step": 7864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.324866230964087e-05,
+      "loss": 0.2598,
+      "step": 7865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3247171268100715e-05,
+      "loss": 0.4941,
+      "step": 7866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3245680145855775e-05,
+      "loss": 0.543,
+      "step": 7867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3244188942943108e-05,
+      "loss": 0.4785,
+      "step": 7868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.324269765939977e-05,
+      "loss": 1.3516,
+      "step": 7869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3241206295262834e-05,
+      "loss": 1.1719,
+      "step": 7870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3239714850569359e-05,
+      "loss": 0.9453,
+      "step": 7871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3238223325356418e-05,
+      "loss": 1.0703,
+      "step": 7872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3236731719661077e-05,
+      "loss": 1.6016,
+      "step": 7873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3235240033520413e-05,
+      "loss": 1.0078,
+      "step": 7874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3233748266971498e-05,
+      "loss": 1.0859,
+      "step": 7875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3232256420051408e-05,
+      "loss": 0.2812,
+      "step": 7876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.323076449279722e-05,
+      "loss": 0.9062,
+      "step": 7877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3229272485246016e-05,
+      "loss": 0.832,
+      "step": 7878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3227780397434877e-05,
+      "loss": 1.6172,
+      "step": 7879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3226288229400887e-05,
+      "loss": 1.2812,
+      "step": 7880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3224795981181135e-05,
+      "loss": 1.2266,
+      "step": 7881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3223303652812706e-05,
+      "loss": 0.7695,
+      "step": 7882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3221811244332692e-05,
+      "loss": 1.2812,
+      "step": 7883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3220318755778183e-05,
+      "loss": 0.6953,
+      "step": 7884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.321882618718627e-05,
+      "loss": 0.7969,
+      "step": 7885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3217333538594061e-05,
+      "loss": 1.2109,
+      "step": 7886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3215840810038642e-05,
+      "loss": 1.3281,
+      "step": 7887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3214348001557119e-05,
+      "loss": 0.2988,
+      "step": 7888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.321285511318659e-05,
+      "loss": 1.0938,
+      "step": 7889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3211362144964166e-05,
+      "loss": 1.3516,
+      "step": 7890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3209869096926948e-05,
+      "loss": 1.0391,
+      "step": 7891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3208375969112044e-05,
+      "loss": 0.9258,
+      "step": 7892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3206882761556565e-05,
+      "loss": 1.3281,
+      "step": 7893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3205389474297622e-05,
+      "loss": 1.0859,
+      "step": 7894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3203896107372332e-05,
+      "loss": 0.9297,
+      "step": 7895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3202402660817809e-05,
+      "loss": 0.918,
+      "step": 7896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3200909134671166e-05,
+      "loss": 1.6094,
+      "step": 7897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3199415528969533e-05,
+      "loss": 0.6289,
+      "step": 7898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3197921843750026e-05,
+      "loss": 0.7109,
+      "step": 7899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3196428079049768e-05,
+      "loss": 0.5234,
+      "step": 7900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3194934234905885e-05,
+      "loss": 1.0078,
+      "step": 7901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3193440311355507e-05,
+      "loss": 1.7344,
+      "step": 7902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3191946308435764e-05,
+      "loss": 1.0703,
+      "step": 7903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3190452226183783e-05,
+      "loss": 1.2031,
+      "step": 7904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3188958064636705e-05,
+      "loss": 0.5547,
+      "step": 7905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.318746382383166e-05,
+      "loss": 0.832,
+      "step": 7906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3185969503805789e-05,
+      "loss": 1.0938,
+      "step": 7907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.318447510459623e-05,
+      "loss": 1.3359,
+      "step": 7908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3182980626240124e-05,
+      "loss": 1.5781,
+      "step": 7909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3181486068774619e-05,
+      "loss": 0.8516,
+      "step": 7910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3179991432236854e-05,
+      "loss": 0.4805,
+      "step": 7911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.317849671666398e-05,
+      "loss": 1.1406,
+      "step": 7912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3177001922093147e-05,
+      "loss": 0.3379,
+      "step": 7913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3175507048561504e-05,
+      "loss": 1.4922,
+      "step": 7914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3174012096106207e-05,
+      "loss": 0.3398,
+      "step": 7915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.317251706476441e-05,
+      "loss": 1.1094,
+      "step": 7916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.317102195457327e-05,
+      "loss": 1.3281,
+      "step": 7917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3169526765569946e-05,
+      "loss": 0.4688,
+      "step": 7918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3168031497791605e-05,
+      "loss": 1.2969,
+      "step": 7919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3166536151275397e-05,
+      "loss": 0.7305,
+      "step": 7920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3165040726058502e-05,
+      "loss": 0.7578,
+      "step": 7921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.316354522217808e-05,
+      "loss": 1.6328,
+      "step": 7922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3162049639671299e-05,
+      "loss": 1.6875,
+      "step": 7923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3160553978575331e-05,
+      "loss": 1.2891,
+      "step": 7924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3159058238927353e-05,
+      "loss": 0.8086,
+      "step": 7925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3157562420764533e-05,
+      "loss": 1.2188,
+      "step": 7926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3156066524124053e-05,
+      "loss": 1.5859,
+      "step": 7927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3154570549043091e-05,
+      "loss": 0.9727,
+      "step": 7928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3153074495558826e-05,
+      "loss": 1.2734,
+      "step": 7929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3151578363708444e-05,
+      "loss": 1.4922,
+      "step": 7930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3150082153529125e-05,
+      "loss": 1.0703,
+      "step": 7931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3148585865058056e-05,
+      "loss": 0.9375,
+      "step": 7932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3147089498332432e-05,
+      "loss": 0.5859,
+      "step": 7933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3145593053389436e-05,
+      "loss": 0.7773,
+      "step": 7934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3144096530266268e-05,
+      "loss": 1.0312,
+      "step": 7935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3142599929000113e-05,
+      "loss": 0.9844,
+      "step": 7936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3141103249628175e-05,
+      "loss": 0.6992,
+      "step": 7937
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3139606492187649e-05,
+      "loss": 1.0156,
+      "step": 7938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3138109656715737e-05,
+      "loss": 1.3047,
+      "step": 7939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3136612743249637e-05,
+      "loss": 0.5859,
+      "step": 7940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.313511575182656e-05,
+      "loss": 0.6445,
+      "step": 7941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3133618682483704e-05,
+      "loss": 0.498,
+      "step": 7942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3132121535258285e-05,
+      "loss": 1.1406,
+      "step": 7943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3130624310187506e-05,
+      "loss": 1.2266,
+      "step": 7944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3129127007308585e-05,
+      "loss": 1.0938,
+      "step": 7945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3127629626658727e-05,
+      "loss": 1.2188,
+      "step": 7946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3126132168275159e-05,
+      "loss": 1.1328,
+      "step": 7947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.312463463219509e-05,
+      "loss": 1.2344,
+      "step": 7948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3123137018455748e-05,
+      "loss": 0.6094,
+      "step": 7949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3121639327094342e-05,
+      "loss": 0.9219,
+      "step": 7950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3120141558148106e-05,
+      "loss": 0.8828,
+      "step": 7951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.311864371165426e-05,
+      "loss": 0.7227,
+      "step": 7952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3117145787650034e-05,
+      "loss": 0.8281,
+      "step": 7953
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3115647786172658e-05,
+      "loss": 0.9062,
+      "step": 7954
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3114149707259361e-05,
+      "loss": 0.582,
+      "step": 7955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3112651550947375e-05,
+      "loss": 1.4062,
+      "step": 7956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3111153317273936e-05,
+      "loss": 0.498,
+      "step": 7957
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3109655006276285e-05,
+      "loss": 1.2656,
+      "step": 7958
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3108156617991656e-05,
+      "loss": 0.8281,
+      "step": 7959
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3106658152457289e-05,
+      "loss": 0.8242,
+      "step": 7960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3105159609710432e-05,
+      "loss": 1.5234,
+      "step": 7961
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3103660989788323e-05,
+      "loss": 0.6641,
+      "step": 7962
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3102162292728216e-05,
+      "loss": 0.6797,
+      "step": 7963
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3100663518567348e-05,
+      "loss": 0.918,
+      "step": 7964
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3099164667342983e-05,
+      "loss": 1.4531,
+      "step": 7965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3097665739092367e-05,
+      "loss": 1.5547,
+      "step": 7966
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3096166733852753e-05,
+      "loss": 1.6328,
+      "step": 7967
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3094667651661396e-05,
+      "loss": 1.0703,
+      "step": 7968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3093168492555559e-05,
+      "loss": 1.2891,
+      "step": 7969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.30916692565725e-05,
+      "loss": 0.7383,
+      "step": 7970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.8164,
+      "step": 7971
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3088670554123757e-05,
+      "loss": 1.6562,
+      "step": 7972
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3087171087732604e-05,
+      "loss": 0.2373,
+      "step": 7973
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3085671544613287e-05,
+      "loss": 0.5312,
+      "step": 7974
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3084171924803079e-05,
+      "loss": 0.9492,
+      "step": 7975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3082672228339247e-05,
+      "loss": 0.5547,
+      "step": 7976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3081172455259063e-05,
+      "loss": 0.5234,
+      "step": 7977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3079672605599802e-05,
+      "loss": 0.1855,
+      "step": 7978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3078172679398747e-05,
+      "loss": 1.125,
+      "step": 7979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3076672676693175e-05,
+      "loss": 0.2314,
+      "step": 7980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.307517259752036e-05,
+      "loss": 0.1895,
+      "step": 7981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.307367244191759e-05,
+      "loss": 0.7891,
+      "step": 7982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3072172209922152e-05,
+      "loss": 0.3066,
+      "step": 7983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.307067190157133e-05,
+      "loss": 0.7227,
+      "step": 7984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3069171516902413e-05,
+      "loss": 1.0703,
+      "step": 7985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3067671055952688e-05,
+      "loss": 0.875,
+      "step": 7986
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3066170518759453e-05,
+      "loss": 0.7695,
+      "step": 7987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.306466990536e-05,
+      "loss": 0.9492,
+      "step": 7988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3063169215791622e-05,
+      "loss": 0.5469,
+      "step": 7989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3061668450091615e-05,
+      "loss": 1.1094,
+      "step": 7990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3060167608297291e-05,
+      "loss": 0.3477,
+      "step": 7991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3058666690445942e-05,
+      "loss": 0.5039,
+      "step": 7992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3057165696574875e-05,
+      "loss": 0.8633,
+      "step": 7993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.305566462672139e-05,
+      "loss": 1.3984,
+      "step": 7994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3054163480922798e-05,
+      "loss": 0.543,
+      "step": 7995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3052662259216416e-05,
+      "loss": 0.3613,
+      "step": 7996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3051160961639541e-05,
+      "loss": 0.9336,
+      "step": 7997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3049659588229494e-05,
+      "loss": 0.3652,
+      "step": 7998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.304815813902359e-05,
+      "loss": 0.7266,
+      "step": 7999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3046656614059145e-05,
+      "loss": 0.9492,
+      "step": 8000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.304515501337348e-05,
+      "loss": 0.6328,
+      "step": 8001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3043653337003908e-05,
+      "loss": 1.0234,
+      "step": 8002
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3042151584987758e-05,
+      "loss": 0.6562,
+      "step": 8003
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3040649757362356e-05,
+      "loss": 0.8125,
+      "step": 8004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3039147854165021e-05,
+      "loss": 0.9766,
+      "step": 8005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3037645875433086e-05,
+      "loss": 2.125,
+      "step": 8006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3036143821203883e-05,
+      "loss": 0.4316,
+      "step": 8007
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3034641691514743e-05,
+      "loss": 0.4922,
+      "step": 8008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3033139486402995e-05,
+      "loss": 0.4023,
+      "step": 8009
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3031637205905973e-05,
+      "loss": 1.0312,
+      "step": 8010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3030134850061024e-05,
+      "loss": 0.5117,
+      "step": 8011
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.302863241890548e-05,
+      "loss": 1.0078,
+      "step": 8012
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.302712991247669e-05,
+      "loss": 0.4531,
+      "step": 8013
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3025627330811985e-05,
+      "loss": 0.9297,
+      "step": 8014
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3024124673948722e-05,
+      "loss": 1.3672,
+      "step": 8015
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3022621941924238e-05,
+      "loss": 0.7812,
+      "step": 8016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3021119134775888e-05,
+      "loss": 1.0547,
+      "step": 8017
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3019616252541018e-05,
+      "loss": 0.9336,
+      "step": 8018
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3018113295256992e-05,
+      "loss": 1.2734,
+      "step": 8019
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3016610262961149e-05,
+      "loss": 1.125,
+      "step": 8020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3015107155690852e-05,
+      "loss": 0.7578,
+      "step": 8021
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3013603973483457e-05,
+      "loss": 1.2031,
+      "step": 8022
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.301210071637633e-05,
+      "loss": 0.625,
+      "step": 8023
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3010597384406828e-05,
+      "loss": 0.4863,
+      "step": 8024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3009093977612315e-05,
+      "loss": 1.2266,
+      "step": 8025
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3007590496030151e-05,
+      "loss": 0.8008,
+      "step": 8026
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3006086939697715e-05,
+      "loss": 0.252,
+      "step": 8027
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3004583308652368e-05,
+      "loss": 0.1426,
+      "step": 8028
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3003079602931486e-05,
+      "loss": 0.9531,
+      "step": 8029
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3001575822572434e-05,
+      "loss": 1.1328,
+      "step": 8030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3000071967612594e-05,
+      "loss": 0.8086,
+      "step": 8031
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.299856803808934e-05,
+      "loss": 1.1016,
+      "step": 8032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2997064034040055e-05,
+      "loss": 0.21,
+      "step": 8033
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2995559955502109e-05,
+      "loss": 1.1641,
+      "step": 8034
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2994055802512896e-05,
+      "loss": 0.1172,
+      "step": 8035
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.299255157510979e-05,
+      "loss": 1.6016,
+      "step": 8036
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2991047273330187e-05,
+      "loss": 1.4375,
+      "step": 8037
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.298954289721146e-05,
+      "loss": 0.9766,
+      "step": 8038
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2988038446791013e-05,
+      "loss": 1.9922,
+      "step": 8039
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2986533922106235e-05,
+      "loss": 0.2598,
+      "step": 8040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2985029323194516e-05,
+      "loss": 0.5234,
+      "step": 8041
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2983524650093246e-05,
+      "loss": 0.8164,
+      "step": 8042
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.298201990283983e-05,
+      "loss": 0.5352,
+      "step": 8043
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2980515081471666e-05,
+      "loss": 0.7734,
+      "step": 8044
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2979010186026151e-05,
+      "loss": 1.0,
+      "step": 8045
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2977505216540693e-05,
+      "loss": 0.3535,
+      "step": 8046
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2976000173052688e-05,
+      "loss": 0.9727,
+      "step": 8047
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2974495055599548e-05,
+      "loss": 0.8125,
+      "step": 8048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.297298986421868e-05,
+      "loss": 1.125,
+      "step": 8049
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2971484598947494e-05,
+      "loss": 1.3047,
+      "step": 8050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.29699792598234e-05,
+      "loss": 0.2402,
+      "step": 8051
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2968473846883818e-05,
+      "loss": 1.5469,
+      "step": 8052
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2966968360166155e-05,
+      "loss": 0.9102,
+      "step": 8053
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2965462799707832e-05,
+      "loss": 0.8945,
+      "step": 8054
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2963957165546265e-05,
+      "loss": 0.4785,
+      "step": 8055
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2962451457718877e-05,
+      "loss": 0.8242,
+      "step": 8056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2960945676263095e-05,
+      "loss": 0.4219,
+      "step": 8057
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2959439821216337e-05,
+      "loss": 0.8867,
+      "step": 8058
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2957933892616027e-05,
+      "loss": 1.2344,
+      "step": 8059
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2956427890499606e-05,
+      "loss": 0.6094,
+      "step": 8060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2954921814904492e-05,
+      "loss": 0.8633,
+      "step": 8061
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2953415665868124e-05,
+      "loss": 1.1094,
+      "step": 8062
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2951909443427928e-05,
+      "loss": 0.8633,
+      "step": 8063
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2950403147621345e-05,
+      "loss": 0.5742,
+      "step": 8064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2948896778485811e-05,
+      "loss": 1.0,
+      "step": 8065
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2947390336058767e-05,
+      "loss": 1.0703,
+      "step": 8066
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.294588382037765e-05,
+      "loss": 0.9922,
+      "step": 8067
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2944377231479908e-05,
+      "loss": 1.0625,
+      "step": 8068
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2942870569402978e-05,
+      "loss": 0.4199,
+      "step": 8069
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2941363834184315e-05,
+      "loss": 1.1641,
+      "step": 8070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.293985702586136e-05,
+      "loss": 0.8516,
+      "step": 8071
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2938350144471567e-05,
+      "loss": 1.2266,
+      "step": 8072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2936843190052391e-05,
+      "loss": 1.1172,
+      "step": 8073
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2935336162641276e-05,
+      "loss": 1.0703,
+      "step": 8074
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2933829062275684e-05,
+      "loss": 0.8047,
+      "step": 8075
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2932321888993071e-05,
+      "loss": 1.2578,
+      "step": 8076
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.29308146428309e-05,
+      "loss": 0.9844,
+      "step": 8077
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2929307323826626e-05,
+      "loss": 1.4219,
+      "step": 8078
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2927799932017713e-05,
+      "loss": 1.5234,
+      "step": 8079
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.292629246744163e-05,
+      "loss": 0.2422,
+      "step": 8080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.292478493013584e-05,
+      "loss": 0.8438,
+      "step": 8081
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2923277320137806e-05,
+      "loss": 1.3516,
+      "step": 8082
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2921769637485006e-05,
+      "loss": 0.6484,
+      "step": 8083
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2920261882214909e-05,
+      "loss": 0.7656,
+      "step": 8084
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.291875405436499e-05,
+      "loss": 0.5898,
+      "step": 8085
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2917246153972722e-05,
+      "loss": 0.2871,
+      "step": 8086
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.291573818107558e-05,
+      "loss": 0.4863,
+      "step": 8087
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2914230135711047e-05,
+      "loss": 0.832,
+      "step": 8088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2912722017916603e-05,
+      "loss": 0.5352,
+      "step": 8089
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2911213827729734e-05,
+      "loss": 1.1406,
+      "step": 8090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2909705565187913e-05,
+      "loss": 0.4238,
+      "step": 8091
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.290819723032864e-05,
+      "loss": 0.2383,
+      "step": 8092
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2906688823189395e-05,
+      "loss": 0.3789,
+      "step": 8093
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2905180343807668e-05,
+      "loss": 0.4277,
+      "step": 8094
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2903671792220951e-05,
+      "loss": 1.6016,
+      "step": 8095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2902163168466744e-05,
+      "loss": 0.1729,
+      "step": 8096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2900654472582532e-05,
+      "loss": 0.7617,
+      "step": 8097
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2899145704605817e-05,
+      "loss": 0.7891,
+      "step": 8098
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2897636864574097e-05,
+      "loss": 0.6602,
+      "step": 8099
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2896127952524872e-05,
+      "loss": 0.291,
+      "step": 8100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2894618968495648e-05,
+      "loss": 1.1016,
+      "step": 8101
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2893109912523927e-05,
+      "loss": 0.6602,
+      "step": 8102
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.289160078464721e-05,
+      "loss": 0.8711,
+      "step": 8103
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.289009158490301e-05,
+      "loss": 1.3047,
+      "step": 8104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2888582313328835e-05,
+      "loss": 0.8203,
+      "step": 8105
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2887072969962198e-05,
+      "loss": 1.2031,
+      "step": 8106
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2885563554840607e-05,
+      "loss": 2.5781,
+      "step": 8107
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2884054068001584e-05,
+      "loss": 1.1172,
+      "step": 8108
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.288254450948264e-05,
+      "loss": 0.7578,
+      "step": 8109
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2881034879321297e-05,
+      "loss": 0.7578,
+      "step": 8110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2879525177555068e-05,
+      "loss": 0.3047,
+      "step": 8111
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2878015404221484e-05,
+      "loss": 0.8555,
+      "step": 8112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.287650555935807e-05,
+      "loss": 1.625,
+      "step": 8113
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2874995643002342e-05,
+      "loss": 0.6836,
+      "step": 8114
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2873485655191828e-05,
+      "loss": 1.3438,
+      "step": 8115
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2871975595964065e-05,
+      "loss": 0.832,
+      "step": 8116
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2870465465356578e-05,
+      "loss": 0.7461,
+      "step": 8117
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2868955263406906e-05,
+      "loss": 0.9766,
+      "step": 8118
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2867444990152574e-05,
+      "loss": 1.1094,
+      "step": 8119
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2865934645631124e-05,
+      "loss": 0.9609,
+      "step": 8120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2864424229880092e-05,
+      "loss": 0.7461,
+      "step": 8121
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.286291374293702e-05,
+      "loss": 0.9062,
+      "step": 8122
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.286140318483945e-05,
+      "loss": 0.7617,
+      "step": 8123
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2859892555624916e-05,
+      "loss": 1.3359,
+      "step": 8124
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2858381855330975e-05,
+      "loss": 1.4219,
+      "step": 8125
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2856871083995168e-05,
+      "loss": 0.4355,
+      "step": 8126
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2855360241655043e-05,
+      "loss": 0.918,
+      "step": 8127
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.285384932834815e-05,
+      "loss": 0.8125,
+      "step": 8128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.285233834411205e-05,
+      "loss": 0.832,
+      "step": 8129
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2850827288984285e-05,
+      "loss": 1.2891,
+      "step": 8130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2849316163002415e-05,
+      "loss": 1.1406,
+      "step": 8131
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2847804966203997e-05,
+      "loss": 0.9297,
+      "step": 8132
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2846293698626588e-05,
+      "loss": 0.7734,
+      "step": 8133
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2844782360307756e-05,
+      "loss": 0.6328,
+      "step": 8134
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2843270951285058e-05,
+      "loss": 1.9609,
+      "step": 8135
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2841759471596058e-05,
+      "loss": 0.8789,
+      "step": 8136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2840247921278324e-05,
+      "loss": 0.9336,
+      "step": 8137
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2838736300369423e-05,
+      "loss": 1.4609,
+      "step": 8138
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2837224608906926e-05,
+      "loss": 0.7344,
+      "step": 8139
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.28357128469284e-05,
+      "loss": 0.4863,
+      "step": 8140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2834201014471425e-05,
+      "loss": 0.4375,
+      "step": 8141
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2832689111573573e-05,
+      "loss": 0.8594,
+      "step": 8142
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2831177138272417e-05,
+      "loss": 1.1172,
+      "step": 8143
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.282966509460554e-05,
+      "loss": 1.2969,
+      "step": 8144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.282815298061052e-05,
+      "loss": 1.0938,
+      "step": 8145
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2826640796324943e-05,
+      "loss": 0.9258,
+      "step": 8146
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2825128541786385e-05,
+      "loss": 1.2891,
+      "step": 8147
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2823616217032436e-05,
+      "loss": 1.0625,
+      "step": 8148
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2822103822100682e-05,
+      "loss": 1.2188,
+      "step": 8149
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2820591357028715e-05,
+      "loss": 0.9492,
+      "step": 8150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2819078821854123e-05,
+      "loss": 0.8008,
+      "step": 8151
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2817566216614496e-05,
+      "loss": 1.2422,
+      "step": 8152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2816053541347433e-05,
+      "loss": 0.5078,
+      "step": 8153
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2814540796090526e-05,
+      "loss": 0.6836,
+      "step": 8154
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2813027980881374e-05,
+      "loss": 0.6406,
+      "step": 8155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2811515095757576e-05,
+      "loss": 0.8359,
+      "step": 8156
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2810002140756737e-05,
+      "loss": 0.6562,
+      "step": 8157
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2808489115916451e-05,
+      "loss": 0.9531,
+      "step": 8158
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2806976021274332e-05,
+      "loss": 1.0469,
+      "step": 8159
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2805462856867978e-05,
+      "loss": 1.5,
+      "step": 8160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2803949622735003e-05,
+      "loss": 1.0859,
+      "step": 8161
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2802436318913017e-05,
+      "loss": 0.2637,
+      "step": 8162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2800922945439627e-05,
+      "loss": 1.1875,
+      "step": 8163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2799409502352447e-05,
+      "loss": 1.3359,
+      "step": 8164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2797895989689096e-05,
+      "loss": 1.1797,
+      "step": 8165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2796382407487184e-05,
+      "loss": 0.9805,
+      "step": 8166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.279486875578434e-05,
+      "loss": 1.2812,
+      "step": 8167
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2793355034618171e-05,
+      "loss": 0.8203,
+      "step": 8168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.2791841244026308e-05,
+      "loss": 1.0312,
+      "step": 8169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.279032738404637e-05,
+      "loss": 0.5117,
+      "step": 8170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2788813454715988e-05,
+      "loss": 0.8477,
+      "step": 8171
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2787299456072779e-05,
+      "loss": 1.4297,
+      "step": 8172
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2785785388154386e-05,
+      "loss": 1.25,
+      "step": 8173
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2784271250998425e-05,
+      "loss": 1.4453,
+      "step": 8174
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2782757044642537e-05,
+      "loss": 0.3984,
+      "step": 8175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2781242769124351e-05,
+      "loss": 0.5352,
+      "step": 8176
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2779728424481507e-05,
+      "loss": 0.8867,
+      "step": 8177
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2778214010751642e-05,
+      "loss": 0.2949,
+      "step": 8178
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2776699527972393e-05,
+      "loss": 0.3633,
+      "step": 8179
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2775184976181398e-05,
+      "loss": 0.2773,
+      "step": 8180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2773670355416303e-05,
+      "loss": 0.8555,
+      "step": 8181
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2772155665714753e-05,
+      "loss": 0.5039,
+      "step": 8182
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2770640907114396e-05,
+      "loss": 1.1875,
+      "step": 8183
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.276912607965287e-05,
+      "loss": 0.8594,
+      "step": 8184
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2767611183367837e-05,
+      "loss": 0.7734,
+      "step": 8185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2766096218296937e-05,
+      "loss": 1.8438,
+      "step": 8186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2764581184477831e-05,
+      "loss": 1.5078,
+      "step": 8187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2763066081948166e-05,
+      "loss": 1.3125,
+      "step": 8188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2761550910745603e-05,
+      "loss": 1.4844,
+      "step": 8189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2760035670907805e-05,
+      "loss": 1.1641,
+      "step": 8190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.275852036247242e-05,
+      "loss": 1.3125,
+      "step": 8191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2757004985477112e-05,
+      "loss": 0.8945,
+      "step": 8192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2755489539959553e-05,
+      "loss": 1.0625,
+      "step": 8193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2753974025957398e-05,
+      "loss": 0.7891,
+      "step": 8194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.275245844350832e-05,
+      "loss": 0.8125,
+      "step": 8195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2750942792649983e-05,
+      "loss": 1.0078,
+      "step": 8196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2749427073420056e-05,
+      "loss": 1.4141,
+      "step": 8197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2747911285856214e-05,
+      "loss": 1.6328,
+      "step": 8198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2746395429996125e-05,
+      "loss": 0.7461,
+      "step": 8199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2744879505877475e-05,
+      "loss": 0.918,
+      "step": 8200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2743363513537925e-05,
+      "loss": 1.0547,
+      "step": 8201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2741847453015166e-05,
+      "loss": 0.5547,
+      "step": 8202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2740331324346871e-05,
+      "loss": 1.4375,
+      "step": 8203
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2738815127570724e-05,
+      "loss": 1.3359,
+      "step": 8204
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2737298862724407e-05,
+      "loss": 1.1562,
+      "step": 8205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2735782529845612e-05,
+      "loss": 0.6094,
+      "step": 8206
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2734266128972018e-05,
+      "loss": 0.5938,
+      "step": 8207
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2732749660141312e-05,
+      "loss": 0.8633,
+      "step": 8208
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2731233123391189e-05,
+      "loss": 0.8516,
+      "step": 8209
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2729716518759342e-05,
+      "loss": 0.7266,
+      "step": 8210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2728199846283457e-05,
+      "loss": 1.0703,
+      "step": 8211
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2726683106001243e-05,
+      "loss": 1.2109,
+      "step": 8212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.272516629795038e-05,
+      "loss": 1.0156,
+      "step": 8213
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2723649422168575e-05,
+      "loss": 0.6133,
+      "step": 8214
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2722132478693532e-05,
+      "loss": 1.25,
+      "step": 8215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2720615467562947e-05,
+      "loss": 0.7305,
+      "step": 8216
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2719098388814524e-05,
+      "loss": 1.6094,
+      "step": 8217
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2717581242485973e-05,
+      "loss": 1.0703,
+      "step": 8218
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2716064028614996e-05,
+      "loss": 0.6992,
+      "step": 8219
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2714546747239302e-05,
+      "loss": 0.7305,
+      "step": 8220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2713029398396601e-05,
+      "loss": 0.7422,
+      "step": 8221
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.271151198212461e-05,
+      "loss": 2.0781,
+      "step": 8222
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.270999449846104e-05,
+      "loss": 1.2891,
+      "step": 8223
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2708476947443604e-05,
+      "loss": 1.4453,
+      "step": 8224
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2706959329110018e-05,
+      "loss": 0.4238,
+      "step": 8225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2705441643498007e-05,
+      "loss": 1.3438,
+      "step": 8226
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2703923890645286e-05,
+      "loss": 0.208,
+      "step": 8227
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2702406070589581e-05,
+      "loss": 0.6445,
+      "step": 8228
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2700888183368609e-05,
+      "loss": 0.291,
+      "step": 8229
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2699370229020105e-05,
+      "loss": 0.1504,
+      "step": 8230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2697852207581786e-05,
+      "loss": 0.0864,
+      "step": 8231
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2696334119091391e-05,
+      "loss": 1.1797,
+      "step": 8232
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.269481596358664e-05,
+      "loss": 1.1719,
+      "step": 8233
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2693297741105273e-05,
+      "loss": 0.793,
+      "step": 8234
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.269177945168502e-05,
+      "loss": 1.4062,
+      "step": 8235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2690261095363615e-05,
+      "loss": 0.9297,
+      "step": 8236
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.26887426721788e-05,
+      "loss": 0.5078,
+      "step": 8237
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.268722418216831e-05,
+      "loss": 0.8711,
+      "step": 8238
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2685705625369888e-05,
+      "loss": 1.6797,
+      "step": 8239
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2684187001821272e-05,
+      "loss": 1.2969,
+      "step": 8240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2682668311560206e-05,
+      "loss": 1.2188,
+      "step": 8241
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2681149554624442e-05,
+      "loss": 1.0312,
+      "step": 8242
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.267963073105172e-05,
+      "loss": 0.498,
+      "step": 8243
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2678111840879791e-05,
+      "loss": 0.5547,
+      "step": 8244
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2676592884146402e-05,
+      "loss": 0.3203,
+      "step": 8245
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2675073860889313e-05,
+      "loss": 0.832,
+      "step": 8246
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2673554771146272e-05,
+      "loss": 1.1094,
+      "step": 8247
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2672035614955035e-05,
+      "loss": 1.6562,
+      "step": 8248
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.2670516392353354e-05,
+      "loss": 0.8281,
+      "step": 8249
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2668997103378998e-05,
+      "loss": 0.6172,
+      "step": 8250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2667477748069721e-05,
+      "loss": 0.5352,
+      "step": 8251
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2665958326463287e-05,
+      "loss": 1.1797,
+      "step": 8252
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2664438838597452e-05,
+      "loss": 1.0625,
+      "step": 8253
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.266291928450999e-05,
+      "loss": 1.2109,
+      "step": 8254
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2661399664238668e-05,
+      "loss": 0.6797,
+      "step": 8255
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2659879977821253e-05,
+      "loss": 1.375,
+      "step": 8256
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2658360225295507e-05,
+      "loss": 0.8281,
+      "step": 8257
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2656840406699214e-05,
+      "loss": 1.3594,
+      "step": 8258
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2655320522070138e-05,
+      "loss": 1.2188,
+      "step": 8259
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.265380057144606e-05,
+      "loss": 1.0234,
+      "step": 8260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.265228055486475e-05,
+      "loss": 0.7461,
+      "step": 8261
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2650760472364e-05,
+      "loss": 0.8633,
+      "step": 8262
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2649240323981575e-05,
+      "loss": 1.1406,
+      "step": 8263
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2647720109755262e-05,
+      "loss": 1.2266,
+      "step": 8264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2646199829722843e-05,
+      "loss": 1.1016,
+      "step": 8265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2644679483922107e-05,
+      "loss": 0.8164,
+      "step": 8266
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.264315907239084e-05,
+      "loss": 0.8086,
+      "step": 8267
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2641638595166825e-05,
+      "loss": 0.4688,
+      "step": 8268
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2640118052287856e-05,
+      "loss": 0.5469,
+      "step": 8269
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.263859744379172e-05,
+      "loss": 0.7344,
+      "step": 8270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2637076769716214e-05,
+      "loss": 0.875,
+      "step": 8271
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2635556030099136e-05,
+      "loss": 0.7305,
+      "step": 8272
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2634035224978277e-05,
+      "loss": 1.0859,
+      "step": 8273
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2632514354391429e-05,
+      "loss": 0.9727,
+      "step": 8274
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2630993418376406e-05,
+      "loss": 1.1094,
+      "step": 8275
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2629472416970998e-05,
+      "loss": 0.6094,
+      "step": 8276
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2627951350213015e-05,
+      "loss": 0.2656,
+      "step": 8277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2626430218140251e-05,
+      "loss": 0.3184,
+      "step": 8278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2624909020790523e-05,
+      "loss": 0.8477,
+      "step": 8279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2623387758201635e-05,
+      "loss": 0.4336,
+      "step": 8280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2621866430411394e-05,
+      "loss": 0.5234,
+      "step": 8281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2620345037457613e-05,
+      "loss": 1.0469,
+      "step": 8282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2618823579378103e-05,
+      "loss": 0.5195,
+      "step": 8283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2617302056210683e-05,
+      "loss": 0.2988,
+      "step": 8284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2615780467993165e-05,
+      "loss": 0.3535,
+      "step": 8285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2614258814763361e-05,
+      "loss": 0.9961,
+      "step": 8286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2612737096559098e-05,
+      "loss": 0.3262,
+      "step": 8287
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2611215313418198e-05,
+      "loss": 1.2266,
+      "step": 8288
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2609693465378476e-05,
+      "loss": 0.8438,
+      "step": 8289
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2608171552477755e-05,
+      "loss": 0.3145,
+      "step": 8290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2606649574753869e-05,
+      "loss": 0.5703,
+      "step": 8291
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2605127532244641e-05,
+      "loss": 0.9375,
+      "step": 8292
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2603605424987896e-05,
+      "loss": 0.4375,
+      "step": 8293
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2602083253021466e-05,
+      "loss": 0.2852,
+      "step": 8294
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.260056101638319e-05,
+      "loss": 0.5898,
+      "step": 8295
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2599038715110894e-05,
+      "loss": 1.5703,
+      "step": 8296
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2597516349242414e-05,
+      "loss": 1.3984,
+      "step": 8297
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2595993918815583e-05,
+      "loss": 0.9102,
+      "step": 8298
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2594471423868248e-05,
+      "loss": 1.6562,
+      "step": 8299
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2592948864438248e-05,
+      "loss": 1.0703,
+      "step": 8300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2591426240563416e-05,
+      "loss": 1.0469,
+      "step": 8301
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2589903552281602e-05,
+      "loss": 0.4961,
+      "step": 8302
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2588380799630649e-05,
+      "loss": 1.1172,
+      "step": 8303
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2586857982648401e-05,
+      "loss": 1.3516,
+      "step": 8304
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2585335101372714e-05,
+      "loss": 0.6211,
+      "step": 8305
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2583812155841425e-05,
+      "loss": 0.5469,
+      "step": 8306
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2582289146092395e-05,
+      "loss": 0.8359,
+      "step": 8307
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2580766072163475e-05,
+      "loss": 1.2188,
+      "step": 8308
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2579242934092515e-05,
+      "loss": 1.5312,
+      "step": 8309
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2577719731917374e-05,
+      "loss": 0.9141,
+      "step": 8310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2576196465675911e-05,
+      "loss": 0.9609,
+      "step": 8311
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2574673135405982e-05,
+      "loss": 0.5234,
+      "step": 8312
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2573149741145449e-05,
+      "loss": 1.0547,
+      "step": 8313
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2571626282932172e-05,
+      "loss": 1.3281,
+      "step": 8314
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.257010276080402e-05,
+      "loss": 0.8359,
+      "step": 8315
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2568579174798853e-05,
+      "loss": 0.9961,
+      "step": 8316
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2567055524954546e-05,
+      "loss": 0.7383,
+      "step": 8317
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2565531811308955e-05,
+      "loss": 1.0703,
+      "step": 8318
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.256400803389996e-05,
+      "loss": 0.5273,
+      "step": 8319
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2562484192765429e-05,
+      "loss": 0.7852,
+      "step": 8320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2560960287943237e-05,
+      "loss": 0.8164,
+      "step": 8321
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2559436319471257e-05,
+      "loss": 0.8594,
+      "step": 8322
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2557912287387371e-05,
+      "loss": 0.3418,
+      "step": 8323
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.255638819172945e-05,
+      "loss": 2.1094,
+      "step": 8324
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2554864032535376e-05,
+      "loss": 0.252,
+      "step": 8325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2553339809843031e-05,
+      "loss": 0.5859,
+      "step": 8326
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2551815523690297e-05,
+      "loss": 1.0312,
+      "step": 8327
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2550291174115064e-05,
+      "loss": 0.5117,
+      "step": 8328
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.2548766761155209e-05,
+      "loss": 0.4766,
+      "step": 8329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2547242284848622e-05,
+      "loss": 0.1328,
+      "step": 8330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2545717745233194e-05,
+      "loss": 0.4805,
+      "step": 8331
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.254419314234682e-05,
+      "loss": 0.4004,
+      "step": 8332
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2542668476227387e-05,
+      "loss": 0.5664,
+      "step": 8333
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2541143746912786e-05,
+      "loss": 0.207,
+      "step": 8334
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2539618954440915e-05,
+      "loss": 1.0703,
+      "step": 8335
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2538094098849676e-05,
+      "loss": 1.25,
+      "step": 8336
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2536569180176962e-05,
+      "loss": 0.3848,
+      "step": 8337
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2535044198460673e-05,
+      "loss": 0.5508,
+      "step": 8338
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2533519153738718e-05,
+      "loss": 1.6641,
+      "step": 8339
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2531994046048992e-05,
+      "loss": 1.125,
+      "step": 8340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2530468875429402e-05,
+      "loss": 0.6836,
+      "step": 8341
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2528943641917852e-05,
+      "loss": 0.9336,
+      "step": 8342
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2527418345552258e-05,
+      "loss": 0.4297,
+      "step": 8343
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2525892986370524e-05,
+      "loss": 1.8594,
+      "step": 8344
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.252436756441056e-05,
+      "loss": 0.8086,
+      "step": 8345
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.252284207971028e-05,
+      "loss": 1.0391,
+      "step": 8346
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2521316532307597e-05,
+      "loss": 0.9062,
+      "step": 8347
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2519790922240431e-05,
+      "loss": 0.3633,
+      "step": 8348
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2518265249546695e-05,
+      "loss": 1.3359,
+      "step": 8349
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2516739514264314e-05,
+      "loss": 0.6094,
+      "step": 8350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2515213716431196e-05,
+      "loss": 1.125,
+      "step": 8351
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2513687856085274e-05,
+      "loss": 1.2734,
+      "step": 8352
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2512161933264471e-05,
+      "loss": 1.0547,
+      "step": 8353
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2510635948006706e-05,
+      "loss": 0.4473,
+      "step": 8354
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.250910990034991e-05,
+      "loss": 0.7461,
+      "step": 8355
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2507583790332012e-05,
+      "loss": 0.5742,
+      "step": 8356
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2506057617990938e-05,
+      "loss": 0.5234,
+      "step": 8357
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2504531383364623e-05,
+      "loss": 0.377,
+      "step": 8358
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2503005086490996e-05,
+      "loss": 1.1172,
+      "step": 8359
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2501478727407996e-05,
+      "loss": 1.3438,
+      "step": 8360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2499952306153558e-05,
+      "loss": 1.0078,
+      "step": 8361
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2498425822765615e-05,
+      "loss": 1.0312,
+      "step": 8362
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2496899277282108e-05,
+      "loss": 1.0469,
+      "step": 8363
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.249537266974098e-05,
+      "loss": 0.5703,
+      "step": 8364
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2493846000180172e-05,
+      "loss": 1.0078,
+      "step": 8365
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.249231926863763e-05,
+      "loss": 1.0469,
+      "step": 8366
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2490792475151293e-05,
+      "loss": 1.2109,
+      "step": 8367
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.248926561975911e-05,
+      "loss": 1.2031,
+      "step": 8368
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2487738702499033e-05,
+      "loss": 0.9297,
+      "step": 8369
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.248621172340901e-05,
+      "loss": 0.7383,
+      "step": 8370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2484684682526988e-05,
+      "loss": 0.6367,
+      "step": 8371
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.248315757989093e-05,
+      "loss": 0.3184,
+      "step": 8372
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.248163041553878e-05,
+      "loss": 0.457,
+      "step": 8373
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2480103189508498e-05,
+      "loss": 1.6953,
+      "step": 8374
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.247857590183804e-05,
+      "loss": 1.1797,
+      "step": 8375
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.247704855256537e-05,
+      "loss": 0.8164,
+      "step": 8376
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2475521141728445e-05,
+      "loss": 1.0781,
+      "step": 8377
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2473993669365225e-05,
+      "loss": 0.9648,
+      "step": 8378
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2472466135513674e-05,
+      "loss": 0.7812,
+      "step": 8379
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.247093854021176e-05,
+      "loss": 0.4922,
+      "step": 8380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.246941088349745e-05,
+      "loss": 0.7188,
+      "step": 8381
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2467883165408712e-05,
+      "loss": 0.2734,
+      "step": 8382
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.246635538598351e-05,
+      "loss": 0.3828,
+      "step": 8383
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2464827545259825e-05,
+      "loss": 0.5195,
+      "step": 8384
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2463299643275621e-05,
+      "loss": 0.1836,
+      "step": 8385
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2461771680068878e-05,
+      "loss": 0.9219,
+      "step": 8386
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2460243655677565e-05,
+      "loss": 1.2891,
+      "step": 8387
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2458715570139668e-05,
+      "loss": 1.3516,
+      "step": 8388
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2457187423493163e-05,
+      "loss": 0.8164,
+      "step": 8389
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2455659215776025e-05,
+      "loss": 0.7656,
+      "step": 8390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2454130947026239e-05,
+      "loss": 0.543,
+      "step": 8391
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2452602617281793e-05,
+      "loss": 1.5859,
+      "step": 8392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2451074226580669e-05,
+      "loss": 0.9531,
+      "step": 8393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2449545774960852e-05,
+      "loss": 0.3301,
+      "step": 8394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2448017262460326e-05,
+      "loss": 0.6016,
+      "step": 8395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2446488689117089e-05,
+      "loss": 1.9844,
+      "step": 8396
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2444960054969129e-05,
+      "loss": 1.4844,
+      "step": 8397
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2443431360054434e-05,
+      "loss": 1.4062,
+      "step": 8398
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2441902604411002e-05,
+      "loss": 1.25,
+      "step": 8399
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2440373788076832e-05,
+      "loss": 0.5586,
+      "step": 8400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2438844911089916e-05,
+      "loss": 1.5312,
+      "step": 8401
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.243731597348825e-05,
+      "loss": 0.957,
+      "step": 8402
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.243578697530984e-05,
+      "loss": 0.9805,
+      "step": 8403
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2434257916592685e-05,
+      "loss": 0.9297,
+      "step": 8404
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2432728797374789e-05,
+      "loss": 1.1953,
+      "step": 8405
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2431199617694156e-05,
+      "loss": 0.9141,
+      "step": 8406
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2429670377588787e-05,
+      "loss": 0.6562,
+      "step": 8407
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2428141077096698e-05,
+      "loss": 0.8984,
+      "step": 8408
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2426611716255894e-05,
+      "loss": 1.1797,
+      "step": 8409
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.242508229510439e-05,
+      "loss": 1.1172,
+      "step": 8410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2423552813680186e-05,
+      "loss": 0.3301,
+      "step": 8411
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.242202327202131e-05,
+      "loss": 0.5234,
+      "step": 8412
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2420493670165766e-05,
+      "loss": 0.8516,
+      "step": 8413
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2418964008151576e-05,
+      "loss": 1.5703,
+      "step": 8414
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2417434286016757e-05,
+      "loss": 1.0859,
+      "step": 8415
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.241590450379933e-05,
+      "loss": 1.0,
+      "step": 8416
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2414374661537316e-05,
+      "loss": 0.8125,
+      "step": 8417
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2412844759268734e-05,
+      "loss": 1.2031,
+      "step": 8418
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.241131479703161e-05,
+      "loss": 0.9648,
+      "step": 8419
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2409784774863967e-05,
+      "loss": 0.9688,
+      "step": 8420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2408254692803837e-05,
+      "loss": 0.7031,
+      "step": 8421
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2406724550889248e-05,
+      "loss": 1.7344,
+      "step": 8422
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2405194349158227e-05,
+      "loss": 0.6875,
+      "step": 8423
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2403664087648804e-05,
+      "loss": 1.75,
+      "step": 8424
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2402133766399015e-05,
+      "loss": 0.8633,
+      "step": 8425
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2400603385446894e-05,
+      "loss": 0.5352,
+      "step": 8426
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2399072944830482e-05,
+      "loss": 0.2598,
+      "step": 8427
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2397542444587802e-05,
+      "loss": 0.9609,
+      "step": 8428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2396011884756906e-05,
+      "loss": 0.3555,
+      "step": 8429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2394481265375833e-05,
+      "loss": 0.4473,
+      "step": 8430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2392950586482619e-05,
+      "loss": 0.457,
+      "step": 8431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.239141984811531e-05,
+      "loss": 0.2217,
+      "step": 8432
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2389889050311954e-05,
+      "loss": 0.2969,
+      "step": 8433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2388358193110593e-05,
+      "loss": 0.2773,
+      "step": 8434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2386827276549273e-05,
+      "loss": 0.3008,
+      "step": 8435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2385296300666048e-05,
+      "loss": 0.9805,
+      "step": 8436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2383765265498968e-05,
+      "loss": 1.0312,
+      "step": 8437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2382234171086088e-05,
+      "loss": 0.4922,
+      "step": 8438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2380703017465452e-05,
+      "loss": 1.1406,
+      "step": 8439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.237917180467512e-05,
+      "loss": 0.9453,
+      "step": 8440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2377640532753155e-05,
+      "loss": 0.6914,
+      "step": 8441
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2376109201737605e-05,
+      "loss": 1.375,
+      "step": 8442
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2374577811666539e-05,
+      "loss": 0.5156,
+      "step": 8443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2373046362578007e-05,
+      "loss": 0.5977,
+      "step": 8444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2371514854510083e-05,
+      "loss": 0.707,
+      "step": 8445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2369983287500821e-05,
+      "loss": 0.6797,
+      "step": 8446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2368451661588292e-05,
+      "loss": 0.7031,
+      "step": 8447
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.236691997681056e-05,
+      "loss": 0.6719,
+      "step": 8448
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.23653882332057e-05,
+      "loss": 1.2422,
+      "step": 8449
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2363856430811774e-05,
+      "loss": 1.25,
+      "step": 8450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2362324569666855e-05,
+      "loss": 0.9844,
+      "step": 8451
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2360792649809015e-05,
+      "loss": 0.5742,
+      "step": 8452
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2359260671276332e-05,
+      "loss": 0.8164,
+      "step": 8453
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2357728634106881e-05,
+      "loss": 0.5938,
+      "step": 8454
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2356196538338736e-05,
+      "loss": 1.125,
+      "step": 8455
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2354664384009976e-05,
+      "loss": 0.4785,
+      "step": 8456
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2353132171158682e-05,
+      "loss": 0.7266,
+      "step": 8457
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2351599899822935e-05,
+      "loss": 1.2578,
+      "step": 8458
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2350067570040817e-05,
+      "loss": 1.4141,
+      "step": 8459
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2348535181850414e-05,
+      "loss": 1.4219,
+      "step": 8460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2347002735289814e-05,
+      "loss": 1.0,
+      "step": 8461
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.23454702303971e-05,
+      "loss": 0.8906,
+      "step": 8462
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2343937667210362e-05,
+      "loss": 0.918,
+      "step": 8463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.234240504576769e-05,
+      "loss": 0.9492,
+      "step": 8464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2340872366107175e-05,
+      "loss": 0.7773,
+      "step": 8465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2339339628266916e-05,
+      "loss": 1.2109,
+      "step": 8466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2337806832284998e-05,
+      "loss": 0.9297,
+      "step": 8467
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.233627397819952e-05,
+      "loss": 0.9492,
+      "step": 8468
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2334741066048581e-05,
+      "loss": 0.793,
+      "step": 8469
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2333208095870282e-05,
+      "loss": 0.7539,
+      "step": 8470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2331675067702723e-05,
+      "loss": 0.75,
+      "step": 8471
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2330141981583998e-05,
+      "loss": 0.8359,
+      "step": 8472
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2328608837552219e-05,
+      "loss": 1.125,
+      "step": 8473
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2327075635645485e-05,
+      "loss": 0.3223,
+      "step": 8474
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2325542375901905e-05,
+      "loss": 1.125,
+      "step": 8475
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2324009058359585e-05,
+      "loss": 1.2422,
+      "step": 8476
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2322475683056636e-05,
+      "loss": 0.6289,
+      "step": 8477
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2320942250031167e-05,
+      "loss": 0.8516,
+      "step": 8478
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.231940875932129e-05,
+      "loss": 0.7539,
+      "step": 8479
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2317875210965113e-05,
+      "loss": 1.0469,
+      "step": 8480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2316341605000758e-05,
+      "loss": 0.2598,
+      "step": 8481
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2314807941466342e-05,
+      "loss": 1.5703,
+      "step": 8482
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2313274220399975e-05,
+      "loss": 0.8672,
+      "step": 8483
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.231174044183978e-05,
+      "loss": 0.7969,
+      "step": 8484
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2310206605823877e-05,
+      "loss": 0.1963,
+      "step": 8485
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.230867271239039e-05,
+      "loss": 0.9062,
+      "step": 8486
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.230713876157744e-05,
+      "loss": 1.1641,
+      "step": 8487
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2305604753423149e-05,
+      "loss": 0.5508,
+      "step": 8488
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2304070687965648e-05,
+      "loss": 0.8633,
+      "step": 8489
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2302536565243064e-05,
+      "loss": 0.9688,
+      "step": 8490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.230100238529352e-05,
+      "loss": 0.3516,
+      "step": 8491
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2299468148155152e-05,
+      "loss": 0.9727,
+      "step": 8492
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2297933853866089e-05,
+      "loss": 1.0859,
+      "step": 8493
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2296399502464471e-05,
+      "loss": 1.1406,
+      "step": 8494
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2294865093988422e-05,
+      "loss": 0.6641,
+      "step": 8495
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2293330628476084e-05,
+      "loss": 1.3516,
+      "step": 8496
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2291796105965589e-05,
+      "loss": 0.4512,
+      "step": 8497
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2290261526495087e-05,
+      "loss": 0.5117,
+      "step": 8498
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2288726890102712e-05,
+      "loss": 1.0234,
+      "step": 8499
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.22871921968266e-05,
+      "loss": 1.0234,
+      "step": 8500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2285657446704899e-05,
+      "loss": 1.2969,
+      "step": 8501
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2284122639775755e-05,
+      "loss": 0.4375,
+      "step": 8502
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2282587776077315e-05,
+      "loss": 0.9766,
+      "step": 8503
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2281052855647723e-05,
+      "loss": 0.373,
+      "step": 8504
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2279517878525127e-05,
+      "loss": 0.2295,
+      "step": 8505
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2277982844747679e-05,
+      "loss": 1.1172,
+      "step": 8506
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2276447754353528e-05,
+      "loss": 0.5508,
+      "step": 8507
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2274912607380834e-05,
+      "loss": 0.8906,
+      "step": 8508
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.227337740386774e-05,
+      "loss": 1.5,
+      "step": 8509
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2271842143852417e-05,
+      "loss": 1.2109,
+      "step": 8510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2270306827373009e-05,
+      "loss": 0.7461,
+      "step": 8511
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2268771454467678e-05,
+      "loss": 0.7383,
+      "step": 8512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2267236025174585e-05,
+      "loss": 0.9258,
+      "step": 8513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2265700539531893e-05,
+      "loss": 0.8359,
+      "step": 8514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2264164997577763e-05,
+      "loss": 1.2188,
+      "step": 8515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2262629399350359e-05,
+      "loss": 1.0703,
+      "step": 8516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2261093744887844e-05,
+      "loss": 1.0547,
+      "step": 8517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2259558034228392e-05,
+      "loss": 0.7031,
+      "step": 8518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2258022267410165e-05,
+      "loss": 0.7422,
+      "step": 8519
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2256486444471336e-05,
+      "loss": 0.625,
+      "step": 8520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2254950565450072e-05,
+      "loss": 0.8008,
+      "step": 8521
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2253414630384554e-05,
+      "loss": 0.793,
+      "step": 8522
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2251878639312947e-05,
+      "loss": 1.4297,
+      "step": 8523
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2250342592273431e-05,
+      "loss": 0.4883,
+      "step": 8524
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.224880648930418e-05,
+      "loss": 0.7383,
+      "step": 8525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2247270330443375e-05,
+      "loss": 1.875,
+      "step": 8526
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2245734115729195e-05,
+      "loss": 0.5664,
+      "step": 8527
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2244197845199821e-05,
+      "loss": 1.3125,
+      "step": 8528
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2242661518893431e-05,
+      "loss": 0.5469,
+      "step": 8529
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2241125136848214e-05,
+      "loss": 0.3574,
+      "step": 8530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2239588699102352e-05,
+      "loss": 0.7578,
+      "step": 8531
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2238052205694036e-05,
+      "loss": 0.6133,
+      "step": 8532
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2236515656661445e-05,
+      "loss": 1.3438,
+      "step": 8533
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2234979052042776e-05,
+      "loss": 0.3574,
+      "step": 8534
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2233442391876217e-05,
+      "loss": 0.373,
+      "step": 8535
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2231905676199962e-05,
+      "loss": 0.4121,
+      "step": 8536
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2230368905052198e-05,
+      "loss": 1.4766,
+      "step": 8537
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2228832078471131e-05,
+      "loss": 0.9883,
+      "step": 8538
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2227295196494946e-05,
+      "loss": 1.1562,
+      "step": 8539
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2225758259161844e-05,
+      "loss": 0.5664,
+      "step": 8540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2224221266510026e-05,
+      "loss": 1.0156,
+      "step": 8541
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.222268421857769e-05,
+      "loss": 0.8828,
+      "step": 8542
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.222114711540304e-05,
+      "loss": 0.7227,
+      "step": 8543
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2219609957024277e-05,
+      "loss": 1.1641,
+      "step": 8544
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2218072743479603e-05,
+      "loss": 0.7734,
+      "step": 8545
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2216535474807227e-05,
+      "loss": 1.0312,
+      "step": 8546
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2214998151045357e-05,
+      "loss": 0.3574,
+      "step": 8547
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2213460772232201e-05,
+      "loss": 1.125,
+      "step": 8548
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2211923338405966e-05,
+      "loss": 0.9766,
+      "step": 8549
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2210385849604864e-05,
+      "loss": 1.5391,
+      "step": 8550
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2208848305867111e-05,
+      "loss": 0.7734,
+      "step": 8551
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2207310707230916e-05,
+      "loss": 0.9336,
+      "step": 8552
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2205773053734498e-05,
+      "loss": 0.3574,
+      "step": 8553
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2204235345416072e-05,
+      "loss": 0.8906,
+      "step": 8554
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2202697582313858e-05,
+      "loss": 1.3828,
+      "step": 8555
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2201159764466073e-05,
+      "loss": 0.8438,
+      "step": 8556
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2199621891910935e-05,
+      "loss": 0.5664,
+      "step": 8557
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2198083964686674e-05,
+      "loss": 0.6016,
+      "step": 8558
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2196545982831506e-05,
+      "loss": 0.8945,
+      "step": 8559
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2195007946383663e-05,
+      "loss": 0.7344,
+      "step": 8560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2193469855381363e-05,
+      "loss": 1.3984,
+      "step": 8561
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2191931709862838e-05,
+      "loss": 1.0938,
+      "step": 8562
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2190393509866318e-05,
+      "loss": 0.875,
+      "step": 8563
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2188855255430032e-05,
+      "loss": 0.7891,
+      "step": 8564
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2187316946592207e-05,
+      "loss": 1.3359,
+      "step": 8565
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2185778583391085e-05,
+      "loss": 0.9844,
+      "step": 8566
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2184240165864893e-05,
+      "loss": 1.5391,
+      "step": 8567
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.218270169405187e-05,
+      "loss": 0.4453,
+      "step": 8568
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2181163167990254e-05,
+      "loss": 0.5703,
+      "step": 8569
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2179624587718278e-05,
+      "loss": 1.2656,
+      "step": 8570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2178085953274186e-05,
+      "loss": 0.5898,
+      "step": 8571
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2176547264696218e-05,
+      "loss": 0.8086,
+      "step": 8572
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2175008522022617e-05,
+      "loss": 1.2422,
+      "step": 8573
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2173469725291624e-05,
+      "loss": 1.3594,
+      "step": 8574
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2171930874541489e-05,
+      "loss": 1.8203,
+      "step": 8575
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2170391969810456e-05,
+      "loss": 1.4062,
+      "step": 8576
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2168853011136772e-05,
+      "loss": 0.6289,
+      "step": 8577
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2167313998558683e-05,
+      "loss": 0.9062,
+      "step": 8578
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2165774932114447e-05,
+      "loss": 1.0234,
+      "step": 8579
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2164235811842308e-05,
+      "loss": 0.9297,
+      "step": 8580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2162696637780529e-05,
+      "loss": 1.4922,
+      "step": 8581
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2161157409967352e-05,
+      "loss": 0.1768,
+      "step": 8582
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.215961812844104e-05,
+      "loss": 0.1318,
+      "step": 8583
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2158078793239851e-05,
+      "loss": 0.7695,
+      "step": 8584
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2156539404402041e-05,
+      "loss": 0.7969,
+      "step": 8585
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2154999961965867e-05,
+      "loss": 0.8789,
+      "step": 8586
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2153460465969598e-05,
+      "loss": 0.4629,
+      "step": 8587
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2151920916451491e-05,
+      "loss": 0.6367,
+      "step": 8588
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.215038131344981e-05,
+      "loss": 0.3184,
+      "step": 8589
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2148841657002822e-05,
+      "loss": 0.7812,
+      "step": 8590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2147301947148788e-05,
+      "loss": 1.0938,
+      "step": 8591
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2145762183925984e-05,
+      "loss": 1.6172,
+      "step": 8592
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2144222367372675e-05,
+      "loss": 0.8516,
+      "step": 8593
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.214268249752713e-05,
+      "loss": 0.334,
+      "step": 8594
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2141142574427622e-05,
+      "loss": 1.5234,
+      "step": 8595
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2139602598112426e-05,
+      "loss": 1.0469,
+      "step": 8596
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2138062568619813e-05,
+      "loss": 0.4258,
+      "step": 8597
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2136522485988059e-05,
+      "loss": 0.8516,
+      "step": 8598
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2134982350255444e-05,
+      "loss": 0.6836,
+      "step": 8599
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2133442161460244e-05,
+      "loss": 1.1094,
+      "step": 8600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2131901919640741e-05,
+      "loss": 0.7461,
+      "step": 8601
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2130361624835212e-05,
+      "loss": 0.832,
+      "step": 8602
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2128821277081942e-05,
+      "loss": 1.0156,
+      "step": 8603
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2127280876419216e-05,
+      "loss": 1.3203,
+      "step": 8604
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2125740422885316e-05,
+      "loss": 1.0391,
+      "step": 8605
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2124199916518528e-05,
+      "loss": 0.6406,
+      "step": 8606
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2122659357357142e-05,
+      "loss": 0.6406,
+      "step": 8607
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2121118745439447e-05,
+      "loss": 0.5625,
+      "step": 8608
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2119578080803733e-05,
+      "loss": 0.8867,
+      "step": 8609
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2118037363488287e-05,
+      "loss": 0.9492,
+      "step": 8610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2116496593531406e-05,
+      "loss": 1.625,
+      "step": 8611
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2114955770971384e-05,
+      "loss": 0.4336,
+      "step": 8612
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2113414895846516e-05,
+      "loss": 0.7305,
+      "step": 8613
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2111873968195096e-05,
+      "loss": 0.5234,
+      "step": 8614
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.211033298805543e-05,
+      "loss": 0.8984,
+      "step": 8615
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2108791955465806e-05,
+      "loss": 1.5469,
+      "step": 8616
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2107250870464535e-05,
+      "loss": 1.1172,
+      "step": 8617
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.210570973308991e-05,
+      "loss": 0.5156,
+      "step": 8618
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.210416854338024e-05,
+      "loss": 0.498,
+      "step": 8619
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2102627301373832e-05,
+      "loss": 1.4766,
+      "step": 8620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2101086007108985e-05,
+      "loss": 0.9844,
+      "step": 8621
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2099544660624008e-05,
+      "loss": 0.9297,
+      "step": 8622
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2098003261957212e-05,
+      "loss": 1.4297,
+      "step": 8623
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2096461811146904e-05,
+      "loss": 0.6445,
+      "step": 8624
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2094920308231396e-05,
+      "loss": 0.707,
+      "step": 8625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2093378753249e-05,
+      "loss": 1.6172,
+      "step": 8626
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2091837146238034e-05,
+      "loss": 1.1094,
+      "step": 8627
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2090295487236809e-05,
+      "loss": 0.8867,
+      "step": 8628
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.208875377628364e-05,
+      "loss": 0.416,
+      "step": 8629
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2087212013416842e-05,
+      "loss": 1.0859,
+      "step": 8630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2085670198674742e-05,
+      "loss": 0.416,
+      "step": 8631
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2084128332095657e-05,
+      "loss": 1.0391,
+      "step": 8632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2082586413717905e-05,
+      "loss": 1.3906,
+      "step": 8633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2081044443579808e-05,
+      "loss": 0.8047,
+      "step": 8634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2079502421719695e-05,
+      "loss": 0.9023,
+      "step": 8635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.207796034817589e-05,
+      "loss": 0.126,
+      "step": 8636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2076418222986722e-05,
+      "loss": 1.1875,
+      "step": 8637
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2074876046190507e-05,
+      "loss": 1.3906,
+      "step": 8638
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2073333817825585e-05,
+      "loss": 1.0312,
+      "step": 8639
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2071791537930285e-05,
+      "loss": 1.0625,
+      "step": 8640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2070249206542937e-05,
+      "loss": 1.125,
+      "step": 8641
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2068706823701875e-05,
+      "loss": 0.8203,
+      "step": 8642
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2067164389445429e-05,
+      "loss": 1.5625,
+      "step": 8643
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.206562190381194e-05,
+      "loss": 1.3281,
+      "step": 8644
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2064079366839742e-05,
+      "loss": 0.543,
+      "step": 8645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2062536778567173e-05,
+      "loss": 1.1953,
+      "step": 8646
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2060994139032573e-05,
+      "loss": 0.5625,
+      "step": 8647
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2059451448274286e-05,
+      "loss": 0.8242,
+      "step": 8648
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2057908706330648e-05,
+      "loss": 0.3809,
+      "step": 8649
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2056365913240003e-05,
+      "loss": 0.8164,
+      "step": 8650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2054823069040695e-05,
+      "loss": 0.6211,
+      "step": 8651
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2053280173771073e-05,
+      "loss": 0.6289,
+      "step": 8652
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2051737227469485e-05,
+      "loss": 0.9531,
+      "step": 8653
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2050194230174275e-05,
+      "loss": 0.7383,
+      "step": 8654
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2048651181923791e-05,
+      "loss": 1.0625,
+      "step": 8655
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2047108082756388e-05,
+      "loss": 0.957,
+      "step": 8656
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2045564932710422e-05,
+      "loss": 1.6016,
+      "step": 8657
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2044021731824236e-05,
+      "loss": 0.2539,
+      "step": 8658
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2042478480136188e-05,
+      "loss": 0.7812,
+      "step": 8659
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2040935177684638e-05,
+      "loss": 1.0703,
+      "step": 8660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2039391824507941e-05,
+      "loss": 0.8477,
+      "step": 8661
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2037848420644454e-05,
+      "loss": 0.7305,
+      "step": 8662
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2036304966132535e-05,
+      "loss": 0.9531,
+      "step": 8663
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2034761461010548e-05,
+      "loss": 1.2422,
+      "step": 8664
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.203321790531686e-05,
+      "loss": 0.7734,
+      "step": 8665
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2031674299089821e-05,
+      "loss": 1.2656,
+      "step": 8666
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2030130642367804e-05,
+      "loss": 1.0781,
+      "step": 8667
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2028586935189176e-05,
+      "loss": 0.7188,
+      "step": 8668
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2027043177592298e-05,
+      "loss": 1.0,
+      "step": 8669
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.202549936961555e-05,
+      "loss": 0.4629,
+      "step": 8670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2023955511297287e-05,
+      "loss": 0.4297,
+      "step": 8671
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2022411602675889e-05,
+      "loss": 1.5078,
+      "step": 8672
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2020867643789726e-05,
+      "loss": 0.6289,
+      "step": 8673
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2019323634677174e-05,
+      "loss": 1.1875,
+      "step": 8674
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.20177795753766e-05,
+      "loss": 1.5469,
+      "step": 8675
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.201623546592639e-05,
+      "loss": 1.0469,
+      "step": 8676
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2014691306364911e-05,
+      "loss": 0.2471,
+      "step": 8677
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2013147096730548e-05,
+      "loss": 1.3984,
+      "step": 8678
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2011602837061678e-05,
+      "loss": 1.6172,
+      "step": 8679
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2010058527396685e-05,
+      "loss": 0.5469,
+      "step": 8680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.200851416777395e-05,
+      "loss": 0.2871,
+      "step": 8681
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2006969758231852e-05,
+      "loss": 0.5664,
+      "step": 8682
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2005425298808777e-05,
+      "loss": 0.7539,
+      "step": 8683
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2003880789543114e-05,
+      "loss": 0.4102,
+      "step": 8684
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2002336230473249e-05,
+      "loss": 0.8398,
+      "step": 8685
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2000791621637572e-05,
+      "loss": 0.7891,
+      "step": 8686
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1999246963074467e-05,
+      "loss": 0.8984,
+      "step": 8687
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1997702254822328e-05,
+      "loss": 1.2188,
+      "step": 8688
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1996157496919549e-05,
+      "loss": 0.9414,
+      "step": 8689
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1994612689404518e-05,
+      "loss": 1.5156,
+      "step": 8690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1993067832315635e-05,
+      "loss": 0.5117,
+      "step": 8691
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1991522925691292e-05,
+      "loss": 1.2812,
+      "step": 8692
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1989977969569891e-05,
+      "loss": 1.2109,
+      "step": 8693
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1988432963989825e-05,
+      "loss": 1.1016,
+      "step": 8694
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1986887908989489e-05,
+      "loss": 1.0,
+      "step": 8695
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1985342804607295e-05,
+      "loss": 0.3262,
+      "step": 8696
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1983797650881637e-05,
+      "loss": 0.8633,
+      "step": 8697
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1982252447850925e-05,
+      "loss": 0.6211,
+      "step": 8698
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.198070719555355e-05,
+      "loss": 0.5312,
+      "step": 8699
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1979161894027933e-05,
+      "loss": 0.5273,
+      "step": 8700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1977616543312471e-05,
+      "loss": 0.8359,
+      "step": 8701
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1976071143445576e-05,
+      "loss": 0.5312,
+      "step": 8702
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1974525694465653e-05,
+      "loss": 0.8633,
+      "step": 8703
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1972980196411119e-05,
+      "loss": 0.625,
+      "step": 8704
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1971434649320382e-05,
+      "loss": 0.9961,
+      "step": 8705
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1969889053231854e-05,
+      "loss": 0.9219,
+      "step": 8706
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1968343408183947e-05,
+      "loss": 1.0547,
+      "step": 8707
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1966797714215081e-05,
+      "loss": 0.9883,
+      "step": 8708
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1965251971363672e-05,
+      "loss": 0.3125,
+      "step": 8709
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1963706179668134e-05,
+      "loss": 0.4805,
+      "step": 8710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1962160339166889e-05,
+      "loss": 1.125,
+      "step": 8711
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1960614449898355e-05,
+      "loss": 0.8398,
+      "step": 8712
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1959068511900958e-05,
+      "loss": 0.5547,
+      "step": 8713
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1957522525213117e-05,
+      "loss": 0.6758,
+      "step": 8714
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1955976489873251e-05,
+      "loss": 1.0,
+      "step": 8715
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1954430405919795e-05,
+      "loss": 0.6016,
+      "step": 8716
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1952884273391168e-05,
+      "loss": 1.0234,
+      "step": 8717
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.19513380923258e-05,
+      "loss": 0.5898,
+      "step": 8718
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1949791862762121e-05,
+      "loss": 0.6172,
+      "step": 8719
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1948245584738553e-05,
+      "loss": 1.1719,
+      "step": 8720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1946699258293537e-05,
+      "loss": 0.6719,
+      "step": 8721
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.19451528834655e-05,
+      "loss": 0.9727,
+      "step": 8722
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1943606460292877e-05,
+      "loss": 0.9258,
+      "step": 8723
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.19420599888141e-05,
+      "loss": 0.7227,
+      "step": 8724
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1940513469067611e-05,
+      "loss": 1.4375,
+      "step": 8725
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1938966901091843e-05,
+      "loss": 1.4219,
+      "step": 8726
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1937420284925232e-05,
+      "loss": 1.1641,
+      "step": 8727
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1935873620606217e-05,
+      "loss": 0.7031,
+      "step": 8728
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1934326908173244e-05,
+      "loss": 1.2109,
+      "step": 8729
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.193278014766475e-05,
+      "loss": 0.2852,
+      "step": 8730
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1931233339119185e-05,
+      "loss": 0.8203,
+      "step": 8731
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1929686482574982e-05,
+      "loss": 0.7852,
+      "step": 8732
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1928139578070596e-05,
+      "loss": 0.2754,
+      "step": 8733
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.192659262564447e-05,
+      "loss": 0.9961,
+      "step": 8734
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.192504562533505e-05,
+      "loss": 0.9531,
+      "step": 8735
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1923498577180786e-05,
+      "loss": 0.4375,
+      "step": 8736
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1921951481220134e-05,
+      "loss": 1.2344,
+      "step": 8737
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1920404337491536e-05,
+      "loss": 1.2734,
+      "step": 8738
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1918857146033448e-05,
+      "loss": 1.2969,
+      "step": 8739
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1917309906884324e-05,
+      "loss": 1.5469,
+      "step": 8740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1915762620082622e-05,
+      "loss": 0.9609,
+      "step": 8741
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1914215285666798e-05,
+      "loss": 0.2617,
+      "step": 8742
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1912667903675304e-05,
+      "loss": 0.332,
+      "step": 8743
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1911120474146597e-05,
+      "loss": 0.5742,
+      "step": 8744
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1909572997119146e-05,
+      "loss": 0.3945,
+      "step": 8745
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1908025472631403e-05,
+      "loss": 0.5234,
+      "step": 8746
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1906477900721838e-05,
+      "loss": 1.6172,
+      "step": 8747
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1904930281428903e-05,
+      "loss": 0.5781,
+      "step": 8748
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1903382614791075e-05,
+      "loss": 0.9297,
+      "step": 8749
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1901834900846812e-05,
+      "loss": 0.7266,
+      "step": 8750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.190028713963458e-05,
+      "loss": 0.7031,
+      "step": 8751
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.189873933119285e-05,
+      "loss": 1.2578,
+      "step": 8752
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1897191475560093e-05,
+      "loss": 1.1094,
+      "step": 8753
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1895643572774773e-05,
+      "loss": 1.2578,
+      "step": 8754
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1894095622875366e-05,
+      "loss": 0.6328,
+      "step": 8755
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1892547625900341e-05,
+      "loss": 1.3359,
+      "step": 8756
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1890999581888176e-05,
+      "loss": 1.1328,
+      "step": 8757
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1889451490877344e-05,
+      "loss": 0.3438,
+      "step": 8758
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1887903352906323e-05,
+      "loss": 0.8789,
+      "step": 8759
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1886355168013583e-05,
+      "loss": 0.832,
+      "step": 8760
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1884806936237612e-05,
+      "loss": 0.8125,
+      "step": 8761
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1883258657616884e-05,
+      "loss": 0.6602,
+      "step": 8762
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.188171033218988e-05,
+      "loss": 1.3438,
+      "step": 8763
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.188016195999508e-05,
+      "loss": 1.1016,
+      "step": 8764
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1878613541070976e-05,
+      "loss": 1.0703,
+      "step": 8765
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1877065075456042e-05,
+      "loss": 0.8672,
+      "step": 8766
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.187551656318877e-05,
+      "loss": 1.3594,
+      "step": 8767
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.187396800430764e-05,
+      "loss": 1.2031,
+      "step": 8768
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1872419398851146e-05,
+      "loss": 0.8555,
+      "step": 8769
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1870870746857774e-05,
+      "loss": 0.9805,
+      "step": 8770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1869322048366017e-05,
+      "loss": 0.668,
+      "step": 8771
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1867773303414359e-05,
+      "loss": 1.8516,
+      "step": 8772
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1866224512041297e-05,
+      "loss": 1.7812,
+      "step": 8773
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1864675674285326e-05,
+      "loss": 0.8867,
+      "step": 8774
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1863126790184942e-05,
+      "loss": 1.0547,
+      "step": 8775
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1861577859778632e-05,
+      "loss": 1.0781,
+      "step": 8776
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1860028883104904e-05,
+      "loss": 1.2109,
+      "step": 8777
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1858479860202247e-05,
+      "loss": 1.0,
+      "step": 8778
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1856930791109168e-05,
+      "loss": 0.5078,
+      "step": 8779
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1855381675864156e-05,
+      "loss": 0.2559,
+      "step": 8780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1853832514505728e-05,
+      "loss": 1.2969,
+      "step": 8781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1852283307072374e-05,
+      "loss": 1.8828,
+      "step": 8782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1850734053602604e-05,
+      "loss": 0.5117,
+      "step": 8783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.184918475413492e-05,
+      "loss": 1.5156,
+      "step": 8784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1847635408707828e-05,
+      "loss": 0.6133,
+      "step": 8785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1846086017359842e-05,
+      "loss": 0.4668,
+      "step": 8786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.184453658012946e-05,
+      "loss": 0.6055,
+      "step": 8787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1842987097055196e-05,
+      "loss": 1.6875,
+      "step": 8788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1841437568175565e-05,
+      "loss": 0.4922,
+      "step": 8789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1839887993529072e-05,
+      "loss": 0.6094,
+      "step": 8790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1838338373154236e-05,
+      "loss": 0.4453,
+      "step": 8791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1836788707089566e-05,
+      "loss": 1.2734,
+      "step": 8792
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1835238995373577e-05,
+      "loss": 0.8789,
+      "step": 8793
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.183368923804479e-05,
+      "loss": 1.4688,
+      "step": 8794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1832139435141721e-05,
+      "loss": 1.4922,
+      "step": 8795
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1830589586702885e-05,
+      "loss": 1.8906,
+      "step": 8796
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1829039692766804e-05,
+      "loss": 1.0938,
+      "step": 8797
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1827489753372001e-05,
+      "loss": 0.832,
+      "step": 8798
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1825939768556998e-05,
+      "loss": 1.2891,
+      "step": 8799
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1824389738360314e-05,
+      "loss": 0.9531,
+      "step": 8800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1822839662820474e-05,
+      "loss": 1.2031,
+      "step": 8801
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1821289541976006e-05,
+      "loss": 1.1641,
+      "step": 8802
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1819739375865437e-05,
+      "loss": 0.7656,
+      "step": 8803
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1818189164527293e-05,
+      "loss": 0.8477,
+      "step": 8804
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1816638908000102e-05,
+      "loss": 0.6133,
+      "step": 8805
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1815088606322393e-05,
+      "loss": 0.9219,
+      "step": 8806
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.1813538259532701e-05,
+      "loss": 1.1406,
+      "step": 8807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1811987867669559e-05,
+      "loss": 0.4023,
+      "step": 8808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1810437430771491e-05,
+      "loss": 0.5156,
+      "step": 8809
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1808886948877041e-05,
+      "loss": 0.875,
+      "step": 8810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.180733642202474e-05,
+      "loss": 1.1016,
+      "step": 8811
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1805785850253126e-05,
+      "loss": 0.4141,
+      "step": 8812
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1804235233600733e-05,
+      "loss": 1.375,
+      "step": 8813
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1802684572106111e-05,
+      "loss": 0.8359,
+      "step": 8814
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1801133865807786e-05,
+      "loss": 1.5938,
+      "step": 8815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1799583114744306e-05,
+      "loss": 0.5234,
+      "step": 8816
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.179803231895421e-05,
+      "loss": 1.1406,
+      "step": 8817
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1796481478476046e-05,
+      "loss": 0.7852,
+      "step": 8818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1794930593348362e-05,
+      "loss": 0.8125,
+      "step": 8819
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.179337966360969e-05,
+      "loss": 1.0156,
+      "step": 8820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1791828689298583e-05,
+      "loss": 0.8164,
+      "step": 8821
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1790277670453593e-05,
+      "loss": 0.2314,
+      "step": 8822
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1788726607113268e-05,
+      "loss": 1.4688,
+      "step": 8823
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1787175499316154e-05,
+      "loss": 0.8633,
+      "step": 8824
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1785624347100798e-05,
+      "loss": 0.5742,
+      "step": 8825
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1784073150505763e-05,
+      "loss": 0.2051,
+      "step": 8826
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1782521909569598e-05,
+      "loss": 0.8203,
+      "step": 8827
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1780970624330856e-05,
+      "loss": 1.0859,
+      "step": 8828
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.177941929482809e-05,
+      "loss": 0.3633,
+      "step": 8829
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1777867921099859e-05,
+      "loss": 0.2969,
+      "step": 8830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1776316503184727e-05,
+      "loss": 0.8242,
+      "step": 8831
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1774765041121241e-05,
+      "loss": 0.7695,
+      "step": 8832
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1773213534947967e-05,
+      "loss": 0.2793,
+      "step": 8833
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1771661984703466e-05,
+      "loss": 0.3906,
+      "step": 8834
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1770110390426301e-05,
+      "loss": 1.0469,
+      "step": 8835
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1768558752155035e-05,
+      "loss": 0.1973,
+      "step": 8836
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1767007069928226e-05,
+      "loss": 0.9414,
+      "step": 8837
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1765455343784449e-05,
+      "loss": 0.1172,
+      "step": 8838
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1763903573762266e-05,
+      "loss": 0.7539,
+      "step": 8839
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.176235175990024e-05,
+      "loss": 0.6914,
+      "step": 8840
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1760799902236944e-05,
+      "loss": 0.416,
+      "step": 8841
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1759248000810952e-05,
+      "loss": 1.6641,
+      "step": 8842
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1757696055660827e-05,
+      "loss": 1.0156,
+      "step": 8843
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1756144066825146e-05,
+      "loss": 1.3906,
+      "step": 8844
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1754592034342476e-05,
+      "loss": 0.8867,
+      "step": 8845
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1753039958251398e-05,
+      "loss": 0.416,
+      "step": 8846
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1751487838590485e-05,
+      "loss": 0.8125,
+      "step": 8847
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.174993567539831e-05,
+      "loss": 1.2656,
+      "step": 8848
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1748383468713451e-05,
+      "loss": 0.4082,
+      "step": 8849
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1746831218574491e-05,
+      "loss": 0.8047,
+      "step": 8850
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1745278925020007e-05,
+      "loss": 0.6055,
+      "step": 8851
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1743726588088577e-05,
+      "loss": 0.8594,
+      "step": 8852
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1742174207818785e-05,
+      "loss": 0.8359,
+      "step": 8853
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1740621784249213e-05,
+      "loss": 1.0469,
+      "step": 8854
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1739069317418446e-05,
+      "loss": 0.6328,
+      "step": 8855
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1737516807365066e-05,
+      "loss": 0.2363,
+      "step": 8856
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1735964254127659e-05,
+      "loss": 0.5,
+      "step": 8857
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1734411657744818e-05,
+      "loss": 0.9453,
+      "step": 8858
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1732859018255122e-05,
+      "loss": 0.7539,
+      "step": 8859
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1731306335697169e-05,
+      "loss": 1.3281,
+      "step": 8860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.172975361010954e-05,
+      "loss": 0.5742,
+      "step": 8861
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1728200841530835e-05,
+      "loss": 1.125,
+      "step": 8862
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1726648029999639e-05,
+      "loss": 1.1406,
+      "step": 8863
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1725095175554553e-05,
+      "loss": 1.25,
+      "step": 8864
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1723542278234166e-05,
+      "loss": 0.8984,
+      "step": 8865
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.172198933807707e-05,
+      "loss": 0.5234,
+      "step": 8866
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.172043635512187e-05,
+      "loss": 0.7305,
+      "step": 8867
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1718883329407162e-05,
+      "loss": 0.9219,
+      "step": 8868
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1717330260971545e-05,
+      "loss": 0.6406,
+      "step": 8869
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1715777149853609e-05,
+      "loss": 1.0078,
+      "step": 8870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1714223996091964e-05,
+      "loss": 0.8359,
+      "step": 8871
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1712670799725212e-05,
+      "loss": 1.7969,
+      "step": 8872
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1711117560791956e-05,
+      "loss": 0.7734,
+      "step": 8873
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1709564279330792e-05,
+      "loss": 0.3301,
+      "step": 8874
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1708010955380338e-05,
+      "loss": 1.0938,
+      "step": 8875
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1706457588979189e-05,
+      "loss": 0.4043,
+      "step": 8876
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1704904180165959e-05,
+      "loss": 1.0312,
+      "step": 8877
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.170335072897925e-05,
+      "loss": 0.4199,
+      "step": 8878
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1701797235457679e-05,
+      "loss": 1.0312,
+      "step": 8879
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1700243699639853e-05,
+      "loss": 0.3008,
+      "step": 8880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.169869012156438e-05,
+      "loss": 0.1475,
+      "step": 8881
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1697136501269875e-05,
+      "loss": 0.4023,
+      "step": 8882
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1695582838794952e-05,
+      "loss": 1.5625,
+      "step": 8883
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1694029134178226e-05,
+      "loss": 0.6484,
+      "step": 8884
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1692475387458313e-05,
+      "loss": 1.2109,
+      "step": 8885
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1690921598673825e-05,
+      "loss": 0.252,
+      "step": 8886
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.1689367767863385e-05,
+      "loss": 0.7734,
+      "step": 8887
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.168781389506561e-05,
+      "loss": 0.126,
+      "step": 8888
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1686259980319119e-05,
+      "loss": 1.7266,
+      "step": 8889
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1684706023662531e-05,
+      "loss": 0.2637,
+      "step": 8890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1683152025134475e-05,
+      "loss": 0.5117,
+      "step": 8891
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1681597984773564e-05,
+      "loss": 1.3281,
+      "step": 8892
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1680043902618432e-05,
+      "loss": 0.7383,
+      "step": 8893
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1678489778707694e-05,
+      "loss": 0.7734,
+      "step": 8894
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1676935613079983e-05,
+      "loss": 0.7188,
+      "step": 8895
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1675381405773925e-05,
+      "loss": 0.7109,
+      "step": 8896
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1673827156828147e-05,
+      "loss": 1.0938,
+      "step": 8897
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1672272866281276e-05,
+      "loss": 1.5469,
+      "step": 8898
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1670718534171946e-05,
+      "loss": 0.2598,
+      "step": 8899
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1669164160538786e-05,
+      "loss": 1.0469,
+      "step": 8900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.166760974542043e-05,
+      "loss": 0.6367,
+      "step": 8901
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1666055288855506e-05,
+      "loss": 0.5586,
+      "step": 8902
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1664500790882659e-05,
+      "loss": 1.125,
+      "step": 8903
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1662946251540512e-05,
+      "loss": 1.0859,
+      "step": 8904
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1661391670867708e-05,
+      "loss": 1.0234,
+      "step": 8905
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1659837048902882e-05,
+      "loss": 0.6328,
+      "step": 8906
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1658282385684676e-05,
+      "loss": 0.6406,
+      "step": 8907
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1656727681251727e-05,
+      "loss": 0.8555,
+      "step": 8908
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1655172935642677e-05,
+      "loss": 0.9609,
+      "step": 8909
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.165361814889616e-05,
+      "loss": 1.2969,
+      "step": 8910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1652063321050825e-05,
+      "loss": 0.832,
+      "step": 8911
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.165050845214532e-05,
+      "loss": 0.2471,
+      "step": 8912
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1648953542218285e-05,
+      "loss": 1.0391,
+      "step": 8913
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1647398591308358e-05,
+      "loss": 1.0938,
+      "step": 8914
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1645843599454196e-05,
+      "loss": 1.1094,
+      "step": 8915
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1644288566694444e-05,
+      "loss": 0.6719,
+      "step": 8916
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1642733493067748e-05,
+      "loss": 0.832,
+      "step": 8917
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1641178378612756e-05,
+      "loss": 0.7734,
+      "step": 8918
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.163962322336813e-05,
+      "loss": 1.1094,
+      "step": 8919
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1638068027372509e-05,
+      "loss": 0.8555,
+      "step": 8920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1636512790664549e-05,
+      "loss": 0.8281,
+      "step": 8921
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1634957513282905e-05,
+      "loss": 1.1172,
+      "step": 8922
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1633402195266233e-05,
+      "loss": 0.625,
+      "step": 8923
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1631846836653187e-05,
+      "loss": 0.7656,
+      "step": 8924
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1630291437482424e-05,
+      "loss": 0.4609,
+      "step": 8925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1628735997792598e-05,
+      "loss": 0.3145,
+      "step": 8926
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1627180517622374e-05,
+      "loss": 0.291,
+      "step": 8927
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.162562499701041e-05,
+      "loss": 0.1582,
+      "step": 8928
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1624069435995366e-05,
+      "loss": 0.5859,
+      "step": 8929
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1622513834615896e-05,
+      "loss": 0.3223,
+      "step": 8930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.162095819291068e-05,
+      "loss": 0.1572,
+      "step": 8931
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1619402510918368e-05,
+      "loss": 0.8281,
+      "step": 8932
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1617846788677627e-05,
+      "loss": 0.1992,
+      "step": 8933
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1616291026227123e-05,
+      "loss": 0.4863,
+      "step": 8934
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1614735223605527e-05,
+      "loss": 0.2393,
+      "step": 8935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1613179380851505e-05,
+      "loss": 1.4219,
+      "step": 8936
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1611623498003721e-05,
+      "loss": 0.1387,
+      "step": 8937
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1610067575100847e-05,
+      "loss": 1.0859,
+      "step": 8938
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1608511612181559e-05,
+      "loss": 1.0156,
+      "step": 8939
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1606955609284522e-05,
+      "loss": 1.4688,
+      "step": 8940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1605399566448414e-05,
+      "loss": 0.7773,
+      "step": 8941
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1603843483711903e-05,
+      "loss": 0.7812,
+      "step": 8942
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1602287361113666e-05,
+      "loss": 1.1016,
+      "step": 8943
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1600731198692381e-05,
+      "loss": 0.8828,
+      "step": 8944
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1599174996486725e-05,
+      "loss": 1.5312,
+      "step": 8945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1597618754535372e-05,
+      "loss": 0.7266,
+      "step": 8946
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1596062472876998e-05,
+      "loss": 1.4375,
+      "step": 8947
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1594506151550292e-05,
+      "loss": 1.2891,
+      "step": 8948
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1592949790593928e-05,
+      "loss": 1.4375,
+      "step": 8949
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1591393390046588e-05,
+      "loss": 0.7227,
+      "step": 8950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1589836949946956e-05,
+      "loss": 0.5938,
+      "step": 8951
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1588280470333717e-05,
+      "loss": 1.0078,
+      "step": 8952
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1586723951245555e-05,
+      "loss": 1.0391,
+      "step": 8953
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1585167392721151e-05,
+      "loss": 0.6445,
+      "step": 8954
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1583610794799195e-05,
+      "loss": 0.8867,
+      "step": 8955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1582054157518375e-05,
+      "loss": 0.3906,
+      "step": 8956
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1580497480917381e-05,
+      "loss": 0.9961,
+      "step": 8957
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.15789407650349e-05,
+      "loss": 0.7852,
+      "step": 8958
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1577384009909618e-05,
+      "loss": 0.8945,
+      "step": 8959
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1575827215580233e-05,
+      "loss": 0.6445,
+      "step": 8960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1574270382085437e-05,
+      "loss": 0.875,
+      "step": 8961
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1572713509463925e-05,
+      "loss": 0.8633,
+      "step": 8962
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.157115659775438e-05,
+      "loss": 0.9805,
+      "step": 8963
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1569599646995513e-05,
+      "loss": 0.8047,
+      "step": 8964
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1568042657226008e-05,
+      "loss": 0.7227,
+      "step": 8965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.156648562848457e-05,
+      "loss": 0.5391,
+      "step": 8966
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.1564928560809889e-05,
+      "loss": 0.9766,
+      "step": 8967
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1563371454240674e-05,
+      "loss": 0.7422,
+      "step": 8968
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1561814308815622e-05,
+      "loss": 1.0,
+      "step": 8969
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1560257124573428e-05,
+      "loss": 0.3867,
+      "step": 8970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.15586999015528e-05,
+      "loss": 0.9648,
+      "step": 8971
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1557142639792438e-05,
+      "loss": 1.4141,
+      "step": 8972
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.155558533933105e-05,
+      "loss": 1.1016,
+      "step": 8973
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.155402800020734e-05,
+      "loss": 1.2109,
+      "step": 8974
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1552470622460008e-05,
+      "loss": 2.2188,
+      "step": 8975
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1550913206127767e-05,
+      "loss": 0.3125,
+      "step": 8976
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1549355751249325e-05,
+      "loss": 0.5312,
+      "step": 8977
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1547798257863388e-05,
+      "loss": 1.3672,
+      "step": 8978
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1546240726008664e-05,
+      "loss": 1.0703,
+      "step": 8979
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.154468315572387e-05,
+      "loss": 0.6055,
+      "step": 8980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1543125547047713e-05,
+      "loss": 0.918,
+      "step": 8981
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1541567900018906e-05,
+      "loss": 0.5742,
+      "step": 8982
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.154001021467616e-05,
+      "loss": 0.6406,
+      "step": 8983
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1538452491058197e-05,
+      "loss": 1.2266,
+      "step": 8984
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1536894729203729e-05,
+      "loss": 0.957,
+      "step": 8985
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.153533692915147e-05,
+      "loss": 0.6719,
+      "step": 8986
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1533779090940138e-05,
+      "loss": 0.1309,
+      "step": 8987
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1532221214608452e-05,
+      "loss": 1.0625,
+      "step": 8988
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1530663300195133e-05,
+      "loss": 1.1094,
+      "step": 8989
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.15291053477389e-05,
+      "loss": 0.8008,
+      "step": 8990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1527547357278473e-05,
+      "loss": 1.1875,
+      "step": 8991
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1525989328852574e-05,
+      "loss": 1.0703,
+      "step": 8992
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.152443126249993e-05,
+      "loss": 0.7109,
+      "step": 8993
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.152287315825926e-05,
+      "loss": 1.125,
+      "step": 8994
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1521315016169288e-05,
+      "loss": 1.0625,
+      "step": 8995
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.151975683626875e-05,
+      "loss": 1.6719,
+      "step": 8996
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1518198618596361e-05,
+      "loss": 1.3438,
+      "step": 8997
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1516640363190853e-05,
+      "loss": 1.1719,
+      "step": 8998
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1515082070090955e-05,
+      "loss": 0.7383,
+      "step": 8999
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1513523739335399e-05,
+      "loss": 0.9805,
+      "step": 9000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1511965370962913e-05,
+      "loss": 0.6797,
+      "step": 9001
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1510406965012228e-05,
+      "loss": 0.9648,
+      "step": 9002
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1508848521522076e-05,
+      "loss": 1.0156,
+      "step": 9003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1507290040531192e-05,
+      "loss": 0.8672,
+      "step": 9004
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1505731522078309e-05,
+      "loss": 0.4492,
+      "step": 9005
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1504172966202166e-05,
+      "loss": 0.832,
+      "step": 9006
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1502614372941492e-05,
+      "loss": 1.0391,
+      "step": 9007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1501055742335034e-05,
+      "loss": 0.6953,
+      "step": 9008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1499497074421522e-05,
+      "loss": 0.75,
+      "step": 9009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1497938369239696e-05,
+      "loss": 1.1719,
+      "step": 9010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1496379626828296e-05,
+      "loss": 0.459,
+      "step": 9011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1494820847226064e-05,
+      "loss": 0.8672,
+      "step": 9012
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1493262030471746e-05,
+      "loss": 1.6562,
+      "step": 9013
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1491703176604076e-05,
+      "loss": 0.9453,
+      "step": 9014
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1490144285661805e-05,
+      "loss": 1.0312,
+      "step": 9015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1488585357683668e-05,
+      "loss": 1.1562,
+      "step": 9016
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1487026392708423e-05,
+      "loss": 1.0859,
+      "step": 9017
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.148546739077481e-05,
+      "loss": 0.8867,
+      "step": 9018
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1483908351921573e-05,
+      "loss": 1.1875,
+      "step": 9019
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1482349276187463e-05,
+      "loss": 0.8945,
+      "step": 9020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.148079016361123e-05,
+      "loss": 0.5664,
+      "step": 9021
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1479231014231624e-05,
+      "loss": 1.7891,
+      "step": 9022
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.14776718280874e-05,
+      "loss": 0.7695,
+      "step": 9023
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1476112605217297e-05,
+      "loss": 1.4844,
+      "step": 9024
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1474553345660082e-05,
+      "loss": 0.6836,
+      "step": 9025
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.14729940494545e-05,
+      "loss": 1.0625,
+      "step": 9026
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.147143471663931e-05,
+      "loss": 0.7773,
+      "step": 9027
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1469875347253261e-05,
+      "loss": 0.3516,
+      "step": 9028
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.146831594133512e-05,
+      "loss": 0.3105,
+      "step": 9029
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1466756498923634e-05,
+      "loss": 0.2754,
+      "step": 9030
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1465197020057566e-05,
+      "loss": 0.8906,
+      "step": 9031
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1463637504775672e-05,
+      "loss": 0.3809,
+      "step": 9032
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.146207795311672e-05,
+      "loss": 0.9141,
+      "step": 9033
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1460518365119462e-05,
+      "loss": 0.5898,
+      "step": 9034
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1458958740822666e-05,
+      "loss": 0.457,
+      "step": 9035
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1457399080265088e-05,
+      "loss": 0.8008,
+      "step": 9036
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1455839383485498e-05,
+      "loss": 0.377,
+      "step": 9037
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1454279650522658e-05,
+      "loss": 0.9961,
+      "step": 9038
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1452719881415333e-05,
+      "loss": 0.208,
+      "step": 9039
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1451160076202288e-05,
+      "loss": 0.5859,
+      "step": 9040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1449600234922295e-05,
+      "loss": 1.1172,
+      "step": 9041
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.144804035761412e-05,
+      "loss": 1.0,
+      "step": 9042
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1446480444316526e-05,
+      "loss": 0.3359,
+      "step": 9043
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.144492049506829e-05,
+      "loss": 1.0312,
+      "step": 9044
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1443360509908181e-05,
+      "loss": 0.7773,
+      "step": 9045
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1441800488874972e-05,
+      "loss": 0.8594,
+      "step": 9046
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.1440240432007432e-05,
+      "loss": 0.4668,
+      "step": 9047
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1438680339344336e-05,
+      "loss": 0.4102,
+      "step": 9048
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1437120210924459e-05,
+      "loss": 0.6328,
+      "step": 9049
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1435560046786578e-05,
+      "loss": 0.9727,
+      "step": 9050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1433999846969469e-05,
+      "loss": 0.5391,
+      "step": 9051
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.14324396115119e-05,
+      "loss": 0.7695,
+      "step": 9052
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.143087934045266e-05,
+      "loss": 0.8516,
+      "step": 9053
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1429319033830526e-05,
+      "loss": 0.9023,
+      "step": 9054
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1427758691684272e-05,
+      "loss": 1.4609,
+      "step": 9055
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1426198314052682e-05,
+      "loss": 0.5469,
+      "step": 9056
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.142463790097454e-05,
+      "loss": 0.918,
+      "step": 9057
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1423077452488625e-05,
+      "loss": 0.9062,
+      "step": 9058
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1421516968633722e-05,
+      "loss": 1.0078,
+      "step": 9059
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.141995644944861e-05,
+      "loss": 1.2891,
+      "step": 9060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.141839589497208e-05,
+      "loss": 0.9219,
+      "step": 9061
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.141683530524292e-05,
+      "loss": 0.6719,
+      "step": 9062
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1415274680299908e-05,
+      "loss": 0.6641,
+      "step": 9063
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1413714020181834e-05,
+      "loss": 0.4707,
+      "step": 9064
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1412153324927494e-05,
+      "loss": 1.0859,
+      "step": 9065
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1410592594575667e-05,
+      "loss": 0.5547,
+      "step": 9066
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1409031829165152e-05,
+      "loss": 0.9336,
+      "step": 9067
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1407471028734732e-05,
+      "loss": 1.0859,
+      "step": 9068
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1405910193323206e-05,
+      "loss": 0.6211,
+      "step": 9069
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1404349322969364e-05,
+      "loss": 0.8906,
+      "step": 9070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1402788417711998e-05,
+      "loss": 0.9297,
+      "step": 9071
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1401227477589904e-05,
+      "loss": 0.3477,
+      "step": 9072
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1399666502641877e-05,
+      "loss": 0.2793,
+      "step": 9073
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.139810549290672e-05,
+      "loss": 0.6797,
+      "step": 9074
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1396544448423219e-05,
+      "loss": 1.0391,
+      "step": 9075
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1394983369230175e-05,
+      "loss": 0.1348,
+      "step": 9076
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1393422255366392e-05,
+      "loss": 1.7109,
+      "step": 9077
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1391861106870664e-05,
+      "loss": 0.9141,
+      "step": 9078
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1390299923781799e-05,
+      "loss": 1.0156,
+      "step": 9079
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1388738706138587e-05,
+      "loss": 0.707,
+      "step": 9080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1387177453979838e-05,
+      "loss": 1.1875,
+      "step": 9081
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1385616167344358e-05,
+      "loss": 0.7109,
+      "step": 9082
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1384054846270944e-05,
+      "loss": 0.377,
+      "step": 9083
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1382493490798404e-05,
+      "loss": 0.5352,
+      "step": 9084
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1380932100965545e-05,
+      "loss": 0.5703,
+      "step": 9085
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1379370676811173e-05,
+      "loss": 0.8125,
+      "step": 9086
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1377809218374094e-05,
+      "loss": 1.3047,
+      "step": 9087
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1376247725693114e-05,
+      "loss": 0.9883,
+      "step": 9088
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1374686198807049e-05,
+      "loss": 0.7539,
+      "step": 9089
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1373124637754705e-05,
+      "loss": 1.2969,
+      "step": 9090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1371563042574892e-05,
+      "loss": 0.7773,
+      "step": 9091
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1370001413306422e-05,
+      "loss": 0.9219,
+      "step": 9092
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1368439749988104e-05,
+      "loss": 1.6641,
+      "step": 9093
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1366878052658764e-05,
+      "loss": 0.2178,
+      "step": 9094
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1365316321357205e-05,
+      "loss": 1.0938,
+      "step": 9095
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1363754556122243e-05,
+      "loss": 0.7461,
+      "step": 9096
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1362192756992694e-05,
+      "loss": 1.0,
+      "step": 9097
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.136063092400738e-05,
+      "loss": 0.7188,
+      "step": 9098
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1359069057205115e-05,
+      "loss": 0.5117,
+      "step": 9099
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1357507156624718e-05,
+      "loss": 1.3906,
+      "step": 9100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1355945222305002e-05,
+      "loss": 0.9297,
+      "step": 9101
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.13543832542848e-05,
+      "loss": 1.0938,
+      "step": 9102
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1352821252602926e-05,
+      "loss": 0.918,
+      "step": 9103
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1351259217298199e-05,
+      "loss": 0.7109,
+      "step": 9104
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1349697148409442e-05,
+      "loss": 1.5,
+      "step": 9105
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1348135045975483e-05,
+      "loss": 1.1484,
+      "step": 9106
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.134657291003515e-05,
+      "loss": 1.4766,
+      "step": 9107
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1345010740627255e-05,
+      "loss": 1.1328,
+      "step": 9108
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.134344853779063e-05,
+      "loss": 0.4531,
+      "step": 9109
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.134188630156411e-05,
+      "loss": 0.8945,
+      "step": 9110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1340324031986511e-05,
+      "loss": 0.6367,
+      "step": 9111
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.133876172909667e-05,
+      "loss": 0.8086,
+      "step": 9112
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1337199392933407e-05,
+      "loss": 0.8906,
+      "step": 9113
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1335637023535559e-05,
+      "loss": 0.4434,
+      "step": 9114
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1334074620941957e-05,
+      "loss": 1.0859,
+      "step": 9115
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1332512185191431e-05,
+      "loss": 1.2891,
+      "step": 9116
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1330949716322812e-05,
+      "loss": 0.9844,
+      "step": 9117
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.132938721437494e-05,
+      "loss": 0.7383,
+      "step": 9118
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.132782467938664e-05,
+      "loss": 0.9336,
+      "step": 9119
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1326262111396753e-05,
+      "loss": 0.7539,
+      "step": 9120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1324699510444113e-05,
+      "loss": 0.875,
+      "step": 9121
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1323136876567555e-05,
+      "loss": 1.1562,
+      "step": 9122
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1321574209805923e-05,
+      "loss": 1.4844,
+      "step": 9123
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1320011510198048e-05,
+      "loss": 1.2891,
+      "step": 9124
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.131844877778277e-05,
+      "loss": 0.7031,
+      "step": 9125
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.1316886012598936e-05,
+      "loss": 0.3066,
+      "step": 9126
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1315323214685379e-05,
+      "loss": 1.1406,
+      "step": 9127
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1313760384080944e-05,
+      "loss": 0.5,
+      "step": 9128
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.131219752082447e-05,
+      "loss": 0.9102,
+      "step": 9129
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1310634624954806e-05,
+      "loss": 0.9453,
+      "step": 9130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1309071696510792e-05,
+      "loss": 0.8828,
+      "step": 9131
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1307508735531273e-05,
+      "loss": 1.5312,
+      "step": 9132
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1305945742055093e-05,
+      "loss": 0.2637,
+      "step": 9133
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1304382716121107e-05,
+      "loss": 0.9258,
+      "step": 9134
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1302819657768153e-05,
+      "loss": 1.6953,
+      "step": 9135
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1301256567035081e-05,
+      "loss": 1.5703,
+      "step": 9136
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1299693443960742e-05,
+      "loss": 0.2598,
+      "step": 9137
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1298130288583984e-05,
+      "loss": 1.25,
+      "step": 9138
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1296567100943658e-05,
+      "loss": 0.1768,
+      "step": 9139
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1295003881078621e-05,
+      "loss": 1.5,
+      "step": 9140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1293440629027712e-05,
+      "loss": 2.1094,
+      "step": 9141
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1291877344829794e-05,
+      "loss": 0.8008,
+      "step": 9142
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1290314028523719e-05,
+      "loss": 0.5938,
+      "step": 9143
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.128875068014834e-05,
+      "loss": 1.0625,
+      "step": 9144
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.128718729974251e-05,
+      "loss": 1.0391,
+      "step": 9145
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1285623887345092e-05,
+      "loss": 0.957,
+      "step": 9146
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1284060442994938e-05,
+      "loss": 0.4707,
+      "step": 9147
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1282496966730905e-05,
+      "loss": 0.5,
+      "step": 9148
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.128093345859185e-05,
+      "loss": 0.498,
+      "step": 9149
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1279369918616638e-05,
+      "loss": 1.4453,
+      "step": 9150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1277806346844129e-05,
+      "loss": 0.8359,
+      "step": 9151
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1276242743313177e-05,
+      "loss": 0.7422,
+      "step": 9152
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1274679108062647e-05,
+      "loss": 1.0234,
+      "step": 9153
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1273115441131403e-05,
+      "loss": 0.7031,
+      "step": 9154
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1271551742558306e-05,
+      "loss": 1.4141,
+      "step": 9155
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1269988012382225e-05,
+      "loss": 0.9844,
+      "step": 9156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1268424250642015e-05,
+      "loss": 0.9297,
+      "step": 9157
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1266860457376552e-05,
+      "loss": 1.4062,
+      "step": 9158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1265296632624696e-05,
+      "loss": 0.9961,
+      "step": 9159
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1263732776425315e-05,
+      "loss": 0.5977,
+      "step": 9160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1262168888817277e-05,
+      "loss": 0.7109,
+      "step": 9161
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1260604969839454e-05,
+      "loss": 0.4512,
+      "step": 9162
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1259041019530712e-05,
+      "loss": 0.2217,
+      "step": 9163
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1257477037929921e-05,
+      "loss": 0.9102,
+      "step": 9164
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1255913025075954e-05,
+      "loss": 0.918,
+      "step": 9165
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1254348981007679e-05,
+      "loss": 0.7422,
+      "step": 9166
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1252784905763974e-05,
+      "loss": 0.7656,
+      "step": 9167
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1251220799383708e-05,
+      "loss": 0.832,
+      "step": 9168
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.124965666190576e-05,
+      "loss": 0.5234,
+      "step": 9169
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1248092493368997e-05,
+      "loss": 0.918,
+      "step": 9170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.12465282938123e-05,
+      "loss": 0.7656,
+      "step": 9171
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1244964063274547e-05,
+      "loss": 0.8125,
+      "step": 9172
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1243399801794612e-05,
+      "loss": 1.3047,
+      "step": 9173
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1241835509411371e-05,
+      "loss": 0.7891,
+      "step": 9174
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1240271186163709e-05,
+      "loss": 1.4844,
+      "step": 9175
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1238706832090503e-05,
+      "loss": 1.2812,
+      "step": 9176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.123714244723063e-05,
+      "loss": 0.291,
+      "step": 9177
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1235578031622972e-05,
+      "loss": 0.6484,
+      "step": 9178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1234013585306416e-05,
+      "loss": 1.3672,
+      "step": 9179
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1232449108319839e-05,
+      "loss": 0.2598,
+      "step": 9180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1230884600702126e-05,
+      "loss": 0.4922,
+      "step": 9181
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1229320062492157e-05,
+      "loss": 1.1094,
+      "step": 9182
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1227755493728825e-05,
+      "loss": 1.25,
+      "step": 9183
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1226190894451012e-05,
+      "loss": 0.2695,
+      "step": 9184
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1224626264697605e-05,
+      "loss": 0.8086,
+      "step": 9185
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1223061604507483e-05,
+      "loss": 0.4062,
+      "step": 9186
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1221496913919548e-05,
+      "loss": 1.2969,
+      "step": 9187
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1219932192972677e-05,
+      "loss": 0.3457,
+      "step": 9188
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1218367441705768e-05,
+      "loss": 1.1328,
+      "step": 9189
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1216802660157704e-05,
+      "loss": 1.0859,
+      "step": 9190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.121523784836738e-05,
+      "loss": 1.1953,
+      "step": 9191
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1213673006373686e-05,
+      "loss": 0.4395,
+      "step": 9192
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1212108134215516e-05,
+      "loss": 0.8594,
+      "step": 9193
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.121054323193176e-05,
+      "loss": 1.375,
+      "step": 9194
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1208978299561315e-05,
+      "loss": 1.0078,
+      "step": 9195
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1207413337143077e-05,
+      "loss": 0.4434,
+      "step": 9196
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1205848344715935e-05,
+      "loss": 0.9453,
+      "step": 9197
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1204283322318787e-05,
+      "loss": 0.5742,
+      "step": 9198
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1202718269990535e-05,
+      "loss": 0.8047,
+      "step": 9199
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1201153187770078e-05,
+      "loss": 0.3281,
+      "step": 9200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1199588075696306e-05,
+      "loss": 1.0,
+      "step": 9201
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1198022933808119e-05,
+      "loss": 0.6953,
+      "step": 9202
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1196457762144424e-05,
+      "loss": 1.5469,
+      "step": 9203
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1194892560744115e-05,
+      "loss": 1.5078,
+      "step": 9204
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1193327329646096e-05,
+      "loss": 1.0625,
+      "step": 9205
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1191762068889267e-05,
+      "loss": 0.7969,
+      "step": 9206
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1190196778512538e-05,
+      "loss": 1.6719,
+      "step": 9207
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1188631458554804e-05,
+      "loss": 0.7656,
+      "step": 9208
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1187066109054974e-05,
+      "loss": 0.4766,
+      "step": 9209
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1185500730051946e-05,
+      "loss": 1.0469,
+      "step": 9210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1183935321584636e-05,
+      "loss": 1.1875,
+      "step": 9211
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1182369883691945e-05,
+      "loss": 1.2031,
+      "step": 9212
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1180804416412783e-05,
+      "loss": 1.2578,
+      "step": 9213
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1179238919786049e-05,
+      "loss": 1.0703,
+      "step": 9214
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1177673393850663e-05,
+      "loss": 0.8789,
+      "step": 9215
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.117610783864553e-05,
+      "loss": 1.0078,
+      "step": 9216
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1174542254209559e-05,
+      "loss": 0.6055,
+      "step": 9217
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.117297664058166e-05,
+      "loss": 0.9492,
+      "step": 9218
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.117141099780075e-05,
+      "loss": 0.9883,
+      "step": 9219
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1169845325905734e-05,
+      "loss": 1.1719,
+      "step": 9220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.116827962493553e-05,
+      "loss": 0.7305,
+      "step": 9221
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1166713894929049e-05,
+      "loss": 1.8047,
+      "step": 9222
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.116514813592521e-05,
+      "loss": 0.3594,
+      "step": 9223
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1163582347962922e-05,
+      "loss": 0.957,
+      "step": 9224
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1162016531081106e-05,
+      "loss": 0.2012,
+      "step": 9225
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1160450685318671e-05,
+      "loss": 0.7656,
+      "step": 9226
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1158884810714545e-05,
+      "loss": 1.2344,
+      "step": 9227
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1157318907307642e-05,
+      "loss": 0.2363,
+      "step": 9228
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1155752975136878e-05,
+      "loss": 0.7969,
+      "step": 9229
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1154187014241172e-05,
+      "loss": 0.166,
+      "step": 9230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1152621024659446e-05,
+      "loss": 0.3691,
+      "step": 9231
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1151055006430624e-05,
+      "loss": 0.6875,
+      "step": 9232
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1149488959593627e-05,
+      "loss": 0.9453,
+      "step": 9233
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.114792288418737e-05,
+      "loss": 0.7969,
+      "step": 9234
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1146356780250783e-05,
+      "loss": 0.7852,
+      "step": 9235
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1144790647822787e-05,
+      "loss": 1.4219,
+      "step": 9236
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1143224486942308e-05,
+      "loss": 0.9375,
+      "step": 9237
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1141658297648273e-05,
+      "loss": 1.0391,
+      "step": 9238
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.11400920799796e-05,
+      "loss": 2.0156,
+      "step": 9239
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1138525833975226e-05,
+      "loss": 1.6172,
+      "step": 9240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1136959559674072e-05,
+      "loss": 1.4766,
+      "step": 9241
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1135393257115068e-05,
+      "loss": 1.1016,
+      "step": 9242
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1133826926337138e-05,
+      "loss": 1.5312,
+      "step": 9243
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1132260567379218e-05,
+      "loss": 1.8125,
+      "step": 9244
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.113069418028024e-05,
+      "loss": 1.4766,
+      "step": 9245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1129127765079127e-05,
+      "loss": 1.0156,
+      "step": 9246
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1127561321814808e-05,
+      "loss": 1.0625,
+      "step": 9247
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1125994850526226e-05,
+      "loss": 0.707,
+      "step": 9248
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.112442835125231e-05,
+      "loss": 1.0859,
+      "step": 9249
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1122861824031993e-05,
+      "loss": 0.3848,
+      "step": 9250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1121295268904203e-05,
+      "loss": 0.625,
+      "step": 9251
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1119728685907885e-05,
+      "loss": 0.4609,
+      "step": 9252
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1118162075081969e-05,
+      "loss": 0.9805,
+      "step": 9253
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1116595436465394e-05,
+      "loss": 1.3281,
+      "step": 9254
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1115028770097091e-05,
+      "loss": 0.9961,
+      "step": 9255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1113462076016009e-05,
+      "loss": 1.25,
+      "step": 9256
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1111895354261077e-05,
+      "loss": 1.0,
+      "step": 9257
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1110328604871236e-05,
+      "loss": 1.25,
+      "step": 9258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1108761827885426e-05,
+      "loss": 1.0469,
+      "step": 9259
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1107195023342589e-05,
+      "loss": 0.875,
+      "step": 9260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.110562819128167e-05,
+      "loss": 0.8047,
+      "step": 9261
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.11040613317416e-05,
+      "loss": 0.6836,
+      "step": 9262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.110249444476133e-05,
+      "loss": 0.8867,
+      "step": 9263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.11009275303798e-05,
+      "loss": 0.6836,
+      "step": 9264
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1099360588635955e-05,
+      "loss": 0.625,
+      "step": 9265
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1097793619568743e-05,
+      "loss": 1.2656,
+      "step": 9266
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1096226623217102e-05,
+      "loss": 0.8086,
+      "step": 9267
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1094659599619982e-05,
+      "loss": 0.7969,
+      "step": 9268
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1093092548816328e-05,
+      "loss": 1.1406,
+      "step": 9269
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.109152547084509e-05,
+      "loss": 0.4258,
+      "step": 9270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1089958365745212e-05,
+      "loss": 0.957,
+      "step": 9271
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1088391233555648e-05,
+      "loss": 1.0234,
+      "step": 9272
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1086824074315346e-05,
+      "loss": 1.5234,
+      "step": 9273
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1085256888063252e-05,
+      "loss": 1.0,
+      "step": 9274
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1083689674838319e-05,
+      "loss": 0.8242,
+      "step": 9275
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1082122434679496e-05,
+      "loss": 0.7539,
+      "step": 9276
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1080555167625742e-05,
+      "loss": 1.1094,
+      "step": 9277
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1078987873716004e-05,
+      "loss": 0.6836,
+      "step": 9278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1077420552989234e-05,
+      "loss": 0.2344,
+      "step": 9279
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1075853205484387e-05,
+      "loss": 0.7656,
+      "step": 9280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1074285831240423e-05,
+      "loss": 0.8789,
+      "step": 9281
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1072718430296292e-05,
+      "loss": 0.7461,
+      "step": 9282
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.107115100269095e-05,
+      "loss": 0.5312,
+      "step": 9283
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1069583548463359e-05,
+      "loss": 0.3945,
+      "step": 9284
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1068016067652467e-05,
+      "loss": 0.8867,
+      "step": 9285
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.106644856029724e-05,
+      "loss": 0.625,
+      "step": 9286
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1064881026436631e-05,
+      "loss": 0.5039,
+      "step": 9287
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1063313466109603e-05,
+      "loss": 0.4629,
+      "step": 9288
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.106174587935512e-05,
+      "loss": 0.8672,
+      "step": 9289
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1060178266212136e-05,
+      "loss": 0.4336,
+      "step": 9290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.105861062671961e-05,
+      "loss": 1.1719,
+      "step": 9291
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.105704296091651e-05,
+      "loss": 0.4609,
+      "step": 9292
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1055475268841796e-05,
+      "loss": 0.2617,
+      "step": 9293
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1053907550534433e-05,
+      "loss": 0.4844,
+      "step": 9294
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.105233980603338e-05,
+      "loss": 0.7422,
+      "step": 9295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1050772035377608e-05,
+      "loss": 0.2139,
+      "step": 9296
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.104920423860608e-05,
+      "loss": 1.25,
+      "step": 9297
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1047636415757761e-05,
+      "loss": 1.0156,
+      "step": 9298
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1046068566871613e-05,
+      "loss": 1.0859,
+      "step": 9299
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1044500691986615e-05,
+      "loss": 0.9336,
+      "step": 9300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1042932791141722e-05,
+      "loss": 0.7734,
+      "step": 9301
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.104136486437591e-05,
+      "loss": 0.7031,
+      "step": 9302
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1039796911728143e-05,
+      "loss": 0.7148,
+      "step": 9303
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1038228933237397e-05,
+      "loss": 0.6211,
+      "step": 9304
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.103666092894264e-05,
+      "loss": 0.8984,
+      "step": 9305
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.103509289888284e-05,
+      "loss": 0.8672,
+      "step": 9306
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1033524843096967e-05,
+      "loss": 0.6328,
+      "step": 9307
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1031956761624001e-05,
+      "loss": 1.0469,
+      "step": 9308
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1030388654502908e-05,
+      "loss": 0.4629,
+      "step": 9309
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1028820521772668e-05,
+      "loss": 1.5938,
+      "step": 9310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1027252363472245e-05,
+      "loss": 0.7305,
+      "step": 9311
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1025684179640628e-05,
+      "loss": 1.3828,
+      "step": 9312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.102411597031678e-05,
+      "loss": 1.1094,
+      "step": 9313
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1022547735539684e-05,
+      "loss": 0.6523,
+      "step": 9314
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1020979475348313e-05,
+      "loss": 1.1875,
+      "step": 9315
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1019411189781643e-05,
+      "loss": 0.5898,
+      "step": 9316
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1017842878878662e-05,
+      "loss": 0.8203,
+      "step": 9317
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1016274542678338e-05,
+      "loss": 0.6016,
+      "step": 9318
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1014706181219653e-05,
+      "loss": 1.0312,
+      "step": 9319
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1013137794541585e-05,
+      "loss": 0.9883,
+      "step": 9320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1011569382683118e-05,
+      "loss": 0.7227,
+      "step": 9321
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1010000945683237e-05,
+      "loss": 0.918,
+      "step": 9322
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1008432483580915e-05,
+      "loss": 1.0938,
+      "step": 9323
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1006863996415136e-05,
+      "loss": 1.0078,
+      "step": 9324
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1005295484224888e-05,
+      "loss": 0.9961,
+      "step": 9325
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1003726947049151e-05,
+      "loss": 0.8438,
+      "step": 9326
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1002158384926913e-05,
+      "loss": 0.5391,
+      "step": 9327
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.100058979789715e-05,
+      "loss": 0.2207,
+      "step": 9328
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0999021185998857e-05,
+      "loss": 1.0469,
+      "step": 9329
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0997452549271015e-05,
+      "loss": 0.6289,
+      "step": 9330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0995883887752614e-05,
+      "loss": 1.5,
+      "step": 9331
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0994315201482635e-05,
+      "loss": 0.2266,
+      "step": 9332
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0992746490500078e-05,
+      "loss": 0.8984,
+      "step": 9333
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0991177754843919e-05,
+      "loss": 0.75,
+      "step": 9334
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0989608994553151e-05,
+      "loss": 0.2871,
+      "step": 9335
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0988040209666763e-05,
+      "loss": 0.252,
+      "step": 9336
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.098647140022375e-05,
+      "loss": 0.3945,
+      "step": 9337
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0984902566263105e-05,
+      "loss": 0.5703,
+      "step": 9338
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.098333370782381e-05,
+      "loss": 0.6836,
+      "step": 9339
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0981764824944861e-05,
+      "loss": 1.1328,
+      "step": 9340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0980195917665253e-05,
+      "loss": 1.125,
+      "step": 9341
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0978626986023977e-05,
+      "loss": 0.5898,
+      "step": 9342
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.097705803006003e-05,
+      "loss": 1.5859,
+      "step": 9343
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0975489049812402e-05,
+      "loss": 1.2266,
+      "step": 9344
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0973920045320096e-05,
+      "loss": 0.4746,
+      "step": 9345
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0972351016622101e-05,
+      "loss": 1.0469,
+      "step": 9346
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0970781963757416e-05,
+      "loss": 0.4395,
+      "step": 9347
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0969212886765033e-05,
+      "loss": 1.75,
+      "step": 9348
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.096764378568396e-05,
+      "loss": 1.6719,
+      "step": 9349
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0966074660553188e-05,
+      "loss": 1.2969,
+      "step": 9350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0964505511411716e-05,
+      "loss": 1.2734,
+      "step": 9351
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0962936338298542e-05,
+      "loss": 0.9883,
+      "step": 9352
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0961367141252672e-05,
+      "loss": 1.0078,
+      "step": 9353
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0959797920313103e-05,
+      "loss": 1.1562,
+      "step": 9354
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0958228675518838e-05,
+      "loss": 0.5352,
+      "step": 9355
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0956659406908874e-05,
+      "loss": 1.6484,
+      "step": 9356
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0955090114522218e-05,
+      "loss": 0.4727,
+      "step": 9357
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0953520798397871e-05,
+      "loss": 0.9258,
+      "step": 9358
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.095195145857484e-05,
+      "loss": 0.875,
+      "step": 9359
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0950382095092122e-05,
+      "loss": 1.0312,
+      "step": 9360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0948812707988732e-05,
+      "loss": 0.9297,
+      "step": 9361
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0947243297303667e-05,
+      "loss": 1.2344,
+      "step": 9362
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0945673863075936e-05,
+      "loss": 1.0391,
+      "step": 9363
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0944104405344542e-05,
+      "loss": 1.2891,
+      "step": 9364
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.0942534924148497e-05,
+      "loss": 0.5664,
+      "step": 9365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.094096541952681e-05,
+      "loss": 0.9336,
+      "step": 9366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0939395891518484e-05,
+      "loss": 0.3945,
+      "step": 9367
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0937826340162528e-05,
+      "loss": 1.1484,
+      "step": 9368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0936256765497954e-05,
+      "loss": 0.9336,
+      "step": 9369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0934687167563773e-05,
+      "loss": 0.7383,
+      "step": 9370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0933117546398994e-05,
+      "loss": 0.8594,
+      "step": 9371
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0931547902042627e-05,
+      "loss": 1.3672,
+      "step": 9372
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0929978234533684e-05,
+      "loss": 0.6016,
+      "step": 9373
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0928408543911183e-05,
+      "loss": 1.4531,
+      "step": 9374
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0926838830214126e-05,
+      "loss": 0.3574,
+      "step": 9375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0925269093481535e-05,
+      "loss": 0.9375,
+      "step": 9376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0923699333752422e-05,
+      "loss": 0.1768,
+      "step": 9377
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0922129551065803e-05,
+      "loss": 1.1797,
+      "step": 9378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0920559745460691e-05,
+      "loss": 0.2852,
+      "step": 9379
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.09189899169761e-05,
+      "loss": 1.2812,
+      "step": 9380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0917420065651048e-05,
+      "loss": 0.1895,
+      "step": 9381
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0915850191524556e-05,
+      "loss": 0.707,
+      "step": 9382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0914280294635637e-05,
+      "loss": 0.416,
+      "step": 9383
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0912710375023306e-05,
+      "loss": 0.5039,
+      "step": 9384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0911140432726587e-05,
+      "loss": 0.4355,
+      "step": 9385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.09095704677845e-05,
+      "loss": 1.5938,
+      "step": 9386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0908000480236061e-05,
+      "loss": 0.4492,
+      "step": 9387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0906430470120296e-05,
+      "loss": 0.2422,
+      "step": 9388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0904860437476215e-05,
+      "loss": 0.375,
+      "step": 9389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0903290382342846e-05,
+      "loss": 1.0469,
+      "step": 9390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0901720304759213e-05,
+      "loss": 0.9453,
+      "step": 9391
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0900150204764338e-05,
+      "loss": 1.75,
+      "step": 9392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0898580082397237e-05,
+      "loss": 1.1562,
+      "step": 9393
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0897009937696945e-05,
+      "loss": 1.0391,
+      "step": 9394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0895439770702479e-05,
+      "loss": 0.9531,
+      "step": 9395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0893869581452862e-05,
+      "loss": 1.5547,
+      "step": 9396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0892299369987123e-05,
+      "loss": 0.9219,
+      "step": 9397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0890729136344286e-05,
+      "loss": 1.2422,
+      "step": 9398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0889158880563384e-05,
+      "loss": 0.8711,
+      "step": 9399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0887588602683435e-05,
+      "loss": 0.625,
+      "step": 9400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0886018302743468e-05,
+      "loss": 0.8594,
+      "step": 9401
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0884447980782515e-05,
+      "loss": 0.6211,
+      "step": 9402
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0882877636839601e-05,
+      "loss": 0.2324,
+      "step": 9403
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.088130727095376e-05,
+      "loss": 0.373,
+      "step": 9404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0879736883164014e-05,
+      "loss": 0.4648,
+      "step": 9405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0878166473509399e-05,
+      "loss": 0.8633,
+      "step": 9406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0876596042028944e-05,
+      "loss": 0.3184,
+      "step": 9407
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0875025588761684e-05,
+      "loss": 0.7109,
+      "step": 9408
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.087345511374664e-05,
+      "loss": 1.4453,
+      "step": 9409
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.087188461702286e-05,
+      "loss": 1.3594,
+      "step": 9410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0870314098629367e-05,
+      "loss": 1.4062,
+      "step": 9411
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0868743558605192e-05,
+      "loss": 1.2812,
+      "step": 9412
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0867172996989375e-05,
+      "loss": 1.3516,
+      "step": 9413
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0865602413820949e-05,
+      "loss": 0.8906,
+      "step": 9414
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0864031809138949e-05,
+      "loss": 0.7422,
+      "step": 9415
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0862461182982413e-05,
+      "loss": 1.1094,
+      "step": 9416
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0860890535390372e-05,
+      "loss": 0.7773,
+      "step": 9417
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0859319866401865e-05,
+      "loss": 0.9648,
+      "step": 9418
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0857749176055929e-05,
+      "loss": 0.6875,
+      "step": 9419
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0856178464391605e-05,
+      "loss": 0.8008,
+      "step": 9420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0854607731447924e-05,
+      "loss": 0.8047,
+      "step": 9421
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0853036977263935e-05,
+      "loss": 1.3516,
+      "step": 9422
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.085146620187867e-05,
+      "loss": 1.2109,
+      "step": 9423
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0849895405331168e-05,
+      "loss": 0.9141,
+      "step": 9424
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0848324587660473e-05,
+      "loss": 0.7812,
+      "step": 9425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0846753748905624e-05,
+      "loss": 0.6211,
+      "step": 9426
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0845182889105668e-05,
+      "loss": 0.1836,
+      "step": 9427
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0843612008299637e-05,
+      "loss": 1.3203,
+      "step": 9428
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0842041106526578e-05,
+      "loss": 0.8906,
+      "step": 9429
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0840470183825538e-05,
+      "loss": 1.5859,
+      "step": 9430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0838899240235553e-05,
+      "loss": 0.582,
+      "step": 9431
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0837328275795678e-05,
+      "loss": 0.2109,
+      "step": 9432
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0835757290544947e-05,
+      "loss": 0.498,
+      "step": 9433
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0834186284522408e-05,
+      "loss": 0.7812,
+      "step": 9434
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0832615257767108e-05,
+      "loss": 0.7734,
+      "step": 9435
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0831044210318092e-05,
+      "loss": 1.6641,
+      "step": 9436
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0829473142214404e-05,
+      "loss": 1.0156,
+      "step": 9437
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.08279020534951e-05,
+      "loss": 1.6172,
+      "step": 9438
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0826330944199221e-05,
+      "loss": 1.2656,
+      "step": 9439
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0824759814365815e-05,
+      "loss": 1.3359,
+      "step": 9440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.082318866403393e-05,
+      "loss": 0.957,
+      "step": 9441
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0821617493242618e-05,
+      "loss": 1.0703,
+      "step": 9442
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0820046302030928e-05,
+      "loss": 0.793,
+      "step": 9443
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0818475090437913e-05,
+      "loss": 1.0469,
+      "step": 9444
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.0816903858502615e-05,
+      "loss": 0.7773,
+      "step": 9445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0815332606264093e-05,
+      "loss": 0.9688,
+      "step": 9446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0813761333761394e-05,
+      "loss": 0.7891,
+      "step": 9447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0812190041033577e-05,
+      "loss": 1.6406,
+      "step": 9448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0810618728119686e-05,
+      "loss": 0.9727,
+      "step": 9449
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0809047395058782e-05,
+      "loss": 1.4141,
+      "step": 9450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0807476041889915e-05,
+      "loss": 0.9688,
+      "step": 9451
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0805904668652135e-05,
+      "loss": 0.8398,
+      "step": 9452
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0804333275384504e-05,
+      "loss": 0.7969,
+      "step": 9453
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0802761862126076e-05,
+      "loss": 0.9492,
+      "step": 9454
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0801190428915906e-05,
+      "loss": 0.6719,
+      "step": 9455
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0799618975793046e-05,
+      "loss": 0.8477,
+      "step": 9456
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0798047502796556e-05,
+      "loss": 0.6328,
+      "step": 9457
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0796476009965493e-05,
+      "loss": 1.2188,
+      "step": 9458
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0794904497338917e-05,
+      "loss": 1.0469,
+      "step": 9459
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0793332964955885e-05,
+      "loss": 0.9961,
+      "step": 9460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0791761412855453e-05,
+      "loss": 1.3047,
+      "step": 9461
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.079018984107668e-05,
+      "loss": 1.1562,
+      "step": 9462
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0788618249658629e-05,
+      "loss": 1.4531,
+      "step": 9463
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.078704663864036e-05,
+      "loss": 1.2031,
+      "step": 9464
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0785475008060933e-05,
+      "loss": 0.5352,
+      "step": 9465
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0783903357959405e-05,
+      "loss": 0.7852,
+      "step": 9466
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0782331688374842e-05,
+      "loss": 0.8008,
+      "step": 9467
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0780759999346305e-05,
+      "loss": 1.1797,
+      "step": 9468
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0779188290912856e-05,
+      "loss": 0.8516,
+      "step": 9469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0777616563113558e-05,
+      "loss": 1.3828,
+      "step": 9470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0776044815987479e-05,
+      "loss": 0.7422,
+      "step": 9471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0774473049573677e-05,
+      "loss": 1.2422,
+      "step": 9472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.077290126391122e-05,
+      "loss": 1.4375,
+      "step": 9473
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0771329459039168e-05,
+      "loss": 1.1484,
+      "step": 9474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0769757634996591e-05,
+      "loss": 0.2422,
+      "step": 9475
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0768185791822559e-05,
+      "loss": 0.3125,
+      "step": 9476
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0766613929556128e-05,
+      "loss": 0.875,
+      "step": 9477
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.076504204823637e-05,
+      "loss": 0.2832,
+      "step": 9478
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0763470147902357e-05,
+      "loss": 1.4062,
+      "step": 9479
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0761898228593148e-05,
+      "loss": 0.4727,
+      "step": 9480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.076032629034782e-05,
+      "loss": 0.5938,
+      "step": 9481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0758754333205433e-05,
+      "loss": 0.6328,
+      "step": 9482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0757182357205065e-05,
+      "loss": 1.375,
+      "step": 9483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.075561036238578e-05,
+      "loss": 1.4375,
+      "step": 9484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0754038348786648e-05,
+      "loss": 1.2422,
+      "step": 9485
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.075246631644674e-05,
+      "loss": 0.5508,
+      "step": 9486
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0750894265405129e-05,
+      "loss": 0.8281,
+      "step": 9487
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0749322195700892e-05,
+      "loss": 1.0391,
+      "step": 9488
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.074775010737309e-05,
+      "loss": 0.2969,
+      "step": 9489
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0746178000460796e-05,
+      "loss": 0.707,
+      "step": 9490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0744605875003094e-05,
+      "loss": 0.4062,
+      "step": 9491
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0743033731039048e-05,
+      "loss": 0.9062,
+      "step": 9492
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0741461568607736e-05,
+      "loss": 1.0234,
+      "step": 9493
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0739889387748227e-05,
+      "loss": 0.7383,
+      "step": 9494
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0738317188499603e-05,
+      "loss": 0.3496,
+      "step": 9495
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0736744970900937e-05,
+      "loss": 1.0312,
+      "step": 9496
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0735172734991302e-05,
+      "loss": 0.8555,
+      "step": 9497
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0733600480809774e-05,
+      "loss": 0.5547,
+      "step": 9498
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0732028208395435e-05,
+      "loss": 1.4375,
+      "step": 9499
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0730455917787356e-05,
+      "loss": 0.9961,
+      "step": 9500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0728883609024619e-05,
+      "loss": 0.5195,
+      "step": 9501
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0727311282146295e-05,
+      "loss": 0.5625,
+      "step": 9502
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0725738937191472e-05,
+      "loss": 1.0625,
+      "step": 9503
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0724166574199224e-05,
+      "loss": 1.0625,
+      "step": 9504
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.072259419320863e-05,
+      "loss": 0.9531,
+      "step": 9505
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0721021794258768e-05,
+      "loss": 0.3047,
+      "step": 9506
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0719449377388721e-05,
+      "loss": 0.7773,
+      "step": 9507
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0717876942637573e-05,
+      "loss": 1.7812,
+      "step": 9508
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.07163044900444e-05,
+      "loss": 1.5781,
+      "step": 9509
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0714732019648281e-05,
+      "loss": 0.5938,
+      "step": 9510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0713159531488304e-05,
+      "loss": 0.6562,
+      "step": 9511
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.071158702560355e-05,
+      "loss": 1.1328,
+      "step": 9512
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0710014502033101e-05,
+      "loss": 1.2656,
+      "step": 9513
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0708441960816038e-05,
+      "loss": 1.0703,
+      "step": 9514
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0706869401991448e-05,
+      "loss": 0.9922,
+      "step": 9515
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0705296825598419e-05,
+      "loss": 1.0781,
+      "step": 9516
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0703724231676026e-05,
+      "loss": 0.6992,
+      "step": 9517
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0702151620263359e-05,
+      "loss": 0.7227,
+      "step": 9518
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0700578991399503e-05,
+      "loss": 0.9961,
+      "step": 9519
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0699006345123547e-05,
+      "loss": 0.6094,
+      "step": 9520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0697433681474576e-05,
+      "loss": 0.6133,
+      "step": 9521
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0695861000491672e-05,
+      "loss": 0.7188,
+      "step": 9522
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0694288302213926e-05,
+      "loss": 1.6875,
+      "step": 9523
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.0692715586680426e-05,
+      "loss": 0.6719,
+      "step": 9524
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.069114285393026e-05,
+      "loss": 1.1719,
+      "step": 9525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0689570104002513e-05,
+      "loss": 1.0156,
+      "step": 9526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0687997336936283e-05,
+      "loss": 0.9258,
+      "step": 9527
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.068642455277065e-05,
+      "loss": 0.7695,
+      "step": 9528
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0684851751544705e-05,
+      "loss": 0.5859,
+      "step": 9529
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.068327893329754e-05,
+      "loss": 1.3359,
+      "step": 9530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0681706098068248e-05,
+      "loss": 0.9375,
+      "step": 9531
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0680133245895918e-05,
+      "loss": 0.3027,
+      "step": 9532
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0678560376819642e-05,
+      "loss": 0.2812,
+      "step": 9533
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0676987490878508e-05,
+      "loss": 1.1875,
+      "step": 9534
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0675414588111614e-05,
+      "loss": 0.5117,
+      "step": 9535
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.067384166855805e-05,
+      "loss": 0.1562,
+      "step": 9536
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0672268732256912e-05,
+      "loss": 1.1328,
+      "step": 9537
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0670695779247288e-05,
+      "loss": 0.21,
+      "step": 9538
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0669122809568273e-05,
+      "loss": 0.2559,
+      "step": 9539
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0667549823258967e-05,
+      "loss": 0.6484,
+      "step": 9540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0665976820358459e-05,
+      "loss": 0.7617,
+      "step": 9541
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.066440380090585e-05,
+      "loss": 0.8125,
+      "step": 9542
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0662830764940228e-05,
+      "loss": 0.9336,
+      "step": 9543
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0661257712500695e-05,
+      "loss": 1.1328,
+      "step": 9544
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0659684643626345e-05,
+      "loss": 0.3164,
+      "step": 9545
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0658111558356274e-05,
+      "loss": 1.0156,
+      "step": 9546
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.065653845672958e-05,
+      "loss": 0.5781,
+      "step": 9547
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0654965338785364e-05,
+      "loss": 0.8125,
+      "step": 9548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0653392204562723e-05,
+      "loss": 1.0703,
+      "step": 9549
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0651819054100752e-05,
+      "loss": 0.5312,
+      "step": 9550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0650245887438548e-05,
+      "loss": 1.125,
+      "step": 9551
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0648672704615216e-05,
+      "loss": 1.2188,
+      "step": 9552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0647099505669856e-05,
+      "loss": 0.8477,
+      "step": 9553
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0645526290641568e-05,
+      "loss": 0.8047,
+      "step": 9554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0643953059569443e-05,
+      "loss": 1.5703,
+      "step": 9555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0642379812492593e-05,
+      "loss": 1.1953,
+      "step": 9556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0640806549450118e-05,
+      "loss": 0.3262,
+      "step": 9557
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0639233270481113e-05,
+      "loss": 0.3438,
+      "step": 9558
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0637659975624684e-05,
+      "loss": 0.8359,
+      "step": 9559
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0636086664919938e-05,
+      "loss": 0.8203,
+      "step": 9560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0634513338405971e-05,
+      "loss": 1.4141,
+      "step": 9561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0632939996121889e-05,
+      "loss": 0.6641,
+      "step": 9562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0631366638106795e-05,
+      "loss": 1.0469,
+      "step": 9563
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0629793264399795e-05,
+      "loss": 1.0391,
+      "step": 9564
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0628219875039993e-05,
+      "loss": 1.2812,
+      "step": 9565
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.062664647006649e-05,
+      "loss": 1.0859,
+      "step": 9566
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0625073049518394e-05,
+      "loss": 0.8594,
+      "step": 9567
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0623499613434811e-05,
+      "loss": 0.6602,
+      "step": 9568
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0621926161854847e-05,
+      "loss": 0.6133,
+      "step": 9569
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0620352694817611e-05,
+      "loss": 0.4766,
+      "step": 9570
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.06187792123622e-05,
+      "loss": 0.6445,
+      "step": 9571
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0617205714527732e-05,
+      "loss": 1.2031,
+      "step": 9572
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.061563220135331e-05,
+      "loss": 1.1562,
+      "step": 9573
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0614058672878043e-05,
+      "loss": 0.418,
+      "step": 9574
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0612485129141035e-05,
+      "loss": 0.4434,
+      "step": 9575
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0610911570181403e-05,
+      "loss": 1.1484,
+      "step": 9576
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.060933799603825e-05,
+      "loss": 0.5312,
+      "step": 9577
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0607764406750685e-05,
+      "loss": 0.3301,
+      "step": 9578
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0606190802357819e-05,
+      "loss": 0.2266,
+      "step": 9579
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0604617182898764e-05,
+      "loss": 0.9219,
+      "step": 9580
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0603043548412628e-05,
+      "loss": 0.2383,
+      "step": 9581
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0601469898938529e-05,
+      "loss": 0.5586,
+      "step": 9582
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0599896234515563e-05,
+      "loss": 0.3262,
+      "step": 9583
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0598322555182856e-05,
+      "loss": 1.3906,
+      "step": 9584
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0596748860979513e-05,
+      "loss": 0.4004,
+      "step": 9585
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.059517515194465e-05,
+      "loss": 0.6914,
+      "step": 9586
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0593601428117378e-05,
+      "loss": 0.3477,
+      "step": 9587
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0592027689536812e-05,
+      "loss": 1.375,
+      "step": 9588
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0590453936242062e-05,
+      "loss": 0.5195,
+      "step": 9589
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0588880168272244e-05,
+      "loss": 0.5039,
+      "step": 9590
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.058730638566647e-05,
+      "loss": 0.334,
+      "step": 9591
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0585732588463857e-05,
+      "loss": 0.5156,
+      "step": 9592
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0584158776703523e-05,
+      "loss": 0.957,
+      "step": 9593
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.058258495042458e-05,
+      "loss": 0.5625,
+      "step": 9594
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0581011109666138e-05,
+      "loss": 0.9688,
+      "step": 9595
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0579437254467322e-05,
+      "loss": 1.1094,
+      "step": 9596
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0577863384867244e-05,
+      "loss": 1.1797,
+      "step": 9597
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0576289500905026e-05,
+      "loss": 0.3359,
+      "step": 9598
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0574715602619774e-05,
+      "loss": 1.8672,
+      "step": 9599
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0573141690050614e-05,
+      "loss": 0.4023,
+      "step": 9600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0571567763236666e-05,
+      "loss": 0.8516,
+      "step": 9601
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0569993822217041e-05,
+      "loss": 1.1094,
+      "step": 9602
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0568419867030859e-05,
+      "loss": 0.373,
+      "step": 9603
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0566845897717246e-05,
+      "loss": 0.8555,
+      "step": 9604
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0565271914315313e-05,
+      "loss": 1.1172,
+      "step": 9605
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0563697916864183e-05,
+      "loss": 0.6172,
+      "step": 9606
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0562123905402975e-05,
+      "loss": 0.7852,
+      "step": 9607
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.056054987997081e-05,
+      "loss": 1.2812,
+      "step": 9608
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0558975840606808e-05,
+      "loss": 0.8086,
+      "step": 9609
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0557401787350092e-05,
+      "loss": 0.6562,
+      "step": 9610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.055582772023978e-05,
+      "loss": 1.0781,
+      "step": 9611
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0554253639314995e-05,
+      "loss": 1.1406,
+      "step": 9612
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0552679544614859e-05,
+      "loss": 0.7539,
+      "step": 9613
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0551105436178496e-05,
+      "loss": 1.3594,
+      "step": 9614
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.054953131404503e-05,
+      "loss": 0.5664,
+      "step": 9615
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0547957178253576e-05,
+      "loss": 0.6641,
+      "step": 9616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0546383028843266e-05,
+      "loss": 1.3125,
+      "step": 9617
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.054480886585322e-05,
+      "loss": 0.8633,
+      "step": 9618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0543234689322564e-05,
+      "loss": 0.6016,
+      "step": 9619
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0541660499290416e-05,
+      "loss": 0.7344,
+      "step": 9620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0540086295795912e-05,
+      "loss": 0.8477,
+      "step": 9621
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.053851207887817e-05,
+      "loss": 1.7266,
+      "step": 9622
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0536937848576313e-05,
+      "loss": 0.6484,
+      "step": 9623
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.053536360492947e-05,
+      "loss": 0.4492,
+      "step": 9624
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0533789347976766e-05,
+      "loss": 0.1836,
+      "step": 9625
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0532215077757333e-05,
+      "loss": 0.793,
+      "step": 9626
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.053064079431029e-05,
+      "loss": 0.418,
+      "step": 9627
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0529066497674765e-05,
+      "loss": 0.3535,
+      "step": 9628
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0527492187889889e-05,
+      "loss": 0.2002,
+      "step": 9629
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0525917864994791e-05,
+      "loss": 0.6016,
+      "step": 9630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0524343529028596e-05,
+      "loss": 0.668,
+      "step": 9631
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0522769180030428e-05,
+      "loss": 0.6406,
+      "step": 9632
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0521194818039422e-05,
+      "loss": 0.8008,
+      "step": 9633
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0519620443094707e-05,
+      "loss": 0.9922,
+      "step": 9634
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0518046055235409e-05,
+      "loss": 1.6406,
+      "step": 9635
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0516471654500657e-05,
+      "loss": 0.2852,
+      "step": 9636
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0514897240929587e-05,
+      "loss": 0.4121,
+      "step": 9637
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0513322814561324e-05,
+      "loss": 1.25,
+      "step": 9638
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0511748375435e-05,
+      "loss": 0.957,
+      "step": 9639
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0510173923589744e-05,
+      "loss": 1.2188,
+      "step": 9640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.050859945906469e-05,
+      "loss": 1.2188,
+      "step": 9641
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0507024981898972e-05,
+      "loss": 0.9453,
+      "step": 9642
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0505450492131717e-05,
+      "loss": 0.5,
+      "step": 9643
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0503875989802054e-05,
+      "loss": 0.8516,
+      "step": 9644
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0502301474949126e-05,
+      "loss": 1.0156,
+      "step": 9645
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0500726947612057e-05,
+      "loss": 1.1094,
+      "step": 9646
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0499152407829987e-05,
+      "loss": 0.3477,
+      "step": 9647
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0497577855642036e-05,
+      "loss": 0.248,
+      "step": 9648
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0496003291087357e-05,
+      "loss": 0.8008,
+      "step": 9649
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0494428714205071e-05,
+      "loss": 0.4492,
+      "step": 9650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0492854125034314e-05,
+      "loss": 0.4961,
+      "step": 9651
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0491279523614221e-05,
+      "loss": 1.5547,
+      "step": 9652
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0489704909983933e-05,
+      "loss": 0.6406,
+      "step": 9653
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.048813028418258e-05,
+      "loss": 0.6953,
+      "step": 9654
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0486555646249295e-05,
+      "loss": 0.6211,
+      "step": 9655
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0484980996223216e-05,
+      "loss": 0.9492,
+      "step": 9656
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0483406334143483e-05,
+      "loss": 0.3594,
+      "step": 9657
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0481831660049226e-05,
+      "loss": 1.0312,
+      "step": 9658
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.048025697397959e-05,
+      "loss": 1.2578,
+      "step": 9659
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0478682275973705e-05,
+      "loss": 1.3281,
+      "step": 9660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0477107566070708e-05,
+      "loss": 0.9141,
+      "step": 9661
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0475532844309742e-05,
+      "loss": 0.4277,
+      "step": 9662
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0473958110729942e-05,
+      "loss": 0.6289,
+      "step": 9663
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0472383365370444e-05,
+      "loss": 1.0781,
+      "step": 9664
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0470808608270394e-05,
+      "loss": 1.5391,
+      "step": 9665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0469233839468923e-05,
+      "loss": 0.8086,
+      "step": 9666
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0467659059005172e-05,
+      "loss": 0.8164,
+      "step": 9667
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.046608426691828e-05,
+      "loss": 0.4863,
+      "step": 9668
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0464509463247388e-05,
+      "loss": 0.6875,
+      "step": 9669
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0462934648031639e-05,
+      "loss": 0.7344,
+      "step": 9670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0461359821310168e-05,
+      "loss": 0.6602,
+      "step": 9671
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0459784983122116e-05,
+      "loss": 1.6094,
+      "step": 9672
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.045821013350663e-05,
+      "loss": 1.2109,
+      "step": 9673
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0456635272502844e-05,
+      "loss": 1.3594,
+      "step": 9674
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0455060400149906e-05,
+      "loss": 0.75,
+      "step": 9675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0453485516486947e-05,
+      "loss": 0.5156,
+      "step": 9676
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0451910621553119e-05,
+      "loss": 0.3496,
+      "step": 9677
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.045033571538756e-05,
+      "loss": 0.373,
+      "step": 9678
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0448760798029416e-05,
+      "loss": 0.5312,
+      "step": 9679
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0447185869517823e-05,
+      "loss": 0.2314,
+      "step": 9680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0445610929891932e-05,
+      "loss": 0.3945,
+      "step": 9681
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0444035979190881e-05,
+      "loss": 0.1797,
+      "step": 9682
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0442461017453817e-05,
+      "loss": 0.4004,
+      "step": 9683
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0440886044719877e-05,
+      "loss": 0.9336,
+      "step": 9684
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0439311061028216e-05,
+      "loss": 0.4746,
+      "step": 9685
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0437736066417971e-05,
+      "loss": 1.0469,
+      "step": 9686
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0436161060928291e-05,
+      "loss": 1.7188,
+      "step": 9687
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0434586044598316e-05,
+      "loss": 0.208,
+      "step": 9688
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0433011017467192e-05,
+      "loss": 1.1562,
+      "step": 9689
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.043143597957407e-05,
+      "loss": 1.0859,
+      "step": 9690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0429860930958089e-05,
+      "loss": 0.2852,
+      "step": 9691
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0428285871658403e-05,
+      "loss": 1.0781,
+      "step": 9692
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0426710801714149e-05,
+      "loss": 2.0,
+      "step": 9693
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0425135721164478e-05,
+      "loss": 1.0312,
+      "step": 9694
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.042356063004854e-05,
+      "loss": 1.1172,
+      "step": 9695
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0421985528405477e-05,
+      "loss": 0.9336,
+      "step": 9696
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0420410416274435e-05,
+      "loss": 1.1016,
+      "step": 9697
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0418835293694571e-05,
+      "loss": 1.0859,
+      "step": 9698
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0417260160705025e-05,
+      "loss": 0.6328,
+      "step": 9699
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0415685017344948e-05,
+      "loss": 0.918,
+      "step": 9700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0414109863653482e-05,
+      "loss": 0.5195,
+      "step": 9701
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0412534699669788e-05,
+      "loss": 1.375,
+      "step": 9702
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0410959525433006e-05,
+      "loss": 1.6328,
+      "step": 9703
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.040938434098229e-05,
+      "loss": 1.25,
+      "step": 9704
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.040780914635678e-05,
+      "loss": 1.3672,
+      "step": 9705
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0406233941595638e-05,
+      "loss": 0.7227,
+      "step": 9706
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0404658726738006e-05,
+      "loss": 0.7695,
+      "step": 9707
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0403083501823039e-05,
+      "loss": 0.6836,
+      "step": 9708
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0401508266889882e-05,
+      "loss": 0.5273,
+      "step": 9709
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.039993302197769e-05,
+      "loss": 1.6016,
+      "step": 9710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0398357767125613e-05,
+      "loss": 0.5977,
+      "step": 9711
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0396782502372802e-05,
+      "loss": 0.498,
+      "step": 9712
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0395207227758405e-05,
+      "loss": 0.9844,
+      "step": 9713
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0393631943321586e-05,
+      "loss": 1.0,
+      "step": 9714
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0392056649101481e-05,
+      "loss": 0.9023,
+      "step": 9715
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0390481345137248e-05,
+      "loss": 0.4355,
+      "step": 9716
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0388906031468043e-05,
+      "loss": 1.6484,
+      "step": 9717
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0387330708133015e-05,
+      "loss": 0.9102,
+      "step": 9718
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0385755375171318e-05,
+      "loss": 1.3203,
+      "step": 9719
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0384180032622108e-05,
+      "loss": 0.4316,
+      "step": 9720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0382604680524532e-05,
+      "loss": 1.0781,
+      "step": 9721
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0381029318917748e-05,
+      "loss": 1.8516,
+      "step": 9722
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0379453947840909e-05,
+      "loss": 0.7188,
+      "step": 9723
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.037787856733317e-05,
+      "loss": 0.1719,
+      "step": 9724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0376303177433683e-05,
+      "loss": 0.7578,
+      "step": 9725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0374727778181608e-05,
+      "loss": 0.5859,
+      "step": 9726
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.037315236961609e-05,
+      "loss": 0.459,
+      "step": 9727
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0371576951776295e-05,
+      "loss": 0.6602,
+      "step": 9728
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0370001524701368e-05,
+      "loss": 0.4434,
+      "step": 9729
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.036842608843047e-05,
+      "loss": 1.4531,
+      "step": 9730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.036685064300276e-05,
+      "loss": 1.0859,
+      "step": 9731
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0365275188457386e-05,
+      "loss": 1.0078,
+      "step": 9732
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0363699724833509e-05,
+      "loss": 0.3438,
+      "step": 9733
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0362124252170285e-05,
+      "loss": 0.5312,
+      "step": 9734
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0360548770506871e-05,
+      "loss": 0.7539,
+      "step": 9735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0358973279882423e-05,
+      "loss": 0.3398,
+      "step": 9736
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0357397780336096e-05,
+      "loss": 0.4863,
+      "step": 9737
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.035582227190705e-05,
+      "loss": 0.832,
+      "step": 9738
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0354246754634442e-05,
+      "loss": 0.7969,
+      "step": 9739
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.035267122855743e-05,
+      "loss": 0.4258,
+      "step": 9740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.035109569371517e-05,
+      "loss": 0.5742,
+      "step": 9741
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0349520150146823e-05,
+      "loss": 0.9961,
+      "step": 9742
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0347944597891547e-05,
+      "loss": 0.6055,
+      "step": 9743
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0346369036988497e-05,
+      "loss": 1.3906,
+      "step": 9744
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0344793467476832e-05,
+      "loss": 1.3516,
+      "step": 9745
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0343217889395717e-05,
+      "loss": 1.8281,
+      "step": 9746
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.034164230278431e-05,
+      "loss": 1.0078,
+      "step": 9747
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0340066707681763e-05,
+      "loss": 1.0234,
+      "step": 9748
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0338491104127242e-05,
+      "loss": 0.918,
+      "step": 9749
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0336915492159907e-05,
+      "loss": 1.1797,
+      "step": 9750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0335339871818914e-05,
+      "loss": 0.5742,
+      "step": 9751
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0333764243143427e-05,
+      "loss": 0.4863,
+      "step": 9752
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0332188606172604e-05,
+      "loss": 0.7305,
+      "step": 9753
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.033061296094561e-05,
+      "loss": 0.8945,
+      "step": 9754
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0329037307501602e-05,
+      "loss": 0.7812,
+      "step": 9755
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.032746164587974e-05,
+      "loss": 1.0469,
+      "step": 9756
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0325885976119187e-05,
+      "loss": 0.9805,
+      "step": 9757
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0324310298259106e-05,
+      "loss": 0.7656,
+      "step": 9758
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0322734612338659e-05,
+      "loss": 1.2031,
+      "step": 9759
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0321158918397005e-05,
+      "loss": 0.4648,
+      "step": 9760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0319583216473307e-05,
+      "loss": 0.7031,
+      "step": 9761
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0318007506606727e-05,
+      "loss": 0.3828,
+      "step": 9762
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0316431788836427e-05,
+      "loss": 0.8828,
+      "step": 9763
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0314856063201575e-05,
+      "loss": 1.2656,
+      "step": 9764
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.031328032974133e-05,
+      "loss": 0.7695,
+      "step": 9765
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0311704588494849e-05,
+      "loss": 1.1016,
+      "step": 9766
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0310128839501303e-05,
+      "loss": 0.7461,
+      "step": 9767
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0308553082799857e-05,
+      "loss": 1.0469,
+      "step": 9768
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0306977318429672e-05,
+      "loss": 1.1641,
+      "step": 9769
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0305401546429905e-05,
+      "loss": 0.832,
+      "step": 9770
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.030382576683973e-05,
+      "loss": 0.793,
+      "step": 9771
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0302249979698307e-05,
+      "loss": 1.1328,
+      "step": 9772
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0300674185044801e-05,
+      "loss": 0.4648,
+      "step": 9773
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0299098382918375e-05,
+      "loss": 1.2891,
+      "step": 9774
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0297522573358199e-05,
+      "loss": 1.5156,
+      "step": 9775
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0295946756403432e-05,
+      "loss": 1.2031,
+      "step": 9776
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.029437093209324e-05,
+      "loss": 1.1562,
+      "step": 9777
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0292795100466788e-05,
+      "loss": 0.1924,
+      "step": 9778
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0291219261563245e-05,
+      "loss": 0.5977,
+      "step": 9779
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.028964341542178e-05,
+      "loss": 0.3164,
+      "step": 9780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.028806756208155e-05,
+      "loss": 1.625,
+      "step": 9781
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0286491701581723e-05,
+      "loss": 0.9453,
+      "step": 9782
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0284915833961467e-05,
+      "loss": 0.543,
+      "step": 9783
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0283339959259951e-05,
+      "loss": 0.6953,
+      "step": 9784
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0281764077516336e-05,
+      "loss": 0.1279,
+      "step": 9785
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0280188188769792e-05,
+      "loss": 0.8242,
+      "step": 9786
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0278612293059491e-05,
+      "loss": 0.3691,
+      "step": 9787
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.027703639042459e-05,
+      "loss": 0.918,
+      "step": 9788
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0275460480904264e-05,
+      "loss": 1.1094,
+      "step": 9789
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0273884564537676e-05,
+      "loss": 1.9688,
+      "step": 9790
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0272308641363995e-05,
+      "loss": 0.625,
+      "step": 9791
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.027073271142239e-05,
+      "loss": 1.0938,
+      "step": 9792
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0269156774752029e-05,
+      "loss": 0.9844,
+      "step": 9793
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0267580831392077e-05,
+      "loss": 1.1328,
+      "step": 9794
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0266004881381704e-05,
+      "loss": 1.75,
+      "step": 9795
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0264428924760081e-05,
+      "loss": 0.8906,
+      "step": 9796
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0262852961566376e-05,
+      "loss": 0.8867,
+      "step": 9797
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0261276991839752e-05,
+      "loss": 1.4609,
+      "step": 9798
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0259701015619386e-05,
+      "loss": 0.832,
+      "step": 9799
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0258125032944441e-05,
+      "loss": 0.8086,
+      "step": 9800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0256549043854089e-05,
+      "loss": 1.2578,
+      "step": 9801
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.02549730483875e-05,
+      "loss": 0.5703,
+      "step": 9802
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0253397046583843e-05,
+      "loss": 0.6523,
+      "step": 9803
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0251821038482289e-05,
+      "loss": 0.6367,
+      "step": 9804
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0250245024122003e-05,
+      "loss": 1.1875,
+      "step": 9805
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0248669003542157e-05,
+      "loss": 0.8672,
+      "step": 9806
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0247092976781927e-05,
+      "loss": 0.8516,
+      "step": 9807
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.024551694388048e-05,
+      "loss": 1.1719,
+      "step": 9808
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0243940904876983e-05,
+      "loss": 0.5742,
+      "step": 9809
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0242364859810607e-05,
+      "loss": 0.5352,
+      "step": 9810
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.024078880872053e-05,
+      "loss": 1.0625,
+      "step": 9811
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0239212751645915e-05,
+      "loss": 0.6523,
+      "step": 9812
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0237636688625938e-05,
+      "loss": 1.3984,
+      "step": 9813
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0236060619699766e-05,
+      "loss": 1.0547,
+      "step": 9814
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0234484544906576e-05,
+      "loss": 1.0547,
+      "step": 9815
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0232908464285533e-05,
+      "loss": 0.8594,
+      "step": 9816
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0231332377875813e-05,
+      "loss": 0.8359,
+      "step": 9817
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0229756285716588e-05,
+      "loss": 1.2578,
+      "step": 9818
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0228180187847027e-05,
+      "loss": 1.1875,
+      "step": 9819
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0226604084306308e-05,
+      "loss": 0.8906,
+      "step": 9820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0225027975133599e-05,
+      "loss": 0.8906,
+      "step": 9821
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.022345186036807e-05,
+      "loss": 1.4609,
+      "step": 9822
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0221875740048896e-05,
+      "loss": 0.8125,
+      "step": 9823
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0220299614215252e-05,
+      "loss": 1.1016,
+      "step": 9824
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0218723482906311e-05,
+      "loss": 0.6133,
+      "step": 9825
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.021714734616124e-05,
+      "loss": 1.6484,
+      "step": 9826
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0215571204019219e-05,
+      "loss": 0.4336,
+      "step": 9827
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0213995056519417e-05,
+      "loss": 0.6328,
+      "step": 9828
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0212418903701006e-05,
+      "loss": 1.2812,
+      "step": 9829
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0210842745603165e-05,
+      "loss": 0.7852,
+      "step": 9830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0209266582265065e-05,
+      "loss": 1.0391,
+      "step": 9831
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0207690413725879e-05,
+      "loss": 0.2773,
+      "step": 9832
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0206114240024782e-05,
+      "loss": 0.8711,
+      "step": 9833
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0204538061200949e-05,
+      "loss": 0.8477,
+      "step": 9834
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0202961877293547e-05,
+      "loss": 0.6953,
+      "step": 9835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0201385688341761e-05,
+      "loss": 0.625,
+      "step": 9836
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0199809494384759e-05,
+      "loss": 0.9414,
+      "step": 9837
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0198233295461716e-05,
+      "loss": 1.3359,
+      "step": 9838
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0196657091611806e-05,
+      "loss": 0.9961,
+      "step": 9839
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0195080882874207e-05,
+      "loss": 0.4316,
+      "step": 9840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0193504669288095e-05,
+      "loss": 0.9023,
+      "step": 9841
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.019192845089264e-05,
+      "loss": 0.8789,
+      "step": 9842
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0190352227727015e-05,
+      "loss": 1.4375,
+      "step": 9843
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0188775999830404e-05,
+      "loss": 1.8203,
+      "step": 9844
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0187199767241978e-05,
+      "loss": 1.0625,
+      "step": 9845
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0185623530000915e-05,
+      "loss": 1.2422,
+      "step": 9846
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0184047288146382e-05,
+      "loss": 1.9453,
+      "step": 9847
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0182471041717564e-05,
+      "loss": 0.7344,
+      "step": 9848
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.018089479075363e-05,
+      "loss": 0.6367,
+      "step": 9849
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0179318535293762e-05,
+      "loss": 1.2266,
+      "step": 9850
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0177742275377134e-05,
+      "loss": 0.7578,
+      "step": 9851
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0176166011042922e-05,
+      "loss": 1.2578,
+      "step": 9852
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0174589742330302e-05,
+      "loss": 0.6953,
+      "step": 9853
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.017301346927845e-05,
+      "loss": 1.4062,
+      "step": 9854
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0171437191926541e-05,
+      "loss": 0.832,
+      "step": 9855
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0169860910313755e-05,
+      "loss": 1.0078,
+      "step": 9856
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0168284624479267e-05,
+      "loss": 0.6094,
+      "step": 9857
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.016670833446226e-05,
+      "loss": 1.25,
+      "step": 9858
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0165132040301895e-05,
+      "loss": 0.4766,
+      "step": 9859
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0163555742037363e-05,
+      "loss": 0.6523,
+      "step": 9860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0161979439707837e-05,
+      "loss": 1.1562,
+      "step": 9861
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0160403133352495e-05,
+      "loss": 1.6719,
+      "step": 9862
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.015882682301051e-05,
+      "loss": 0.5039,
+      "step": 9863
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.015725050872107e-05,
+      "loss": 1.3438,
+      "step": 9864
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0155674190523342e-05,
+      "loss": 0.8438,
+      "step": 9865
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0154097868456504e-05,
+      "loss": 0.2988,
+      "step": 9866
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0152521542559738e-05,
+      "loss": 0.8906,
+      "step": 9867
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0150945212872223e-05,
+      "loss": 0.6602,
+      "step": 9868
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0149368879433137e-05,
+      "loss": 0.9141,
+      "step": 9869
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0147792542281651e-05,
+      "loss": 1.2344,
+      "step": 9870
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0146216201456949e-05,
+      "loss": 0.4648,
+      "step": 9871
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0144639856998206e-05,
+      "loss": 1.2812,
+      "step": 9872
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0143063508944607e-05,
+      "loss": 1.4297,
+      "step": 9873
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0141487157335324e-05,
+      "loss": 0.8633,
+      "step": 9874
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0139910802209532e-05,
+      "loss": 0.3301,
+      "step": 9875
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0138334443606417e-05,
+      "loss": 0.3223,
+      "step": 9876
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0136758081565157e-05,
+      "loss": 0.6406,
+      "step": 9877
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0135181716124927e-05,
+      "loss": 1.0625,
+      "step": 9878
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0133605347324906e-05,
+      "loss": 1.125,
+      "step": 9879
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0132028975204278e-05,
+      "loss": 1.1875,
+      "step": 9880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0130452599802218e-05,
+      "loss": 0.1367,
+      "step": 9881
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0128876221157904e-05,
+      "loss": 0.5742,
+      "step": 9882
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0127299839310513e-05,
+      "loss": 0.3203,
+      "step": 9883
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0125723454299231e-05,
+      "loss": 0.6445,
+      "step": 9884
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.012414706616324e-05,
+      "loss": 0.8398,
+      "step": 9885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0122570674941705e-05,
+      "loss": 0.2451,
+      "step": 9886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0120994280673815e-05,
+      "loss": 1.3281,
+      "step": 9887
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.011941788339875e-05,
+      "loss": 0.1738,
+      "step": 9888
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0117841483155688e-05,
+      "loss": 1.1562,
+      "step": 9889
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0116265079983808e-05,
+      "loss": 0.8281,
+      "step": 9890
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0114688673922288e-05,
+      "loss": 0.5547,
+      "step": 9891
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0113112265010315e-05,
+      "loss": 1.2031,
+      "step": 9892
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0111535853287063e-05,
+      "loss": 1.5078,
+      "step": 9893
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.010995943879171e-05,
+      "loss": 0.3926,
+      "step": 9894
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0108383021563438e-05,
+      "loss": 0.3945,
+      "step": 9895
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.010680660164143e-05,
+      "loss": 0.7344,
+      "step": 9896
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0105230179064867e-05,
+      "loss": 1.625,
+      "step": 9897
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0103653753872922e-05,
+      "loss": 0.1377,
+      "step": 9898
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.010207732610478e-05,
+      "loss": 0.9219,
+      "step": 9899
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0100500895799622e-05,
+      "loss": 0.7461,
+      "step": 9900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0098924462996627e-05,
+      "loss": 1.5078,
+      "step": 9901
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0097348027734976e-05,
+      "loss": 0.9766,
+      "step": 9902
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0095771590053849e-05,
+      "loss": 1.0234,
+      "step": 9903
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0094195149992425e-05,
+      "loss": 0.6992,
+      "step": 9904
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0092618707589889e-05,
+      "loss": 0.8281,
+      "step": 9905
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.009104226288542e-05,
+      "loss": 0.7812,
+      "step": 9906
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0089465815918193e-05,
+      "loss": 1.1328,
+      "step": 9907
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.00878893667274e-05,
+      "loss": 0.2275,
+      "step": 9908
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0086312915352214e-05,
+      "loss": 0.8477,
+      "step": 9909
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0084736461831816e-05,
+      "loss": 1.2109,
+      "step": 9910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.008316000620539e-05,
+      "loss": 1.4688,
+      "step": 9911
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0081583548512113e-05,
+      "loss": 0.6953,
+      "step": 9912
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0080007088791173e-05,
+      "loss": 1.2578,
+      "step": 9913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0078430627081746e-05,
+      "loss": 0.8438,
+      "step": 9914
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.007685416342301e-05,
+      "loss": 1.1719,
+      "step": 9915
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0075277697854154e-05,
+      "loss": 0.7344,
+      "step": 9916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0073701230414352e-05,
+      "loss": 0.9531,
+      "step": 9917
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0072124761142794e-05,
+      "loss": 0.3965,
+      "step": 9918
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0070548290078655e-05,
+      "loss": 0.8164,
+      "step": 9919
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0068971817261113e-05,
+      "loss": 0.7148,
+      "step": 9920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0067395342729355e-05,
+      "loss": 0.7773,
+      "step": 9921
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0065818866522566e-05,
+      "loss": 1.2656,
+      "step": 9922
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0064242388679919e-05,
+      "loss": 1.6562,
+      "step": 9923
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.00626659092406e-05,
+      "loss": 1.6953,
+      "step": 9924
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0061089428243792e-05,
+      "loss": 1.1094,
+      "step": 9925
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0059512945728678e-05,
+      "loss": 0.7344,
+      "step": 9926
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0057936461734432e-05,
+      "loss": 0.7969,
+      "step": 9927
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0056359976300241e-05,
+      "loss": 0.582,
+      "step": 9928
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0054783489465286e-05,
+      "loss": 0.9102,
+      "step": 9929
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0053207001268752e-05,
+      "loss": 0.8867,
+      "step": 9930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0051630511749816e-05,
+      "loss": 0.377,
+      "step": 9931
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.005005402094766e-05,
+      "loss": 0.2051,
+      "step": 9932
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0048477528901468e-05,
+      "loss": 0.3145,
+      "step": 9933
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0046901035650426e-05,
+      "loss": 0.832,
+      "step": 9934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0045324541233707e-05,
+      "loss": 0.7305,
+      "step": 9935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0043748045690498e-05,
+      "loss": 0.6328,
+      "step": 9936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0042171549059983e-05,
+      "loss": 0.1206,
+      "step": 9937
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0040595051381338e-05,
+      "loss": 1.0625,
+      "step": 9938
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.003901855269375e-05,
+      "loss": 0.9609,
+      "step": 9939
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0037442053036398e-05,
+      "loss": 0.6992,
+      "step": 9940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0035865552448472e-05,
+      "loss": 1.7891,
+      "step": 9941
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0034289050969143e-05,
+      "loss": 1.5938,
+      "step": 9942
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0032712548637601e-05,
+      "loss": 0.6641,
+      "step": 9943
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0031136045493021e-05,
+      "loss": 1.1562,
+      "step": 9944
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0029559541574593e-05,
+      "loss": 0.6445,
+      "step": 9945
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0027983036921498e-05,
+      "loss": 0.4414,
+      "step": 9946
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0026406531572915e-05,
+      "loss": 0.6406,
+      "step": 9947
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0024830025568021e-05,
+      "loss": 0.9219,
+      "step": 9948
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0023253518946012e-05,
+      "loss": 0.6562,
+      "step": 9949
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.002167701174606e-05,
+      "loss": 1.0547,
+      "step": 9950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0020100504007352e-05,
+      "loss": 0.5078,
+      "step": 9951
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0018523995769067e-05,
+      "loss": 0.9883,
+      "step": 9952
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0016947487070388e-05,
+      "loss": 0.6367,
+      "step": 9953
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.00153709779505e-05,
+      "loss": 0.8203,
+      "step": 9954
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0013794468448584e-05,
+      "loss": 0.9102,
+      "step": 9955
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0012217958603817e-05,
+      "loss": 0.9883,
+      "step": 9956
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0010641448455391e-05,
+      "loss": 0.5742,
+      "step": 9957
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0009064938042487e-05,
+      "loss": 0.4531,
+      "step": 9958
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.000748842740428e-05,
+      "loss": 1.5781,
+      "step": 9959
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0005911916579956e-05,
+      "loss": 0.4199,
+      "step": 9960
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0004335405608699e-05,
+      "loss": 1.1562,
+      "step": 9961
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.000275889452969e-05,
+      "loss": 0.9766,
+      "step": 9962
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.0001182383382115e-05,
+      "loss": 0.5273,
+      "step": 9963
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.999605872205147e-06,
+      "loss": 0.6133,
+      "step": 9964
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.99802936103798e-06,
+      "loss": 0.8281,
+      "step": 9965
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996452849919788e-06,
+      "loss": 1.4688,
+      "step": 9966
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.994876338889757e-06,
+      "loss": 0.4473,
+      "step": 9967
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.993299827987068e-06,
+      "loss": 0.6094,
+      "step": 9968
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.991723317250906e-06,
+      "loss": 0.416,
+      "step": 9969
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.990146806720452e-06,
+      "loss": 0.498,
+      "step": 9970
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.988570296434888e-06,
+      "loss": 1.0547,
+      "step": 9971
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.986993786433393e-06,
+      "loss": 1.2578,
+      "step": 9972
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.985417276755155e-06,
+      "loss": 0.2695,
+      "step": 9973
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.983840767439357e-06,
+      "loss": 0.7031,
+      "step": 9974
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.982264258525176e-06,
+      "loss": 0.2637,
+      "step": 9975
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.980687750051795e-06,
+      "loss": 2.0625,
+      "step": 9976
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.9791112420584e-06,
+      "loss": 0.2275,
+      "step": 9977
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.977534734584172e-06,
+      "loss": 0.6016,
+      "step": 9978
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.975958227668295e-06,
+      "loss": 0.5117,
+      "step": 9979
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.974381721349945e-06,
+      "loss": 0.4043,
+      "step": 9980
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.972805215668311e-06,
+      "loss": 1.2812,
+      "step": 9981
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.971228710662571e-06,
+      "loss": 1.8281,
+      "step": 9982
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.969652206371911e-06,
+      "loss": 0.7578,
+      "step": 9983
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.968075702835514e-06,
+      "loss": 1.3906,
+      "step": 9984
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.966499200092553e-06,
+      "loss": 1.1406,
+      "step": 9985
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.96492269818222e-06,
+      "loss": 0.8125,
+      "step": 9986
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.963346197143692e-06,
+      "loss": 0.7031,
+      "step": 9987
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.961769697016155e-06,
+      "loss": 1.0703,
+      "step": 9988
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.960193197838788e-06,
+      "loss": 0.7305,
+      "step": 9989
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.958616699650775e-06,
+      "loss": 0.6055,
+      "step": 9990
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.957040202491299e-06,
+      "loss": 0.4746,
+      "step": 9991
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.95546370639954e-06,
+      "loss": 0.957,
+      "step": 9992
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.953887211414676e-06,
+      "loss": 1.4766,
+      "step": 9993
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.952310717575896e-06,
+      "loss": 1.2656,
+      "step": 9994
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.950734224922382e-06,
+      "loss": 1.2969,
+      "step": 9995
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.949157733493314e-06,
+      "loss": 0.543,
+      "step": 9996
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.947581243327869e-06,
+      "loss": 1.0469,
+      "step": 9997
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.946004754465235e-06,
+      "loss": 0.5625,
+      "step": 9998
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.944428266944592e-06,
+      "loss": 1.0547,
+      "step": 9999
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.942851780805124e-06,
+      "loss": 0.9414,
+      "step": 10000
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 0.9734683036804199,
+      "eval_runtime": 354.251,
+      "eval_samples_per_second": 2.371,
+      "eval_steps_per_second": 2.371,
+      "step": 10000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.941275296086005e-06,
+      "loss": 1.5781,
+      "step": 10001
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.93969881282643e-06,
+      "loss": 0.5703,
+      "step": 10002
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.938122331065568e-06,
+      "loss": 1.0703,
+      "step": 10003
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.936545850842609e-06,
+      "loss": 0.8945,
+      "step": 10004
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.934969372196727e-06,
+      "loss": 1.0234,
+      "step": 10005
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.933392895167111e-06,
+      "loss": 0.3184,
+      "step": 10006
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.931816419792941e-06,
+      "loss": 1.125,
+      "step": 10007
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.930239946113397e-06,
+      "loss": 0.4922,
+      "step": 10008
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.928663474167657e-06,
+      "loss": 0.4082,
+      "step": 10009
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.927087003994909e-06,
+      "loss": 0.4766,
+      "step": 10010
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.925510535634331e-06,
+      "loss": 1.2188,
+      "step": 10011
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.923934069125107e-06,
+      "loss": 0.3008,
+      "step": 10012
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.92235760450641e-06,
+      "loss": 0.7773,
+      "step": 10013
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.920781141817433e-06,
+      "loss": 1.125,
+      "step": 10014
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.91920468109735e-06,
+      "loss": 0.8438,
+      "step": 10015
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.917628222385344e-06,
+      "loss": 1.0,
+      "step": 10016
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.916051765720592e-06,
+      "loss": 0.8906,
+      "step": 10017
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.914475311142286e-06,
+      "loss": 0.7695,
+      "step": 10018
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.912898858689595e-06,
+      "loss": 0.6523,
+      "step": 10019
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.911322408401707e-06,
+      "loss": 0.7891,
+      "step": 10020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.909745960317797e-06,
+      "loss": 1.0781,
+      "step": 10021
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.908169514477054e-06,
+      "loss": 0.5742,
+      "step": 10022
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.906593070918655e-06,
+      "loss": 1.1172,
+      "step": 10023
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.905016629681779e-06,
+      "loss": 0.7891,
+      "step": 10024
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.903440190805605e-06,
+      "loss": 1.3203,
+      "step": 10025
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.901863754329318e-06,
+      "loss": 0.2891,
+      "step": 10026
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.900287320292097e-06,
+      "loss": 0.957,
+      "step": 10027
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.898710888733124e-06,
+      "loss": 1.2578,
+      "step": 10028
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.897134459691574e-06,
+      "loss": 0.4004,
+      "step": 10029
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.895558033206637e-06,
+      "loss": 0.377,
+      "step": 10030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.893981609317484e-06,
+      "loss": 0.8281,
+      "step": 10031
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.8924051880633e-06,
+      "loss": 1.3125,
+      "step": 10032
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.890828769483261e-06,
+      "loss": 0.5508,
+      "step": 10033
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.889252353616553e-06,
+      "loss": 0.9766,
+      "step": 10034
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.887675940502356e-06,
+      "loss": 1.8047,
+      "step": 10035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.886099530179842e-06,
+      "loss": 0.7266,
+      "step": 10036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.884523122688197e-06,
+      "loss": 0.875,
+      "step": 10037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.8829467180666e-06,
+      "loss": 0.8359,
+      "step": 10038
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.881370316354232e-06,
+      "loss": 0.6211,
+      "step": 10039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.879793917590272e-06,
+      "loss": 1.1797,
+      "step": 10040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.878217521813896e-06,
+      "loss": 1.0625,
+      "step": 10041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.876641129064289e-06,
+      "loss": 1.2266,
+      "step": 10042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.875064739380626e-06,
+      "loss": 1.6094,
+      "step": 10043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.87348835280209e-06,
+      "loss": 0.2871,
+      "step": 10044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.871911969367857e-06,
+      "loss": 1.4922,
+      "step": 10045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.87033558911711e-06,
+      "loss": 0.6719,
+      "step": 10046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.868759212089026e-06,
+      "loss": 1.0234,
+      "step": 10047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.867182838322782e-06,
+      "loss": 0.9492,
+      "step": 10048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.86560646785756e-06,
+      "loss": 1.0625,
+      "step": 10049
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.864030100732537e-06,
+      "loss": 1.3828,
+      "step": 10050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.862453736986897e-06,
+      "loss": 0.9609,
+      "step": 10051
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.860877376659812e-06,
+      "loss": 0.875,
+      "step": 10052
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.859301019790462e-06,
+      "loss": 0.7148,
+      "step": 10053
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.857724666418028e-06,
+      "loss": 0.7656,
+      "step": 10054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.856148316581686e-06,
+      "loss": 1.125,
+      "step": 10055
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.85457197032062e-06,
+      "loss": 0.5273,
+      "step": 10056
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.852995627674001e-06,
+      "loss": 0.7383,
+      "step": 10057
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.851419288681008e-06,
+      "loss": 0.7773,
+      "step": 10058
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.849842953380826e-06,
+      "loss": 1.0312,
+      "step": 10059
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.848266621812625e-06,
+      "loss": 0.4395,
+      "step": 10060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.846690294015588e-06,
+      "loss": 1.1016,
+      "step": 10061
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.845113970028888e-06,
+      "loss": 0.5312,
+      "step": 10062
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.843537649891712e-06,
+      "loss": 0.875,
+      "step": 10063
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.841961333643229e-06,
+      "loss": 0.7734,
+      "step": 10064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.840385021322618e-06,
+      "loss": 1.0312,
+      "step": 10065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.838808712969056e-06,
+      "loss": 0.3613,
+      "step": 10066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.837232408621727e-06,
+      "loss": 0.7148,
+      "step": 10067
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.835656108319803e-06,
+      "loss": 0.8672,
+      "step": 10068
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.83407981210246e-06,
+      "loss": 0.5664,
+      "step": 10069
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.832503520008872e-06,
+      "loss": 0.6406,
+      "step": 10070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.830927232078225e-06,
+      "loss": 0.6992,
+      "step": 10071
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.829350948349691e-06,
+      "loss": 1.2578,
+      "step": 10072
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.827774668862452e-06,
+      "loss": 1.5156,
+      "step": 10073
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.826198393655673e-06,
+      "loss": 0.7656,
+      "step": 10074
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.82462212276854e-06,
+      "loss": 0.7969,
+      "step": 10075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.823045856240225e-06,
+      "loss": 1.8281,
+      "step": 10076
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.821469594109908e-06,
+      "loss": 0.4727,
+      "step": 10077
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.819893336416762e-06,
+      "loss": 0.8125,
+      "step": 10078
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.818317083199966e-06,
+      "loss": 0.7188,
+      "step": 10079
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.816740834498694e-06,
+      "loss": 0.4023,
+      "step": 10080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.815164590352124e-06,
+      "loss": 0.5156,
+      "step": 10081
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.813588350799426e-06,
+      "loss": 0.9531,
+      "step": 10082
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.812012115879782e-06,
+      "loss": 0.7656,
+      "step": 10083
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.810435885632368e-06,
+      "loss": 0.7773,
+      "step": 10084
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.808859660096355e-06,
+      "loss": 0.7539,
+      "step": 10085
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.807283439310918e-06,
+      "loss": 0.8594,
+      "step": 10086
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.805707223315236e-06,
+      "loss": 0.3184,
+      "step": 10087
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.804131012148482e-06,
+      "loss": 0.5781,
+      "step": 10088
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.802554805849833e-06,
+      "loss": 0.5664,
+      "step": 10089
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.80097860445846e-06,
+      "loss": 1.3047,
+      "step": 10090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.799402408013542e-06,
+      "loss": 0.8906,
+      "step": 10091
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.79782621655425e-06,
+      "loss": 0.9648,
+      "step": 10092
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.79625003011976e-06,
+      "loss": 0.4219,
+      "step": 10093
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.794673848749243e-06,
+      "loss": 1.5938,
+      "step": 10094
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.79309767248188e-06,
+      "loss": 1.0547,
+      "step": 10095
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.791521501356845e-06,
+      "loss": 0.7031,
+      "step": 10096
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.789945335413303e-06,
+      "loss": 0.1138,
+      "step": 10097
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.788369174690435e-06,
+      "loss": 0.8086,
+      "step": 10098
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.786793019227412e-06,
+      "loss": 0.9922,
+      "step": 10099
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.785216869063411e-06,
+      "loss": 1.2344,
+      "step": 10100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.783640724237603e-06,
+      "loss": 0.7656,
+      "step": 10101
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.782064584789159e-06,
+      "loss": 1.0156,
+      "step": 10102
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.780488450757256e-06,
+      "loss": 0.4453,
+      "step": 10103
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.778912322181067e-06,
+      "loss": 0.8867,
+      "step": 10104
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.777336199099762e-06,
+      "loss": 0.9688,
+      "step": 10105
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.775760081552513e-06,
+      "loss": 1.4062,
+      "step": 10106
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.774183969578498e-06,
+      "loss": 0.5586,
+      "step": 10107
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.772607863216887e-06,
+      "loss": 0.8203,
+      "step": 10108
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.77103176250685e-06,
+      "loss": 1.3672,
+      "step": 10109
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.76945566748756e-06,
+      "loss": 1.1562,
+      "step": 10110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.76787957819819e-06,
+      "loss": 0.4023,
+      "step": 10111
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.766303494677917e-06,
+      "loss": 1.2188,
+      "step": 10112
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.764727416965904e-06,
+      "loss": 1.1172,
+      "step": 10113
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.763151345101324e-06,
+      "loss": 0.5469,
+      "step": 10114
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.761575279123353e-06,
+      "loss": 0.4785,
+      "step": 10115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.75999921907116e-06,
+      "loss": 0.7695,
+      "step": 10116
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.75842316498392e-06,
+      "loss": 0.6406,
+      "step": 10117
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.756847116900796e-06,
+      "loss": 0.4785,
+      "step": 10118
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.755271074860964e-06,
+      "loss": 1.3125,
+      "step": 10119
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.753695038903594e-06,
+      "loss": 0.6641,
+      "step": 10120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.752119009067858e-06,
+      "loss": 0.9844,
+      "step": 10121
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.750542985392924e-06,
+      "loss": 1.8438,
+      "step": 10122
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.748966967917965e-06,
+      "loss": 0.8438,
+      "step": 10123
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.747390956682149e-06,
+      "loss": 1.2891,
+      "step": 10124
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745814951724646e-06,
+      "loss": 0.8477,
+      "step": 10125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.744238953084624e-06,
+      "loss": 1.1797,
+      "step": 10126
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.742662960801258e-06,
+      "loss": 0.8945,
+      "step": 10127
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.741086974913715e-06,
+      "loss": 0.3379,
+      "step": 10128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.739510995461166e-06,
+      "loss": 0.7578,
+      "step": 10129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.737935022482773e-06,
+      "loss": 1.6641,
+      "step": 10130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.736359056017713e-06,
+      "loss": 0.3555,
+      "step": 10131
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.73478309610515e-06,
+      "loss": 0.4785,
+      "step": 10132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.733207142784257e-06,
+      "loss": 0.9258,
+      "step": 10133
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.731631196094202e-06,
+      "loss": 0.2793,
+      "step": 10134
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.730055256074147e-06,
+      "loss": 1.5234,
+      "step": 10135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.728479322763268e-06,
+      "loss": 1.2734,
+      "step": 10136
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.726903396200728e-06,
+      "loss": 0.7031,
+      "step": 10137
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.7253274764257e-06,
+      "loss": 0.6211,
+      "step": 10138
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.723751563477344e-06,
+      "loss": 1.3359,
+      "step": 10139
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.722175657394837e-06,
+      "loss": 0.3574,
+      "step": 10140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.72059975821734e-06,
+      "loss": 1.1641,
+      "step": 10141
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.71902386598402e-06,
+      "loss": 1.6719,
+      "step": 10142
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.717447980734043e-06,
+      "loss": 1.3516,
+      "step": 10143
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.715872102506584e-06,
+      "loss": 1.0312,
+      "step": 10144
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.714296231340804e-06,
+      "loss": 1.5938,
+      "step": 10145
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.712720367275869e-06,
+      "loss": 0.6758,
+      "step": 10146
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.711144510350944e-06,
+      "loss": 1.0625,
+      "step": 10147
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.709568660605197e-06,
+      "loss": 1.4766,
+      "step": 10148
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.707992818077794e-06,
+      "loss": 1.0781,
+      "step": 10149
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.706416982807905e-06,
+      "loss": 0.9453,
+      "step": 10150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.704841154834688e-06,
+      "loss": 0.7812,
+      "step": 10151
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.703265334197314e-06,
+      "loss": 0.6992,
+      "step": 10152
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.701689520934945e-06,
+      "loss": 1.0078,
+      "step": 10153
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.700113715086747e-06,
+      "loss": 0.3926,
+      "step": 10154
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.698537916691885e-06,
+      "loss": 0.9023,
+      "step": 10155
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.696962125789526e-06,
+      "loss": 0.7773,
+      "step": 10156
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.695386342418833e-06,
+      "loss": 0.4883,
+      "step": 10157
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.693810566618967e-06,
+      "loss": 1.3125,
+      "step": 10158
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.692234798429094e-06,
+      "loss": 1.3359,
+      "step": 10159
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.690659037888381e-06,
+      "loss": 0.707,
+      "step": 10160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.689083285035989e-06,
+      "loss": 1.0234,
+      "step": 10161
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.687507539911085e-06,
+      "loss": 1.5547,
+      "step": 10162
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.685931802552826e-06,
+      "loss": 0.625,
+      "step": 10163
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.684356073000379e-06,
+      "loss": 1.3359,
+      "step": 10164
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.682780351292908e-06,
+      "loss": 0.4629,
+      "step": 10165
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.681204637469575e-06,
+      "loss": 1.2578,
+      "step": 10166
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.679628931569539e-06,
+      "loss": 0.7656,
+      "step": 10167
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.67805323363197e-06,
+      "loss": 0.9258,
+      "step": 10168
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.676477543696022e-06,
+      "loss": 0.5781,
+      "step": 10169
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.674901861800862e-06,
+      "loss": 0.8125,
+      "step": 10170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.67332618798565e-06,
+      "loss": 0.7344,
+      "step": 10171
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.671750522289547e-06,
+      "loss": 1.1953,
+      "step": 10172
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.67017486475172e-06,
+      "loss": 1.1797,
+      "step": 10173
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.668599215411322e-06,
+      "loss": 0.8828,
+      "step": 10174
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.667023574307515e-06,
+      "loss": 0.6094,
+      "step": 10175
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.665447941479466e-06,
+      "loss": 0.4277,
+      "step": 10176
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.663872316966331e-06,
+      "loss": 0.7461,
+      "step": 10177
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.662296700807276e-06,
+      "loss": 0.2334,
+      "step": 10178
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.66072109304145e-06,
+      "loss": 1.0391,
+      "step": 10179
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.659145493708022e-06,
+      "loss": 0.1455,
+      "step": 10180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.65756990284615e-06,
+      "loss": 0.334,
+      "step": 10181
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.655994320494993e-06,
+      "loss": 1.1562,
+      "step": 10182
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.654418746693707e-06,
+      "loss": 1.3906,
+      "step": 10183
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.652843181481459e-06,
+      "loss": 0.5312,
+      "step": 10184
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.6512676248974e-06,
+      "loss": 0.3652,
+      "step": 10185
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.649692076980694e-06,
+      "loss": 1.2969,
+      "step": 10186
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.648116537770494e-06,
+      "loss": 1.1484,
+      "step": 10187
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.646541007305963e-06,
+      "loss": 1.5938,
+      "step": 10188
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.64496548562626e-06,
+      "loss": 0.9258,
+      "step": 10189
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.643389972770538e-06,
+      "loss": 0.5977,
+      "step": 10190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.641814468777955e-06,
+      "loss": 1.1172,
+      "step": 10191
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.640238973687675e-06,
+      "loss": 1.6641,
+      "step": 10192
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.638663487538848e-06,
+      "loss": 0.8594,
+      "step": 10193
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.637088010370636e-06,
+      "loss": 0.4707,
+      "step": 10194
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.635512542222187e-06,
+      "loss": 0.668,
+      "step": 10195
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.633937083132671e-06,
+      "loss": 0.9805,
+      "step": 10196
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.632361633141232e-06,
+      "loss": 0.6328,
+      "step": 10197
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.630786192287034e-06,
+      "loss": 0.4141,
+      "step": 10198
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.629210760609227e-06,
+      "loss": 1.3984,
+      "step": 10199
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.62763533814697e-06,
+      "loss": 0.9961,
+      "step": 10200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.626059924939423e-06,
+      "loss": 0.4629,
+      "step": 10201
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.624484521025731e-06,
+      "loss": 1.6719,
+      "step": 10202
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.622909126445053e-06,
+      "loss": 0.5195,
+      "step": 10203
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.621333741236544e-06,
+      "loss": 1.0078,
+      "step": 10204
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.619758365439362e-06,
+      "loss": 1.0312,
+      "step": 10205
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.618182999092659e-06,
+      "loss": 0.9336,
+      "step": 10206
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.616607642235586e-06,
+      "loss": 1.2422,
+      "step": 10207
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.615032294907297e-06,
+      "loss": 0.918,
+      "step": 10208
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.613456957146947e-06,
+      "loss": 0.5078,
+      "step": 10209
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.611881628993692e-06,
+      "loss": 0.5156,
+      "step": 10210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.610306310486684e-06,
+      "loss": 0.8828,
+      "step": 10211
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.60873100166507e-06,
+      "loss": 1.2188,
+      "step": 10212
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.607155702568009e-06,
+      "loss": 0.7852,
+      "step": 10213
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.60558041323465e-06,
+      "loss": 0.8203,
+      "step": 10214
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.604005133704148e-06,
+      "loss": 0.8281,
+      "step": 10215
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.602429864015649e-06,
+      "loss": 1.0859,
+      "step": 10216
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.600854604208312e-06,
+      "loss": 1.3828,
+      "step": 10217
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.599279354321286e-06,
+      "loss": 1.1875,
+      "step": 10218
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.597704114393718e-06,
+      "loss": 0.9375,
+      "step": 10219
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.596128884464759e-06,
+      "loss": 0.668,
+      "step": 10220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.594553664573566e-06,
+      "loss": 0.7812,
+      "step": 10221
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.592978454759288e-06,
+      "loss": 1.3203,
+      "step": 10222
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.59140325506107e-06,
+      "loss": 0.9141,
+      "step": 10223
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.58982806551806e-06,
+      "loss": 1.3203,
+      "step": 10224
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.588252886169416e-06,
+      "loss": 0.5703,
+      "step": 10225
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.586677717054284e-06,
+      "loss": 0.6172,
+      "step": 10226
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.585102558211813e-06,
+      "loss": 1.2656,
+      "step": 10227
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.583527409681149e-06,
+      "loss": 1.2656,
+      "step": 10228
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.581952271501442e-06,
+      "loss": 0.7461,
+      "step": 10229
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.580377143711843e-06,
+      "loss": 0.7773,
+      "step": 10230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.578802026351498e-06,
+      "loss": 0.1108,
+      "step": 10231
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.577226919459551e-06,
+      "loss": 0.5742,
+      "step": 10232
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.575651823075155e-06,
+      "loss": 0.3242,
+      "step": 10233
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.574076737237462e-06,
+      "loss": 1.3828,
+      "step": 10234
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.572501661985606e-06,
+      "loss": 0.9883,
+      "step": 10235
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.570926597358739e-06,
+      "loss": 1.1172,
+      "step": 10236
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.569351543396011e-06,
+      "loss": 1.1484,
+      "step": 10237
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.567776500136568e-06,
+      "loss": 0.9102,
+      "step": 10238
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.566201467619554e-06,
+      "loss": 0.7305,
+      "step": 10239
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.56462644588411e-06,
+      "loss": 1.0938,
+      "step": 10240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.563051434969388e-06,
+      "loss": 1.3516,
+      "step": 10241
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.561476434914531e-06,
+      "loss": 1.6484,
+      "step": 10242
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.559901445758685e-06,
+      "loss": 1.6719,
+      "step": 10243
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.55832646754099e-06,
+      "loss": 0.9492,
+      "step": 10244
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.556751500300599e-06,
+      "loss": 0.6484,
+      "step": 10245
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.555176544076649e-06,
+      "loss": 0.8555,
+      "step": 10246
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.553601598908285e-06,
+      "loss": 0.9531,
+      "step": 10247
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.552026664834648e-06,
+      "loss": 1.0078,
+      "step": 10248
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.550451741894887e-06,
+      "loss": 1.4297,
+      "step": 10249
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.548876830128144e-06,
+      "loss": 0.9062,
+      "step": 10250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.547301929573559e-06,
+      "loss": 1.0547,
+      "step": 10251
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.545727040270272e-06,
+      "loss": 1.0547,
+      "step": 10252
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.544152162257431e-06,
+      "loss": 0.8672,
+      "step": 10253
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.542577295574174e-06,
+      "loss": 1.1953,
+      "step": 10254
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.541002440259646e-06,
+      "loss": 1.0938,
+      "step": 10255
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.539427596352983e-06,
+      "loss": 1.7734,
+      "step": 10256
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.53785276389333e-06,
+      "loss": 0.875,
+      "step": 10257
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.536277942919829e-06,
+      "loss": 1.2266,
+      "step": 10258
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.534703133471615e-06,
+      "loss": 1.0703,
+      "step": 10259
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.53312833558783e-06,
+      "loss": 0.8086,
+      "step": 10260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.531553549307619e-06,
+      "loss": 1.1016,
+      "step": 10261
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.529978774670116e-06,
+      "loss": 1.1953,
+      "step": 10262
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.528404011714462e-06,
+      "loss": 1.1172,
+      "step": 10263
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.526829260479791e-06,
+      "loss": 1.2188,
+      "step": 10264
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.525254521005252e-06,
+      "loss": 0.6953,
+      "step": 10265
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.523679793329977e-06,
+      "loss": 0.9805,
+      "step": 10266
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.522105077493105e-06,
+      "loss": 1.0703,
+      "step": 10267
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.520530373533771e-06,
+      "loss": 0.7383,
+      "step": 10268
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.518955681491117e-06,
+      "loss": 0.7109,
+      "step": 10269
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.517381001404278e-06,
+      "loss": 0.5742,
+      "step": 10270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.515806333312392e-06,
+      "loss": 0.7578,
+      "step": 10271
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.514231677254591e-06,
+      "loss": 1.3281,
+      "step": 10272
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.512657033270021e-06,
+      "loss": 0.8242,
+      "step": 10273
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.51108240139781e-06,
+      "loss": 1.0469,
+      "step": 10274
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.509507781677095e-06,
+      "loss": 0.8477,
+      "step": 10275
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.50793317414701e-06,
+      "loss": 0.3613,
+      "step": 10276
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.506358578846696e-06,
+      "loss": 0.4551,
+      "step": 10277
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.504783995815285e-06,
+      "loss": 0.7773,
+      "step": 10278
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.50320942509191e-06,
+      "loss": 0.3301,
+      "step": 10279
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.501634866715703e-06,
+      "loss": 0.7305,
+      "step": 10280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.500060320725801e-06,
+      "loss": 0.1943,
+      "step": 10281
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.49848578716134e-06,
+      "loss": 0.3145,
+      "step": 10282
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.49691126606145e-06,
+      "loss": 0.3262,
+      "step": 10283
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.495336757465265e-06,
+      "loss": 0.3008,
+      "step": 10284
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.493762261411913e-06,
+      "loss": 0.4863,
+      "step": 10285
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.492187777940534e-06,
+      "loss": 0.7969,
+      "step": 10286
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.490613307090254e-06,
+      "loss": 0.9805,
+      "step": 10287
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.489038848900212e-06,
+      "loss": 1.8594,
+      "step": 10288
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.48746440340953e-06,
+      "loss": 1.3594,
+      "step": 10289
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.485889970657345e-06,
+      "loss": 1.7812,
+      "step": 10290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.484315550682786e-06,
+      "loss": 1.2812,
+      "step": 10291
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.482741143524983e-06,
+      "loss": 0.3477,
+      "step": 10292
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.481166749223068e-06,
+      "loss": 1.5234,
+      "step": 10293
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.479592367816171e-06,
+      "loss": 0.5664,
+      "step": 10294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.47801799934342e-06,
+      "loss": 0.8086,
+      "step": 10295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.476443643843943e-06,
+      "loss": 0.8945,
+      "step": 10296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.474869301356869e-06,
+      "loss": 0.4434,
+      "step": 10297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.47329497192133e-06,
+      "loss": 0.8555,
+      "step": 10298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.471720655576451e-06,
+      "loss": 0.3945,
+      "step": 10299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.470146352361363e-06,
+      "loss": 1.2109,
+      "step": 10300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.468572062315187e-06,
+      "loss": 1.0781,
+      "step": 10301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.466997785477057e-06,
+      "loss": 0.6289,
+      "step": 10302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.465423521886097e-06,
+      "loss": 1.4453,
+      "step": 10303
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.463849271581434e-06,
+      "loss": 0.9922,
+      "step": 10304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.462275034602193e-06,
+      "loss": 0.5312,
+      "step": 10305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.460700810987505e-06,
+      "loss": 1.0078,
+      "step": 10306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.459126600776488e-06,
+      "loss": 0.8203,
+      "step": 10307
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.457552404008273e-06,
+      "loss": 0.9805,
+      "step": 10308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.455978220721979e-06,
+      "loss": 1.3438,
+      "step": 10309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.454404050956736e-06,
+      "loss": 0.3691,
+      "step": 10310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.452829894751671e-06,
+      "loss": 1.2656,
+      "step": 10311
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.4512557521459e-06,
+      "loss": 0.8633,
+      "step": 10312
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.449681623178546e-06,
+      "loss": 1.3047,
+      "step": 10313
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.44810750788874e-06,
+      "loss": 0.9727,
+      "step": 10314
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.446533406315601e-06,
+      "loss": 0.7539,
+      "step": 10315
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.444959318498254e-06,
+      "loss": 0.8438,
+      "step": 10316
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.443385244475813e-06,
+      "loss": 0.8516,
+      "step": 10317
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.44181118428741e-06,
+      "loss": 1.2656,
+      "step": 10318
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.44023713797216e-06,
+      "loss": 0.7148,
+      "step": 10319
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.438663105569186e-06,
+      "loss": 0.8242,
+      "step": 10320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.437089087117608e-06,
+      "loss": 0.4766,
+      "step": 10321
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.43551508265655e-06,
+      "loss": 0.7227,
+      "step": 10322
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.43394109222513e-06,
+      "loss": 0.1572,
+      "step": 10323
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.432367115862465e-06,
+      "loss": 0.8438,
+      "step": 10324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.430793153607675e-06,
+      "loss": 0.6289,
+      "step": 10325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.429219205499885e-06,
+      "loss": 0.2461,
+      "step": 10326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.42764527157821e-06,
+      "loss": 0.8594,
+      "step": 10327
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.426071351881766e-06,
+      "loss": 0.9883,
+      "step": 10328
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.42449744644967e-06,
+      "loss": 0.8672,
+      "step": 10329
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.422923555321045e-06,
+      "loss": 0.793,
+      "step": 10330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.421349678535005e-06,
+      "loss": 0.8203,
+      "step": 10331
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.41977581613067e-06,
+      "loss": 0.5781,
+      "step": 10332
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.418201968147151e-06,
+      "loss": 1.0234,
+      "step": 10333
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.416628134623572e-06,
+      "loss": 0.9297,
+      "step": 10334
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.415054315599041e-06,
+      "loss": 0.2295,
+      "step": 10335
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.413480511112678e-06,
+      "loss": 0.6797,
+      "step": 10336
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.411906721203594e-06,
+      "loss": 0.3965,
+      "step": 10337
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.410332945910911e-06,
+      "loss": 0.4785,
+      "step": 10338
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.408759185273739e-06,
+      "loss": 0.2617,
+      "step": 10339
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.407185439331191e-06,
+      "loss": 0.7891,
+      "step": 10340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.40561170812238e-06,
+      "loss": 0.875,
+      "step": 10341
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.404037991686423e-06,
+      "loss": 0.1592,
+      "step": 10342
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.402464290062431e-06,
+      "loss": 0.8242,
+      "step": 10343
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.400890603289521e-06,
+      "loss": 1.2031,
+      "step": 10344
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.399316931406796e-06,
+      "loss": 0.8594,
+      "step": 10345
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.397743274453373e-06,
+      "loss": 0.7422,
+      "step": 10346
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.396169632468365e-06,
+      "loss": 0.5352,
+      "step": 10347
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.394596005490881e-06,
+      "loss": 1.0078,
+      "step": 10348
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.39302239356003e-06,
+      "loss": 0.6875,
+      "step": 10349
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.391448796714929e-06,
+      "loss": 1.0078,
+      "step": 10350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.389875214994682e-06,
+      "loss": 0.6914,
+      "step": 10351
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.3883016484384e-06,
+      "loss": 0.5898,
+      "step": 10352
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.38672809708519e-06,
+      "loss": 1.1875,
+      "step": 10353
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.385154560974163e-06,
+      "loss": 0.2412,
+      "step": 10354
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.383581040144433e-06,
+      "loss": 0.4121,
+      "step": 10355
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.382007534635099e-06,
+      "loss": 0.9844,
+      "step": 10356
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.380434044485273e-06,
+      "loss": 1.3047,
+      "step": 10357
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.378860569734059e-06,
+      "loss": 0.6055,
+      "step": 10358
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.37728711042057e-06,
+      "loss": 0.6641,
+      "step": 10359
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.37571366658391e-06,
+      "loss": 1.4531,
+      "step": 10360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.374140238263184e-06,
+      "loss": 0.7617,
+      "step": 10361
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.372566825497495e-06,
+      "loss": 0.7539,
+      "step": 10362
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.370993428325954e-06,
+      "loss": 1.4453,
+      "step": 10363
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.369420046787662e-06,
+      "loss": 1.4062,
+      "step": 10364
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.36784668092173e-06,
+      "loss": 0.4238,
+      "step": 10365
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.36627333076725e-06,
+      "loss": 0.4355,
+      "step": 10366
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.36469999636334e-06,
+      "loss": 1.3203,
+      "step": 10367
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.363126677749094e-06,
+      "loss": 1.2734,
+      "step": 10368
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.36155337496362e-06,
+      "loss": 0.7305,
+      "step": 10369
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.359980088046014e-06,
+      "loss": 0.9258,
+      "step": 10370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.358406817035387e-06,
+      "loss": 0.3555,
+      "step": 10371
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.356833561970838e-06,
+      "loss": 1.5625,
+      "step": 10372
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.355260322891466e-06,
+      "loss": 1.4375,
+      "step": 10373
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.35368709983637e-06,
+      "loss": 0.7344,
+      "step": 10374
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.352113892844655e-06,
+      "loss": 0.7656,
+      "step": 10375
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.350540701955422e-06,
+      "loss": 1.375,
+      "step": 10376
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.348967527207771e-06,
+      "loss": 0.9258,
+      "step": 10377
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.347394368640795e-06,
+      "loss": 0.3027,
+      "step": 10378
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.3458212262936e-06,
+      "loss": 0.9648,
+      "step": 10379
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.344248100205282e-06,
+      "loss": 0.3691,
+      "step": 10380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.342674990414942e-06,
+      "loss": 0.875,
+      "step": 10381
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.34110189696167e-06,
+      "loss": 0.2617,
+      "step": 10382
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.339528819884577e-06,
+      "loss": 0.2617,
+      "step": 10383
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.337955759222747e-06,
+      "loss": 0.6055,
+      "step": 10384
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.336382715015283e-06,
+      "loss": 0.9414,
+      "step": 10385
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.334809687301279e-06,
+      "loss": 0.3164,
+      "step": 10386
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.333236676119833e-06,
+      "loss": 0.252,
+      "step": 10387
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.33166368151004e-06,
+      "loss": 0.793,
+      "step": 10388
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.330090703510993e-06,
+      "loss": 0.4648,
+      "step": 10389
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.328517742161787e-06,
+      "loss": 0.2168,
+      "step": 10390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.326944797501517e-06,
+      "loss": 1.8281,
+      "step": 10391
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.32537186956928e-06,
+      "loss": 1.0,
+      "step": 10392
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.323798958404166e-06,
+      "loss": 0.5586,
+      "step": 10393
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.322226064045264e-06,
+      "loss": 1.4219,
+      "step": 10394
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.320653186531673e-06,
+      "loss": 0.5625,
+      "step": 10395
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.319080325902483e-06,
+      "loss": 1.2344,
+      "step": 10396
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.317507482196784e-06,
+      "loss": 0.7578,
+      "step": 10397
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.31593465545367e-06,
+      "loss": 0.9453,
+      "step": 10398
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.314361845712232e-06,
+      "loss": 0.8555,
+      "step": 10399
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.312789053011557e-06,
+      "loss": 0.3027,
+      "step": 10400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.311216277390738e-06,
+      "loss": 1.6094,
+      "step": 10401
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.30964351888886e-06,
+      "loss": 1.2891,
+      "step": 10402
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.308070777545018e-06,
+      "loss": 0.4961,
+      "step": 10403
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.3064980533983e-06,
+      "loss": 1.4688,
+      "step": 10404
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.304925346487793e-06,
+      "loss": 1.0,
+      "step": 10405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.303352656852582e-06,
+      "loss": 1.8359,
+      "step": 10406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.301779984531757e-06,
+      "loss": 0.9336,
+      "step": 10407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.300207329564406e-06,
+      "loss": 1.1484,
+      "step": 10408
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.298634691989615e-06,
+      "loss": 0.4336,
+      "step": 10409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.297062071846467e-06,
+      "loss": 0.9492,
+      "step": 10410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.295489469174054e-06,
+      "loss": 1.3984,
+      "step": 10411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.293916884011456e-06,
+      "loss": 1.3359,
+      "step": 10412
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.29234431639776e-06,
+      "loss": 0.9648,
+      "step": 10413
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.290771766372048e-06,
+      "loss": 0.5312,
+      "step": 10414
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.289199233973407e-06,
+      "loss": 0.6172,
+      "step": 10415
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.287626719240922e-06,
+      "loss": 0.5234,
+      "step": 10416
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.286054222213671e-06,
+      "loss": 0.8125,
+      "step": 10417
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.284481742930737e-06,
+      "loss": 0.9375,
+      "step": 10418
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.282909281431206e-06,
+      "loss": 0.6133,
+      "step": 10419
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.28133683775416e-06,
+      "loss": 0.6602,
+      "step": 10420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.279764411938677e-06,
+      "loss": 0.668,
+      "step": 10421
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.278192004023839e-06,
+      "loss": 1.8359,
+      "step": 10422
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.276619614048726e-06,
+      "loss": 0.7422,
+      "step": 10423
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.27504724205242e-06,
+      "loss": 1.2812,
+      "step": 10424
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.273474888073998e-06,
+      "loss": 0.6133,
+      "step": 10425
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.271902552152539e-06,
+      "loss": 1.3828,
+      "step": 10426
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.270330234327127e-06,
+      "loss": 1.0312,
+      "step": 10427
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.268757934636834e-06,
+      "loss": 1.3516,
+      "step": 10428
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.26718565312074e-06,
+      "loss": 1.6328,
+      "step": 10429
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.26561338981792e-06,
+      "loss": 0.6367,
+      "step": 10430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.264041144767453e-06,
+      "loss": 1.6875,
+      "step": 10431
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.26246891800842e-06,
+      "loss": 0.21,
+      "step": 10432
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.26089670957989e-06,
+      "loss": 0.8086,
+      "step": 10433
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.25932451952094e-06,
+      "loss": 0.6758,
+      "step": 10434
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.257752347870645e-06,
+      "loss": 0.1113,
+      "step": 10435
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.256180194668082e-06,
+      "loss": 1.6094,
+      "step": 10436
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.25460805995232e-06,
+      "loss": 0.9219,
+      "step": 10437
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.253035943762443e-06,
+      "loss": 0.8711,
+      "step": 10438
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.251463846137511e-06,
+      "loss": 0.2891,
+      "step": 10439
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.249891767116605e-06,
+      "loss": 0.5938,
+      "step": 10440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.248319706738795e-06,
+      "loss": 1.2969,
+      "step": 10441
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.246747665043154e-06,
+      "loss": 0.918,
+      "step": 10442
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.245175642068749e-06,
+      "loss": 0.7148,
+      "step": 10443
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.243603637854658e-06,
+      "loss": 0.9258,
+      "step": 10444
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.242031652439947e-06,
+      "loss": 1.6172,
+      "step": 10445
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.240459685863686e-06,
+      "loss": 0.8164,
+      "step": 10446
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.238887738164943e-06,
+      "loss": 0.5859,
+      "step": 10447
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.237315809382792e-06,
+      "loss": 1.5859,
+      "step": 10448
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.2357438995563e-06,
+      "loss": 0.7188,
+      "step": 10449
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.23417200872453e-06,
+      "loss": 1.1484,
+      "step": 10450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.232600136926552e-06,
+      "loss": 0.5859,
+      "step": 10451
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.231028284201436e-06,
+      "loss": 0.5273,
+      "step": 10452
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.229456450588248e-06,
+      "loss": 0.9688,
+      "step": 10453
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.227884636126055e-06,
+      "loss": 0.8047,
+      "step": 10454
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.226312840853914e-06,
+      "loss": 1.6328,
+      "step": 10455
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.224741064810903e-06,
+      "loss": 0.8594,
+      "step": 10456
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.223169308036078e-06,
+      "loss": 1.0391,
+      "step": 10457
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.221597570568505e-06,
+      "loss": 0.9062,
+      "step": 10458
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.220025852447248e-06,
+      "loss": 0.2656,
+      "step": 10459
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.218454153711376e-06,
+      "loss": 0.2695,
+      "step": 10460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.216882474399944e-06,
+      "loss": 1.1484,
+      "step": 10461
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.215310814552016e-06,
+      "loss": 1.1016,
+      "step": 10462
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.213739174206654e-06,
+      "loss": 1.9844,
+      "step": 10463
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.212167553402923e-06,
+      "loss": 0.4766,
+      "step": 10464
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.210595952179881e-06,
+      "loss": 1.3672,
+      "step": 10465
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.209024370576588e-06,
+      "loss": 0.6719,
+      "step": 10466
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.207452808632102e-06,
+      "loss": 1.4844,
+      "step": 10467
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.205881266385486e-06,
+      "loss": 0.8242,
+      "step": 10468
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.204309743875799e-06,
+      "loss": 1.2031,
+      "step": 10469
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.202738241142096e-06,
+      "loss": 0.6094,
+      "step": 10470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.201166758223436e-06,
+      "loss": 0.8711,
+      "step": 10471
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.199595295158883e-06,
+      "loss": 1.1016,
+      "step": 10472
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.198023851987485e-06,
+      "loss": 1.4766,
+      "step": 10473
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.196452428748303e-06,
+      "loss": 1.2344,
+      "step": 10474
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.19488102548039e-06,
+      "loss": 0.6211,
+      "step": 10475
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.193309642222803e-06,
+      "loss": 1.4531,
+      "step": 10476
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.191738279014602e-06,
+      "loss": 1.1016,
+      "step": 10477
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.190166935894832e-06,
+      "loss": 0.2314,
+      "step": 10478
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.188595612902553e-06,
+      "loss": 0.9922,
+      "step": 10479
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.187024310076816e-06,
+      "loss": 0.2656,
+      "step": 10480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.185453027456676e-06,
+      "loss": 0.5977,
+      "step": 10481
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.183881765081188e-06,
+      "loss": 0.8281,
+      "step": 10482
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.182310522989396e-06,
+      "loss": 0.7109,
+      "step": 10483
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.180739301220358e-06,
+      "loss": 0.8477,
+      "step": 10484
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.179168099813122e-06,
+      "loss": 0.5391,
+      "step": 10485
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.17759691880674e-06,
+      "loss": 1.5234,
+      "step": 10486
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.176025758240259e-06,
+      "loss": 0.7969,
+      "step": 10487
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.174454618152736e-06,
+      "loss": 0.6484,
+      "step": 10488
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.172883498583212e-06,
+      "loss": 0.3418,
+      "step": 10489
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.171312399570738e-06,
+      "loss": 1.0938,
+      "step": 10490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.16974132115436e-06,
+      "loss": 0.4863,
+      "step": 10491
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.168170263373129e-06,
+      "loss": 0.6641,
+      "step": 10492
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.166599226266092e-06,
+      "loss": 0.9414,
+      "step": 10493
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.165028209872292e-06,
+      "loss": 0.8945,
+      "step": 10494
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.163457214230775e-06,
+      "loss": 0.9766,
+      "step": 10495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.161886239380589e-06,
+      "loss": 1.4062,
+      "step": 10496
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.160315285360777e-06,
+      "loss": 1.2031,
+      "step": 10497
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.158744352210389e-06,
+      "loss": 0.9375,
+      "step": 10498
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.157173439968456e-06,
+      "loss": 1.3672,
+      "step": 10499
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.155602548674033e-06,
+      "loss": 0.7109,
+      "step": 10500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.154031678366159e-06,
+      "loss": 0.8047,
+      "step": 10501
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.152460829083875e-06,
+      "loss": 1.0312,
+      "step": 10502
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.150890000866222e-06,
+      "loss": 0.9648,
+      "step": 10503
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.149319193752243e-06,
+      "loss": 0.5938,
+      "step": 10504
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.147748407780983e-06,
+      "loss": 0.9141,
+      "step": 10505
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.146177642991475e-06,
+      "loss": 0.6055,
+      "step": 10506
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.144606899422759e-06,
+      "loss": 1.8672,
+      "step": 10507
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.143036177113875e-06,
+      "loss": 0.8359,
+      "step": 10508
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.141465476103865e-06,
+      "loss": 1.25,
+      "step": 10509
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.139894796431767e-06,
+      "loss": 1.4062,
+      "step": 10510
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.138324138136616e-06,
+      "loss": 0.5977,
+      "step": 10511
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.136753501257444e-06,
+      "loss": 0.9609,
+      "step": 10512
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.135182885833296e-06,
+      "loss": 1.6406,
+      "step": 10513
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.133612291903203e-06,
+      "loss": 0.7578,
+      "step": 10514
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.132041719506207e-06,
+      "loss": 0.6562,
+      "step": 10515
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.13047116868133e-06,
+      "loss": 1.625,
+      "step": 10516
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.128900639467617e-06,
+      "loss": 0.5938,
+      "step": 10517
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.1273301319041e-06,
+      "loss": 1.0625,
+      "step": 10518
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.125759646029807e-06,
+      "loss": 0.5898,
+      "step": 10519
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.124189181883776e-06,
+      "loss": 0.6211,
+      "step": 10520
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.122618739505041e-06,
+      "loss": 0.5703,
+      "step": 10521
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.121048318932629e-06,
+      "loss": 1.4062,
+      "step": 10522
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.119477920205569e-06,
+      "loss": 0.1699,
+      "step": 10523
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.117907543362895e-06,
+      "loss": 1.6641,
+      "step": 10524
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.116337188443639e-06,
+      "loss": 0.8945,
+      "step": 10525
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.114766855486829e-06,
+      "loss": 1.2422,
+      "step": 10526
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.11319654453149e-06,
+      "loss": 0.9492,
+      "step": 10527
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.111626255616653e-06,
+      "loss": 0.6836,
+      "step": 10528
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.110055988781347e-06,
+      "loss": 0.4492,
+      "step": 10529
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.108485744064598e-06,
+      "loss": 0.4004,
+      "step": 10530
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.106915521505436e-06,
+      "loss": 0.5547,
+      "step": 10531
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.10534532114288e-06,
+      "loss": 0.3223,
+      "step": 10532
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.103775143015959e-06,
+      "loss": 0.7266,
+      "step": 10533
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.102204987163701e-06,
+      "loss": 1.2188,
+      "step": 10534
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.100634853625127e-06,
+      "loss": 0.1719,
+      "step": 10535
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.09906474243926e-06,
+      "loss": 0.4082,
+      "step": 10536
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.097494653645126e-06,
+      "loss": 0.6836,
+      "step": 10537
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.095924587281752e-06,
+      "loss": 0.2578,
+      "step": 10538
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.09435454338815e-06,
+      "loss": 1.3828,
+      "step": 10539
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.092784522003347e-06,
+      "loss": 0.2295,
+      "step": 10540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.091214523166366e-06,
+      "loss": 0.1572,
+      "step": 10541
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.089644546916225e-06,
+      "loss": 0.9297,
+      "step": 10542
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.088074593291947e-06,
+      "loss": 1.4688,
+      "step": 10543
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.086504662332545e-06,
+      "loss": 1.7578,
+      "step": 10544
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.084934754077042e-06,
+      "loss": 1.5,
+      "step": 10545
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.083364868564459e-06,
+      "loss": 0.3184,
+      "step": 10546
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.08179500583381e-06,
+      "loss": 0.9766,
+      "step": 10547
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.08022516592411e-06,
+      "loss": 0.5234,
+      "step": 10548
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.078655348874384e-06,
+      "loss": 1.4297,
+      "step": 10549
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.07708555472364e-06,
+      "loss": 0.6758,
+      "step": 10550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.075515783510896e-06,
+      "loss": 1.2812,
+      "step": 10551
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.073946035275167e-06,
+      "loss": 1.1953,
+      "step": 10552
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.072376310055468e-06,
+      "loss": 0.4961,
+      "step": 10553
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.070806607890814e-06,
+      "loss": 1.5391,
+      "step": 10554
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.069236928820214e-06,
+      "loss": 0.6484,
+      "step": 10555
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.067667272882681e-06,
+      "loss": 1.625,
+      "step": 10556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.066097640117233e-06,
+      "loss": 0.5469,
+      "step": 10557
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.064528030562876e-06,
+      "loss": 1.0625,
+      "step": 10558
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.062958444258624e-06,
+      "loss": 1.25,
+      "step": 10559
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.061388881243483e-06,
+      "loss": 0.7773,
+      "step": 10560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.059819341556466e-06,
+      "loss": 1.2031,
+      "step": 10561
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.058249825236582e-06,
+      "loss": 1.0547,
+      "step": 10562
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.056680332322838e-06,
+      "loss": 0.6172,
+      "step": 10563
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.055110862854242e-06,
+      "loss": 0.9141,
+      "step": 10564
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.053541416869807e-06,
+      "loss": 0.6016,
+      "step": 10565
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.051971994408533e-06,
+      "loss": 1.0469,
+      "step": 10566
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.050402595509428e-06,
+      "loss": 1.2578,
+      "step": 10567
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.048833220211496e-06,
+      "loss": 0.5586,
+      "step": 10568
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.047263868553746e-06,
+      "loss": 1.0391,
+      "step": 10569
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.045694540575184e-06,
+      "loss": 1.0859,
+      "step": 10570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.044125236314808e-06,
+      "loss": 0.793,
+      "step": 10571
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.042555955811621e-06,
+      "loss": 1.2109,
+      "step": 10572
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.040986699104632e-06,
+      "loss": 0.3887,
+      "step": 10573
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.03941746623284e-06,
+      "loss": 0.9141,
+      "step": 10574
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.037848257235244e-06,
+      "loss": 0.3145,
+      "step": 10575
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.036279072150847e-06,
+      "loss": 1.1094,
+      "step": 10576
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.034709911018653e-06,
+      "loss": 0.2637,
+      "step": 10577
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.033140773877655e-06,
+      "loss": 1.3438,
+      "step": 10578
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.031571660766856e-06,
+      "loss": 0.6953,
+      "step": 10579
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.030002571725254e-06,
+      "loss": 0.416,
+      "step": 10580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.028433506791845e-06,
+      "loss": 0.1572,
+      "step": 10581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.026864466005631e-06,
+      "loss": 0.6406,
+      "step": 10582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.025295449405605e-06,
+      "loss": 0.875,
+      "step": 10583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.023726457030764e-06,
+      "loss": 1.0,
+      "step": 10584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.022157488920101e-06,
+      "loss": 1.6641,
+      "step": 10585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.020588545112615e-06,
+      "loss": 0.3887,
+      "step": 10586
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.019019625647303e-06,
+      "loss": 0.1572,
+      "step": 10587
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.01745073056315e-06,
+      "loss": 0.6016,
+      "step": 10588
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.015881859899153e-06,
+      "loss": 0.9258,
+      "step": 10589
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.014313013694307e-06,
+      "loss": 0.6484,
+      "step": 10590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.012744191987603e-06,
+      "loss": 1.5078,
+      "step": 10591
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.011175394818033e-06,
+      "loss": 0.8164,
+      "step": 10592
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.00960662222458e-06,
+      "loss": 1.4219,
+      "step": 10593
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.008037874246244e-06,
+      "loss": 0.6445,
+      "step": 10594
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.00646915092201e-06,
+      "loss": 0.6719,
+      "step": 10595
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.00490045229087e-06,
+      "loss": 0.6562,
+      "step": 10596
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.003331778391803e-06,
+      "loss": 0.7031,
+      "step": 10597
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.00176312926381e-06,
+      "loss": 1.7578,
+      "step": 10598
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.00019450494587e-06,
+      "loss": 0.5859,
+      "step": 10599
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.99862590547697e-06,
+      "loss": 1.0781,
+      "step": 10600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.997057330896095e-06,
+      "loss": 0.832,
+      "step": 10601
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.995488781242232e-06,
+      "loss": 0.6836,
+      "step": 10602
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.993920256554369e-06,
+      "loss": 0.6445,
+      "step": 10603
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.992351756871484e-06,
+      "loss": 0.2598,
+      "step": 10604
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.990783282232561e-06,
+      "loss": 0.7148,
+      "step": 10605
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.989214832676584e-06,
+      "loss": 1.3672,
+      "step": 10606
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.987646408242536e-06,
+      "loss": 1.2656,
+      "step": 10607
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.986078008969398e-06,
+      "loss": 1.1328,
+      "step": 10608
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.984509634896147e-06,
+      "loss": 0.9844,
+      "step": 10609
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.982941286061773e-06,
+      "loss": 1.1484,
+      "step": 10610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.981372962505246e-06,
+      "loss": 0.9688,
+      "step": 10611
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.979804664265547e-06,
+      "loss": 1.3984,
+      "step": 10612
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.978236391381655e-06,
+      "loss": 0.9141,
+      "step": 10613
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.976668143892548e-06,
+      "loss": 1.0703,
+      "step": 10614
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.975099921837207e-06,
+      "loss": 1.0156,
+      "step": 10615
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.973531725254602e-06,
+      "loss": 1.1094,
+      "step": 10616
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.971963554183708e-06,
+      "loss": 0.9922,
+      "step": 10617
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.970395408663506e-06,
+      "loss": 1.3203,
+      "step": 10618
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.968827288732968e-06,
+      "loss": 0.7305,
+      "step": 10619
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.967259194431069e-06,
+      "loss": 0.9492,
+      "step": 10620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.965691125796778e-06,
+      "loss": 1.0078,
+      "step": 10621
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.964123082869071e-06,
+      "loss": 1.4375,
+      "step": 10622
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.96255506568692e-06,
+      "loss": 1.0859,
+      "step": 10623
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.960987074289295e-06,
+      "loss": 0.9062,
+      "step": 10624
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.959419108715166e-06,
+      "loss": 1.1797,
+      "step": 10625
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.957851169003509e-06,
+      "loss": 0.3477,
+      "step": 10626
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.956283255193286e-06,
+      "loss": 0.75,
+      "step": 10627
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.6289,
+      "step": 10628
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.95314750543302e-06,
+      "loss": 0.4766,
+      "step": 10629
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.951579669560919e-06,
+      "loss": 0.3047,
+      "step": 10630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.950011859746125e-06,
+      "loss": 1.0234,
+      "step": 10631
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.948444076027603e-06,
+      "loss": 0.5117,
+      "step": 10632
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.94687631844432e-06,
+      "loss": 0.2139,
+      "step": 10633
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.945308587035241e-06,
+      "loss": 0.6523,
+      "step": 10634
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.943740881839333e-06,
+      "loss": 0.4434,
+      "step": 10635
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.942173202895558e-06,
+      "loss": 0.4961,
+      "step": 10636
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.940605550242873e-06,
+      "loss": 0.5508,
+      "step": 10637
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.939037923920249e-06,
+      "loss": 1.4062,
+      "step": 10638
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.937470323966643e-06,
+      "loss": 2.25,
+      "step": 10639
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.935902750421015e-06,
+      "loss": 1.5234,
+      "step": 10640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.934335203322327e-06,
+      "loss": 0.7852,
+      "step": 10641
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.93276768270954e-06,
+      "loss": 1.5547,
+      "step": 10642
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.931200188621613e-06,
+      "loss": 0.8828,
+      "step": 10643
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.929632721097502e-06,
+      "loss": 1.3125,
+      "step": 10644
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.928065280176164e-06,
+      "loss": 0.8359,
+      "step": 10645
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.926497865896557e-06,
+      "loss": 0.332,
+      "step": 10646
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.92493047829764e-06,
+      "loss": 1.1875,
+      "step": 10647
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.923363117418367e-06,
+      "loss": 0.3262,
+      "step": 10648
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.92179578329769e-06,
+      "loss": 0.7383,
+      "step": 10649
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.920228475974565e-06,
+      "loss": 1.4531,
+      "step": 10650
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.91866119548795e-06,
+      "loss": 1.0391,
+      "step": 10651
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.917093941876794e-06,
+      "loss": 0.9102,
+      "step": 10652
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.915526715180049e-06,
+      "loss": 0.7539,
+      "step": 10653
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.913959515436666e-06,
+      "loss": 0.7891,
+      "step": 10654
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.912392342685598e-06,
+      "loss": 1.3594,
+      "step": 10655
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.910825196965795e-06,
+      "loss": 1.0234,
+      "step": 10656
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.909258078316207e-06,
+      "loss": 0.6445,
+      "step": 10657
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.90769098677578e-06,
+      "loss": 1.5312,
+      "step": 10658
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.906123922383467e-06,
+      "loss": 1.3594,
+      "step": 10659
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.904556885178213e-06,
+      "loss": 0.6875,
+      "step": 10660
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.902989875198964e-06,
+      "loss": 1.1641,
+      "step": 10661
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.901422892484667e-06,
+      "loss": 0.7422,
+      "step": 10662
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.899855937074269e-06,
+      "loss": 0.6797,
+      "step": 10663
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.898289009006717e-06,
+      "loss": 0.9648,
+      "step": 10664
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.89672210832095e-06,
+      "loss": 0.8984,
+      "step": 10665
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.89515523505591e-06,
+      "loss": 1.25,
+      "step": 10666
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.893588389250546e-06,
+      "loss": 0.9258,
+      "step": 10667
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.8920215709438e-06,
+      "loss": 1.2578,
+      "step": 10668
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.89045478017461e-06,
+      "loss": 0.6875,
+      "step": 10669
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.888888016981917e-06,
+      "loss": 1.0625,
+      "step": 10670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.887321281404663e-06,
+      "loss": 0.668,
+      "step": 10671
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.885754573481788e-06,
+      "loss": 1.5781,
+      "step": 10672
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.884187893252228e-06,
+      "loss": 0.5352,
+      "step": 10673
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.882621240754922e-06,
+      "loss": 0.6992,
+      "step": 10674
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.881054616028809e-06,
+      "loss": 1.1953,
+      "step": 10675
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.879488019112827e-06,
+      "loss": 1.5,
+      "step": 10676
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.877921450045907e-06,
+      "loss": 1.2344,
+      "step": 10677
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.876354908866985e-06,
+      "loss": 1.2266,
+      "step": 10678
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.874788395615e-06,
+      "loss": 0.3848,
+      "step": 10679
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.873221910328884e-06,
+      "loss": 1.2422,
+      "step": 10680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.871655453047571e-06,
+      "loss": 0.3418,
+      "step": 10681
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.870089023809988e-06,
+      "loss": 0.3027,
+      "step": 10682
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.868522622655073e-06,
+      "loss": 1.0391,
+      "step": 10683
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.866956249621756e-06,
+      "loss": 0.9102,
+      "step": 10684
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.865389904748967e-06,
+      "loss": 0.6172,
+      "step": 10685
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.863823588075633e-06,
+      "loss": 1.5781,
+      "step": 10686
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.862257299640689e-06,
+      "loss": 0.793,
+      "step": 10687
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.860691039483059e-06,
+      "loss": 0.7969,
+      "step": 10688
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.85912480764167e-06,
+      "loss": 0.6172,
+      "step": 10689
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.85755860415545e-06,
+      "loss": 0.7109,
+      "step": 10690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.855992429063326e-06,
+      "loss": 1.3203,
+      "step": 10691
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.854426282404226e-06,
+      "loss": 0.7031,
+      "step": 10692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.85286016421707e-06,
+      "loss": 0.5938,
+      "step": 10693
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.851294074540781e-06,
+      "loss": 0.5742,
+      "step": 10694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.849728013414289e-06,
+      "loss": 0.9922,
+      "step": 10695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.848161980876512e-06,
+      "loss": 1.0078,
+      "step": 10696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.846595976966374e-06,
+      "loss": 1.0078,
+      "step": 10697
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.845030001722793e-06,
+      "loss": 0.5859,
+      "step": 10698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.843464055184694e-06,
+      "loss": 1.3125,
+      "step": 10699
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.841898137390994e-06,
+      "loss": 1.1484,
+      "step": 10700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.840332248380613e-06,
+      "loss": 1.7969,
+      "step": 10701
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.838766388192468e-06,
+      "loss": 1.3672,
+      "step": 10702
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.837200556865481e-06,
+      "loss": 0.6289,
+      "step": 10703
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.835634754438565e-06,
+      "loss": 1.125,
+      "step": 10704
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.834068980950636e-06,
+      "loss": 1.0625,
+      "step": 10705
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.83250323644061e-06,
+      "loss": 1.3594,
+      "step": 10706
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.830937520947404e-06,
+      "loss": 0.7578,
+      "step": 10707
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.82937183450993e-06,
+      "loss": 0.3652,
+      "step": 10708
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.827806177167104e-06,
+      "loss": 1.0547,
+      "step": 10709
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.826240548957835e-06,
+      "loss": 0.9258,
+      "step": 10710
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.824674949921036e-06,
+      "loss": 1.3516,
+      "step": 10711
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.82310938009562e-06,
+      "loss": 0.9336,
+      "step": 10712
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.821543839520496e-06,
+      "loss": 1.2812,
+      "step": 10713
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.819978328234571e-06,
+      "loss": 1.1406,
+      "step": 10714
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.818412846276763e-06,
+      "loss": 0.6875,
+      "step": 10715
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.81684739368597e-06,
+      "loss": 0.9336,
+      "step": 10716
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.815281970501106e-06,
+      "loss": 0.6406,
+      "step": 10717
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.813716576761073e-06,
+      "loss": 1.0,
+      "step": 10718
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.812151212504782e-06,
+      "loss": 0.5898,
+      "step": 10719
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.81058587777114e-06,
+      "loss": 0.6289,
+      "step": 10720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.809020572599042e-06,
+      "loss": 0.7852,
+      "step": 10721
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.807455297027399e-06,
+      "loss": 1.3438,
+      "step": 10722
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.805890051095114e-06,
+      "loss": 1.0,
+      "step": 10723
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.804324834841087e-06,
+      "loss": 1.6484,
+      "step": 10724
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.802759648304226e-06,
+      "loss": 0.5117,
+      "step": 10725
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.80119449152342e-06,
+      "loss": 1.4062,
+      "step": 10726
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.79962936453758e-06,
+      "loss": 0.3203,
+      "step": 10727
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.7980642673856e-06,
+      "loss": 0.6484,
+      "step": 10728
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.796499200106382e-06,
+      "loss": 1.0781,
+      "step": 10729
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.794934162738825e-06,
+      "loss": 1.1953,
+      "step": 10730
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.793369155321817e-06,
+      "loss": 0.1738,
+      "step": 10731
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.791804177894265e-06,
+      "loss": 0.1328,
+      "step": 10732
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.79023923049506e-06,
+      "loss": 1.6875,
+      "step": 10733
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.788674313163098e-06,
+      "loss": 0.8945,
+      "step": 10734
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.78710942593727e-06,
+      "loss": 0.9062,
+      "step": 10735
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.78554456885648e-06,
+      "loss": 0.6914,
+      "step": 10736
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.78397974195961e-06,
+      "loss": 0.4316,
+      "step": 10737
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.782414945285555e-06,
+      "loss": 0.3711,
+      "step": 10738
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.780850178873203e-06,
+      "loss": 0.9062,
+      "step": 10739
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.779285442761453e-06,
+      "loss": 1.6016,
+      "step": 10740
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.77772073698919e-06,
+      "loss": 0.9023,
+      "step": 10741
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.776156061595303e-06,
+      "loss": 0.707,
+      "step": 10742
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.774591416618677e-06,
+      "loss": 1.3047,
+      "step": 10743
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.773026802098203e-06,
+      "loss": 1.0938,
+      "step": 10744
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.771462218072771e-06,
+      "loss": 1.1797,
+      "step": 10745
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.76989766458126e-06,
+      "loss": 1.4766,
+      "step": 10746
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.768333141662558e-06,
+      "loss": 0.543,
+      "step": 10747
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.766768649355555e-06,
+      "loss": 0.9844,
+      "step": 10748
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.765204187699126e-06,
+      "loss": 0.6133,
+      "step": 10749
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.76363975673216e-06,
+      "loss": 1.1641,
+      "step": 10750
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.762075356493533e-06,
+      "loss": 0.6406,
+      "step": 10751
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.760510987022133e-06,
+      "loss": 1.1875,
+      "step": 10752
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.75894664835684e-06,
+      "loss": 0.2793,
+      "step": 10753
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.75738234053653e-06,
+      "loss": 0.9844,
+      "step": 10754
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.755818063600081e-06,
+      "loss": 0.9688,
+      "step": 10755
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.754253817586377e-06,
+      "loss": 0.6641,
+      "step": 10756
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.752689602534292e-06,
+      "loss": 0.3613,
+      "step": 10757
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.751125418482706e-06,
+      "loss": 1.2422,
+      "step": 10758
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.749561265470488e-06,
+      "loss": 1.0,
+      "step": 10759
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.747997143536521e-06,
+      "loss": 0.2773,
+      "step": 10760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.746433052719678e-06,
+      "loss": 0.7578,
+      "step": 10761
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.744868993058828e-06,
+      "loss": 1.0703,
+      "step": 10762
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.743304964592847e-06,
+      "loss": 0.5078,
+      "step": 10763
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.74174096736061e-06,
+      "loss": 1.5625,
+      "step": 10764
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.740177001400984e-06,
+      "loss": 1.0469,
+      "step": 10765
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.738613066752844e-06,
+      "loss": 0.6992,
+      "step": 10766
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.737049163455052e-06,
+      "loss": 1.5781,
+      "step": 10767
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.735485291546484e-06,
+      "loss": 1.2188,
+      "step": 10768
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.73392145106601e-06,
+      "loss": 0.4902,
+      "step": 10769
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.73235764205249e-06,
+      "loss": 1.5625,
+      "step": 10770
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.730793864544795e-06,
+      "loss": 0.957,
+      "step": 10771
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.729230118581792e-06,
+      "loss": 1.125,
+      "step": 10772
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.727666404202343e-06,
+      "loss": 1.0547,
+      "step": 10773
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.726102721445318e-06,
+      "loss": 0.8711,
+      "step": 10774
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.724539070349572e-06,
+      "loss": 0.8594,
+      "step": 10775
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.722975450953974e-06,
+      "loss": 0.9922,
+      "step": 10776
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.721411863297385e-06,
+      "loss": 0.7656,
+      "step": 10777
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.719848307418665e-06,
+      "loss": 0.6719,
+      "step": 10778
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.718284783356675e-06,
+      "loss": 0.3672,
+      "step": 10779
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.716721291150276e-06,
+      "loss": 0.4629,
+      "step": 10780
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.715157830838326e-06,
+      "loss": 0.8242,
+      "step": 10781
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.713594402459681e-06,
+      "loss": 1.2109,
+      "step": 10782
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.7120310060532e-06,
+      "loss": 0.5234,
+      "step": 10783
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.710467641657738e-06,
+      "loss": 0.2891,
+      "step": 10784
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.708904309312155e-06,
+      "loss": 0.6992,
+      "step": 10785
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.707341009055303e-06,
+      "loss": 1.1406,
+      "step": 10786
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.705777740926032e-06,
+      "loss": 0.8555,
+      "step": 10787
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.704214504963202e-06,
+      "loss": 0.2119,
+      "step": 10788
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.702651301205663e-06,
+      "loss": 1.1953,
+      "step": 10789
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.701088129692265e-06,
+      "loss": 1.6719,
+      "step": 10790
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.699524990461859e-06,
+      "loss": 0.9531,
+      "step": 10791
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.6979618835533e-06,
+      "loss": 0.8789,
+      "step": 10792
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.69639880900543e-06,
+      "loss": 0.9219,
+      "step": 10793
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.694835766857104e-06,
+      "loss": 0.6172,
+      "step": 10794
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.693272757147161e-06,
+      "loss": 0.8711,
+      "step": 10795
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.69170977991446e-06,
+      "loss": 0.8945,
+      "step": 10796
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.69014683519784e-06,
+      "loss": 1.1875,
+      "step": 10797
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.688583923036145e-06,
+      "loss": 0.6094,
+      "step": 10798
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.687021043468219e-06,
+      "loss": 0.9023,
+      "step": 10799
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.68545819653291e-06,
+      "loss": 0.9844,
+      "step": 10800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.683895382269059e-06,
+      "loss": 0.5781,
+      "step": 10801
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.682332600715508e-06,
+      "loss": 1.1172,
+      "step": 10802
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.680769851911097e-06,
+      "loss": 0.543,
+      "step": 10803
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.679207135894665e-06,
+      "loss": 0.6875,
+      "step": 10804
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.677644452705056e-06,
+      "loss": 0.5742,
+      "step": 10805
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.676081802381107e-06,
+      "loss": 0.9727,
+      "step": 10806
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.674519184961657e-06,
+      "loss": 0.5547,
+      "step": 10807
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.672956600485535e-06,
+      "loss": 0.9336,
+      "step": 10808
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.671394048991588e-06,
+      "loss": 1.4844,
+      "step": 10809
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.669831530518647e-06,
+      "loss": 1.3125,
+      "step": 10810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.668269045105546e-06,
+      "loss": 0.9062,
+      "step": 10811
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.666706592791118e-06,
+      "loss": 1.0312,
+      "step": 10812
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.6651441736142e-06,
+      "loss": 0.9102,
+      "step": 10813
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.663581787613624e-06,
+      "loss": 0.4355,
+      "step": 10814
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.662019434828217e-06,
+      "loss": 0.7109,
+      "step": 10815
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.660457115296809e-06,
+      "loss": 0.6602,
+      "step": 10816
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.658894829058234e-06,
+      "loss": 1.2891,
+      "step": 10817
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.657332576151321e-06,
+      "loss": 0.9688,
+      "step": 10818
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.655770356614897e-06,
+      "loss": 0.543,
+      "step": 10819
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.654208170487784e-06,
+      "loss": 0.6797,
+      "step": 10820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.652646017808815e-06,
+      "loss": 0.8125,
+      "step": 10821
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.651083898616814e-06,
+      "loss": 0.7734,
+      "step": 10822
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.649521812950605e-06,
+      "loss": 0.498,
+      "step": 10823
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.647959760849009e-06,
+      "loss": 0.3203,
+      "step": 10824
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.646397742350858e-06,
+      "loss": 0.6133,
+      "step": 10825
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.644835757494966e-06,
+      "loss": 0.8633,
+      "step": 10826
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.643273806320156e-06,
+      "loss": 0.2949,
+      "step": 10827
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.641711888865246e-06,
+      "loss": 0.8984,
+      "step": 10828
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.640150005169062e-06,
+      "loss": 0.375,
+      "step": 10829
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.638588155270422e-06,
+      "loss": 0.6562,
+      "step": 10830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.637026339208138e-06,
+      "loss": 0.8984,
+      "step": 10831
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.635464557021032e-06,
+      "loss": 0.7422,
+      "step": 10832
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.633902808747918e-06,
+      "loss": 0.3379,
+      "step": 10833
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.632341094427615e-06,
+      "loss": 1.125,
+      "step": 10834
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.630779414098936e-06,
+      "loss": 0.3438,
+      "step": 10835
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.629217767800691e-06,
+      "loss": 1.8125,
+      "step": 10836
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.627656155571699e-06,
+      "loss": 1.0703,
+      "step": 10837
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.626094577450767e-06,
+      "loss": 1.1016,
+      "step": 10838
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.624533033476711e-06,
+      "loss": 1.5938,
+      "step": 10839
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.622971523688334e-06,
+      "loss": 0.6406,
+      "step": 10840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.621410048124457e-06,
+      "loss": 1.0625,
+      "step": 10841
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.619848606823881e-06,
+      "loss": 0.9844,
+      "step": 10842
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.618287199825414e-06,
+      "loss": 1.3203,
+      "step": 10843
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.616725827167861e-06,
+      "loss": 0.5234,
+      "step": 10844
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.615164488890036e-06,
+      "loss": 0.5703,
+      "step": 10845
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.613603185030737e-06,
+      "loss": 0.1826,
+      "step": 10846
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.612041915628774e-06,
+      "loss": 0.7227,
+      "step": 10847
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.610480680722943e-06,
+      "loss": 0.6836,
+      "step": 10848
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.608919480352054e-06,
+      "loss": 0.1602,
+      "step": 10849
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.607358314554908e-06,
+      "loss": 0.4688,
+      "step": 10850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.605797183370302e-06,
+      "loss": 0.75,
+      "step": 10851
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.604236086837037e-06,
+      "loss": 1.0938,
+      "step": 10852
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.602675024993918e-06,
+      "loss": 1.3281,
+      "step": 10853
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.601113997879737e-06,
+      "loss": 1.2344,
+      "step": 10854
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.599553005533295e-06,
+      "loss": 0.4316,
+      "step": 10855
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.597992047993383e-06,
+      "loss": 0.4863,
+      "step": 10856
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.596431125298805e-06,
+      "loss": 0.8555,
+      "step": 10857
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.594870237488354e-06,
+      "loss": 0.8789,
+      "step": 10858
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.59330938460082e-06,
+      "loss": 0.5547,
+      "step": 10859
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.591748566674997e-06,
+      "loss": 0.9531,
+      "step": 10860
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.59018778374968e-06,
+      "loss": 1.3828,
+      "step": 10861
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.588627035863661e-06,
+      "loss": 0.6484,
+      "step": 10862
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.587066323055732e-06,
+      "loss": 0.8633,
+      "step": 10863
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.585505645364673e-06,
+      "loss": 0.7852,
+      "step": 10864
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.583945002829284e-06,
+      "loss": 0.7969,
+      "step": 10865
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.582384395488347e-06,
+      "loss": 0.9961,
+      "step": 10866
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.580823823380654e-06,
+      "loss": 0.9766,
+      "step": 10867
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.579263286544983e-06,
+      "loss": 0.9766,
+      "step": 10868
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.577702785020132e-06,
+      "loss": 0.4082,
+      "step": 10869
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.576142318844874e-06,
+      "loss": 1.0234,
+      "step": 10870
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.574581888057996e-06,
+      "loss": 0.6602,
+      "step": 10871
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.57302149269828e-06,
+      "loss": 1.0234,
+      "step": 10872
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.571461132804513e-06,
+      "loss": 0.8945,
+      "step": 10873
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.569900808415472e-06,
+      "loss": 1.1094,
+      "step": 10874
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.568340519569937e-06,
+      "loss": 0.793,
+      "step": 10875
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.566780266306685e-06,
+      "loss": 0.582,
+      "step": 10876
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.5652200486645e-06,
+      "loss": 0.8594,
+      "step": 10877
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.563659866682155e-06,
+      "loss": 0.7148,
+      "step": 10878
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.562099720398431e-06,
+      "loss": 0.5781,
+      "step": 10879
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.5605396098521e-06,
+      "loss": 0.0767,
+      "step": 10880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.558979535081932e-06,
+      "loss": 1.4844,
+      "step": 10881
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.557419496126712e-06,
+      "loss": 0.4141,
+      "step": 10882
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.555859493025205e-06,
+      "loss": 0.9531,
+      "step": 10883
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.554299525816188e-06,
+      "loss": 1.0703,
+      "step": 10884
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.552739594538426e-06,
+      "loss": 0.6016,
+      "step": 10885
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.551179699230698e-06,
+      "loss": 0.2422,
+      "step": 10886
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.549619839931766e-06,
+      "loss": 1.1641,
+      "step": 10887
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.548060016680402e-06,
+      "loss": 0.4473,
+      "step": 10888
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.546500229515368e-06,
+      "loss": 0.7812,
+      "step": 10889
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.54494047847544e-06,
+      "loss": 1.2266,
+      "step": 10890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.543380763599381e-06,
+      "loss": 0.5273,
+      "step": 10891
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.541821084925953e-06,
+      "loss": 0.2471,
+      "step": 10892
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.540261442493918e-06,
+      "loss": 0.9805,
+      "step": 10893
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.538701836342046e-06,
+      "loss": 0.8242,
+      "step": 10894
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.537142266509095e-06,
+      "loss": 0.8398,
+      "step": 10895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.53558273303383e-06,
+      "loss": 1.4609,
+      "step": 10896
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.534023235955003e-06,
+      "loss": 0.9102,
+      "step": 10897
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.532463775311383e-06,
+      "loss": 0.5508,
+      "step": 10898
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.530904351141724e-06,
+      "loss": 0.8789,
+      "step": 10899
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.529344963484784e-06,
+      "loss": 1.0,
+      "step": 10900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.527785612379319e-06,
+      "loss": 0.7109,
+      "step": 10901
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.52622629786409e-06,
+      "loss": 0.9883,
+      "step": 10902
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.524667019977846e-06,
+      "loss": 0.3047,
+      "step": 10903
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.523107778759345e-06,
+      "loss": 1.5547,
+      "step": 10904
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.521548574247334e-06,
+      "loss": 1.1094,
+      "step": 10905
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.519989406480574e-06,
+      "loss": 0.7227,
+      "step": 10906
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.518430275497813e-06,
+      "loss": 0.582,
+      "step": 10907
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.5168711813378e-06,
+      "loss": 1.0156,
+      "step": 10908
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.51531212403928e-06,
+      "loss": 0.8789,
+      "step": 10909
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.513753103641013e-06,
+      "loss": 1.1484,
+      "step": 10910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.512194120181739e-06,
+      "loss": 1.2891,
+      "step": 10911
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.510635173700208e-06,
+      "loss": 0.7852,
+      "step": 10912
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.509076264235157e-06,
+      "loss": 0.8281,
+      "step": 10913
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.507517391825347e-06,
+      "loss": 1.1172,
+      "step": 10914
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.505958556509508e-06,
+      "loss": 0.9492,
+      "step": 10915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.504399758326388e-06,
+      "loss": 0.3027,
+      "step": 10916
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.50284099731473e-06,
+      "loss": 1.2578,
+      "step": 10917
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.501282273513276e-06,
+      "loss": 0.6914,
+      "step": 10918
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.499723586960767e-06,
+      "loss": 0.4648,
+      "step": 10919
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.498164937695938e-06,
+      "loss": 0.7969,
+      "step": 10920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.496606325757525e-06,
+      "loss": 1.0859,
+      "step": 10921
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.495047751184277e-06,
+      "loss": 1.5234,
+      "step": 10922
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.493489214014921e-06,
+      "loss": 0.4688,
+      "step": 10923
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.4919307142882e-06,
+      "loss": 0.4961,
+      "step": 10924
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.490372252042838e-06,
+      "loss": 1.0781,
+      "step": 10925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.488813827317579e-06,
+      "loss": 1.1797,
+      "step": 10926
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.487255440151152e-06,
+      "loss": 0.707,
+      "step": 10927
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.485697090582287e-06,
+      "loss": 1.2812,
+      "step": 10928
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.484138778649717e-06,
+      "loss": 0.2754,
+      "step": 10929
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.482580504392177e-06,
+      "loss": 0.875,
+      "step": 10930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.481022267848388e-06,
+      "loss": 1.0,
+      "step": 10931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.479464069057082e-06,
+      "loss": 0.3086,
+      "step": 10932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.477905908056985e-06,
+      "loss": 1.4297,
+      "step": 10933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.476347784886825e-06,
+      "loss": 0.7539,
+      "step": 10934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.474789699585328e-06,
+      "loss": 0.5039,
+      "step": 10935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.473231652191216e-06,
+      "loss": 0.4512,
+      "step": 10936
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.471673642743212e-06,
+      "loss": 1.5625,
+      "step": 10937
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.470115671280041e-06,
+      "loss": 0.875,
+      "step": 10938
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.468557737840425e-06,
+      "loss": 0.4473,
+      "step": 10939
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.466999842463085e-06,
+      "loss": 0.8633,
+      "step": 10940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.465441985186736e-06,
+      "loss": 0.8438,
+      "step": 10941
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.463884166050101e-06,
+      "loss": 0.2852,
+      "step": 10942
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.462326385091896e-06,
+      "loss": 1.3984,
+      "step": 10943
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.46076864235084e-06,
+      "loss": 1.0781,
+      "step": 10944
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.459210937865644e-06,
+      "loss": 0.7305,
+      "step": 10945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.457653271675034e-06,
+      "loss": 1.6719,
+      "step": 10946
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.456095643817712e-06,
+      "loss": 1.2188,
+      "step": 10947
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.454538054332396e-06,
+      "loss": 0.7969,
+      "step": 10948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.452980503257797e-06,
+      "loss": 1.2109,
+      "step": 10949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.451422990632628e-06,
+      "loss": 1.3203,
+      "step": 10950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.449865516495598e-06,
+      "loss": 0.9297,
+      "step": 10951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.448308080885421e-06,
+      "loss": 1.7422,
+      "step": 10952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.446750683840798e-06,
+      "loss": 1.6328,
+      "step": 10953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.445193325400435e-06,
+      "loss": 1.0781,
+      "step": 10954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.443636005603048e-06,
+      "loss": 1.0,
+      "step": 10955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.442078724487336e-06,
+      "loss": 0.832,
+      "step": 10956
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.440521482092006e-06,
+      "loss": 1.5625,
+      "step": 10957
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.438964278455756e-06,
+      "loss": 0.3418,
+      "step": 10958
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.437407113617294e-06,
+      "loss": 1.3828,
+      "step": 10959
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.435849987615321e-06,
+      "loss": 1.3047,
+      "step": 10960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.434292900488537e-06,
+      "loss": 0.9258,
+      "step": 10961
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.43273585227564e-06,
+      "loss": 0.6836,
+      "step": 10962
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.431178843015332e-06,
+      "loss": 0.5039,
+      "step": 10963
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.42962187274631e-06,
+      "loss": 1.0078,
+      "step": 10964
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.428064941507266e-06,
+      "loss": 1.1875,
+      "step": 10965
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.4265080493369e-06,
+      "loss": 1.0547,
+      "step": 10966
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.424951196273908e-06,
+      "loss": 1.3203,
+      "step": 10967
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.423394382356983e-06,
+      "loss": 0.9727,
+      "step": 10968
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.421837607624815e-06,
+      "loss": 0.8555,
+      "step": 10969
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.420280872116095e-06,
+      "loss": 0.8789,
+      "step": 10970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.41872417586952e-06,
+      "loss": 0.9453,
+      "step": 10971
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.417167518923777e-06,
+      "loss": 1.2188,
+      "step": 10972
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.415610901317556e-06,
+      "loss": 0.416,
+      "step": 10973
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.414054323089539e-06,
+      "loss": 1.0703,
+      "step": 10974
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.412497784278419e-06,
+      "loss": 0.9141,
+      "step": 10975
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.410941284922883e-06,
+      "loss": 0.5859,
+      "step": 10976
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.409384825061611e-06,
+      "loss": 0.8516,
+      "step": 10977
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.407828404733289e-06,
+      "loss": 0.8047,
+      "step": 10978
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.406272023976605e-06,
+      "loss": 0.2793,
+      "step": 10979
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.404715682830234e-06,
+      "loss": 0.9531,
+      "step": 10980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.40315938133286e-06,
+      "loss": 0.0898,
+      "step": 10981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.40160311952316e-06,
+      "loss": 1.0312,
+      "step": 10982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.40004689743982e-06,
+      "loss": 0.3301,
+      "step": 10983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.398490715121515e-06,
+      "loss": 0.8594,
+      "step": 10984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.396934572606922e-06,
+      "loss": 0.918,
+      "step": 10985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.395378469934712e-06,
+      "loss": 0.6641,
+      "step": 10986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.393822407143568e-06,
+      "loss": 0.1416,
+      "step": 10987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.392266384272161e-06,
+      "loss": 0.2715,
+      "step": 10988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.390710401359164e-06,
+      "loss": 1.6719,
+      "step": 10989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.389154458443247e-06,
+      "loss": 1.2188,
+      "step": 10990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.387598555563088e-06,
+      "loss": 0.9023,
+      "step": 10991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.386042692757352e-06,
+      "loss": 1.0469,
+      "step": 10992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.384486870064707e-06,
+      "loss": 0.6836,
+      "step": 10993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.382931087523822e-06,
+      "loss": 1.0859,
+      "step": 10994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.381375345173369e-06,
+      "loss": 1.0781,
+      "step": 10995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.379819643052012e-06,
+      "loss": 0.8711,
+      "step": 10996
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.378263981198412e-06,
+      "loss": 1.5703,
+      "step": 10997
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.376708359651234e-06,
+      "loss": 0.7109,
+      "step": 10998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.375152778449146e-06,
+      "loss": 0.252,
+      "step": 10999
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.373597237630808e-06,
+      "loss": 0.5859,
+      "step": 11000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.372041737234885e-06,
+      "loss": 1.2969,
+      "step": 11001
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.370486277300026e-06,
+      "loss": 1.6641,
+      "step": 11002
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.3689308578649e-06,
+      "loss": 0.5625,
+      "step": 11003
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.367375478968162e-06,
+      "loss": 0.5039,
+      "step": 11004
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.365820140648471e-06,
+      "loss": 0.957,
+      "step": 11005
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.364264842944479e-06,
+      "loss": 1.3281,
+      "step": 11006
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.362709585894848e-06,
+      "loss": 1.0234,
+      "step": 11007
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.361154369538226e-06,
+      "loss": 1.0,
+      "step": 11008
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.35959919391327e-06,
+      "loss": 0.7812,
+      "step": 11009
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.358044059058628e-06,
+      "loss": 0.9023,
+      "step": 11010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.356488965012954e-06,
+      "loss": 0.8516,
+      "step": 11011
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.3549339118149e-06,
+      "loss": 1.2578,
+      "step": 11012
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.353378899503113e-06,
+      "loss": 1.1172,
+      "step": 11013
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.351823928116236e-06,
+      "loss": 0.793,
+      "step": 11014
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.350268997692925e-06,
+      "loss": 1.1094,
+      "step": 11015
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.348714108271822e-06,
+      "loss": 0.8633,
+      "step": 11016
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.347159259891572e-06,
+      "loss": 0.4766,
+      "step": 11017
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.345604452590816e-06,
+      "loss": 0.5273,
+      "step": 11018
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.344049686408205e-06,
+      "loss": 0.8984,
+      "step": 11019
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.342494961382374e-06,
+      "loss": 0.6602,
+      "step": 11020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.340940277551966e-06,
+      "loss": 0.793,
+      "step": 11021
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.339385634955619e-06,
+      "loss": 1.3047,
+      "step": 11022
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.337831033631973e-06,
+      "loss": 0.6836,
+      "step": 11023
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.336276473619673e-06,
+      "loss": 0.832,
+      "step": 11024
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.334721954957344e-06,
+      "loss": 1.0391,
+      "step": 11025
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.333167477683629e-06,
+      "loss": 0.457,
+      "step": 11026
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.331613041837158e-06,
+      "loss": 1.4141,
+      "step": 11027
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.33005864745657e-06,
+      "loss": 1.75,
+      "step": 11028
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.3285042945805e-06,
+      "loss": 0.4395,
+      "step": 11029
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.32694998324757e-06,
+      "loss": 0.7773,
+      "step": 11030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.325395713496415e-06,
+      "loss": 0.9961,
+      "step": 11031
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.32384148536567e-06,
+      "loss": 0.7812,
+      "step": 11032
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.322287298893957e-06,
+      "loss": 0.4512,
+      "step": 11033
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.320733154119906e-06,
+      "loss": 0.8477,
+      "step": 11034
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.319179051082142e-06,
+      "loss": 1.0156,
+      "step": 11035
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.317624989819292e-06,
+      "loss": 0.5977,
+      "step": 11036
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.31607097036998e-06,
+      "loss": 0.1147,
+      "step": 11037
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.314516992772832e-06,
+      "loss": 0.8281,
+      "step": 11038
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.312963057066463e-06,
+      "loss": 1.4609,
+      "step": 11039
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.311409163289505e-06,
+      "loss": 0.668,
+      "step": 11040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.30985531148057e-06,
+      "loss": 0.8477,
+      "step": 11041
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.30830150167828e-06,
+      "loss": 1.2344,
+      "step": 11042
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.306747733921248e-06,
+      "loss": 0.8594,
+      "step": 11043
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.3051940082481e-06,
+      "loss": 1.2344,
+      "step": 11044
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.303640324697452e-06,
+      "loss": 1.0781,
+      "step": 11045
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.302086683307912e-06,
+      "loss": 1.3828,
+      "step": 11046
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.300533084118095e-06,
+      "loss": 0.6172,
+      "step": 11047
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.298979527166619e-06,
+      "loss": 1.4297,
+      "step": 11048
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.297426012492092e-06,
+      "loss": 0.6133,
+      "step": 11049
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.295872540133129e-06,
+      "loss": 0.7148,
+      "step": 11050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.29431911012833e-06,
+      "loss": 1.0859,
+      "step": 11051
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.292765722516316e-06,
+      "loss": 0.7266,
+      "step": 11052
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.291212377335687e-06,
+      "loss": 0.4199,
+      "step": 11053
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.289659074625054e-06,
+      "loss": 1.2188,
+      "step": 11054
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.288105814423017e-06,
+      "loss": 1.4141,
+      "step": 11055
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.286552596768188e-06,
+      "loss": 0.8984,
+      "step": 11056
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.284999421699166e-06,
+      "loss": 0.8789,
+      "step": 11057
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.283446289254554e-06,
+      "loss": 1.2344,
+      "step": 11058
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.28189319947295e-06,
+      "loss": 0.6914,
+      "step": 11059
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.280340152392959e-06,
+      "loss": 0.3613,
+      "step": 11060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.278787148053179e-06,
+      "loss": 1.4688,
+      "step": 11061
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.27723418649221e-06,
+      "loss": 1.2109,
+      "step": 11062
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.275681267748643e-06,
+      "loss": 0.9805,
+      "step": 11063
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.274128391861081e-06,
+      "loss": 0.918,
+      "step": 11064
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.272575558868116e-06,
+      "loss": 1.1641,
+      "step": 11065
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.271022768808343e-06,
+      "loss": 1.1406,
+      "step": 11066
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.26947002172035e-06,
+      "loss": 0.8281,
+      "step": 11067
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.267917317642738e-06,
+      "loss": 1.2734,
+      "step": 11068
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.26636465661409e-06,
+      "loss": 0.9336,
+      "step": 11069
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.264812038672997e-06,
+      "loss": 0.5234,
+      "step": 11070
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.263259463858048e-06,
+      "loss": 0.9727,
+      "step": 11071
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.261706932207832e-06,
+      "loss": 1.5156,
+      "step": 11072
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.260154443760937e-06,
+      "loss": 0.543,
+      "step": 11073
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.258601998555944e-06,
+      "loss": 0.2754,
+      "step": 11074
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.257049596631437e-06,
+      "loss": 0.2314,
+      "step": 11075
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.255497238026003e-06,
+      "loss": 0.4082,
+      "step": 11076
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.253944922778223e-06,
+      "loss": 1.7422,
+      "step": 11077
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.25239265092668e-06,
+      "loss": 0.8906,
+      "step": 11078
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.250840422509946e-06,
+      "loss": 0.5469,
+      "step": 11079
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.24928823756661e-06,
+      "loss": 1.2969,
+      "step": 11080
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.247736096135244e-06,
+      "loss": 1.3828,
+      "step": 11081
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.246183998254426e-06,
+      "loss": 0.1904,
+      "step": 11082
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.24463194396273e-06,
+      "loss": 0.5273,
+      "step": 11083
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.243079933298731e-06,
+      "loss": 0.9492,
+      "step": 11084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.241527966301008e-06,
+      "loss": 2.0781,
+      "step": 11085
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.239976043008128e-06,
+      "loss": 0.8047,
+      "step": 11086
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.238424163458661e-06,
+      "loss": 0.7305,
+      "step": 11087
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.23687232769118e-06,
+      "loss": 0.9453,
+      "step": 11088
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.235320535744255e-06,
+      "loss": 1.2969,
+      "step": 11089
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.233768787656454e-06,
+      "loss": 0.6523,
+      "step": 11090
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.232217083466339e-06,
+      "loss": 2.0312,
+      "step": 11091
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.230665423212482e-06,
+      "loss": 1.6562,
+      "step": 11092
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.229113806933445e-06,
+      "loss": 0.7617,
+      "step": 11093
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.227562234667791e-06,
+      "loss": 0.6953,
+      "step": 11094
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.226010706454084e-06,
+      "loss": 0.6523,
+      "step": 11095
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.224459222330887e-06,
+      "loss": 0.9023,
+      "step": 11096
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.222907782336757e-06,
+      "loss": 0.8945,
+      "step": 11097
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.221356386510255e-06,
+      "loss": 1.0625,
+      "step": 11098
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.219805034889937e-06,
+      "loss": 0.9062,
+      "step": 11099
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.218253727514363e-06,
+      "loss": 0.8438,
+      "step": 11100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.216702464422091e-06,
+      "loss": 1.3672,
+      "step": 11101
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.21515124565167e-06,
+      "loss": 0.4863,
+      "step": 11102
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.213600071241658e-06,
+      "loss": 0.4824,
+      "step": 11103
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.212048941230604e-06,
+      "loss": 1.0312,
+      "step": 11104
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.210497855657063e-06,
+      "loss": 0.4492,
+      "step": 11105
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.208946814559589e-06,
+      "loss": 0.7891,
+      "step": 11106
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.207395817976722e-06,
+      "loss": 1.0703,
+      "step": 11107
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.205844865947014e-06,
+      "loss": 1.3594,
+      "step": 11108
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.204293958509017e-06,
+      "loss": 1.1797,
+      "step": 11109
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.202743095701272e-06,
+      "loss": 0.7891,
+      "step": 11110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.201192277562328e-06,
+      "loss": 0.8086,
+      "step": 11111
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.199641504130723e-06,
+      "loss": 1.0469,
+      "step": 11112
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.198090775445002e-06,
+      "loss": 0.6836,
+      "step": 11113
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.19654009154371e-06,
+      "loss": 1.0,
+      "step": 11114
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.194989452465383e-06,
+      "loss": 0.6992,
+      "step": 11115
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.193438858248562e-06,
+      "loss": 1.2969,
+      "step": 11116
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.19188830893179e-06,
+      "loss": 1.0703,
+      "step": 11117
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.190337804553598e-06,
+      "loss": 0.9492,
+      "step": 11118
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.188787345152522e-06,
+      "loss": 0.8555,
+      "step": 11119
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.187236930767098e-06,
+      "loss": 1.1094,
+      "step": 11120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.185686561435864e-06,
+      "loss": 0.9336,
+      "step": 11121
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.184136237197347e-06,
+      "loss": 1.6797,
+      "step": 11122
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.182585958090084e-06,
+      "loss": 1.1797,
+      "step": 11123
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.181035724152598e-06,
+      "loss": 1.5781,
+      "step": 11124
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.179485535423427e-06,
+      "loss": 0.543,
+      "step": 11125
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.17793539194109e-06,
+      "loss": 1.1719,
+      "step": 11126
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.176385293744122e-06,
+      "loss": 1.1406,
+      "step": 11127
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.174835240871045e-06,
+      "loss": 0.4883,
+      "step": 11128
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.173285233360387e-06,
+      "loss": 1.0391,
+      "step": 11129
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.171735271250667e-06,
+      "loss": 1.3594,
+      "step": 11130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.170185354580412e-06,
+      "loss": 0.1826,
+      "step": 11131
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.168635483388136e-06,
+      "loss": 0.5195,
+      "step": 11132
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.167085657712368e-06,
+      "loss": 1.9375,
+      "step": 11133
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.165535877591628e-06,
+      "loss": 0.7812,
+      "step": 11134
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.163986143064426e-06,
+      "loss": 0.3887,
+      "step": 11135
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.162436454169282e-06,
+      "loss": 1.3828,
+      "step": 11136
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.160886810944712e-06,
+      "loss": 0.4062,
+      "step": 11137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.159337213429234e-06,
+      "loss": 0.6172,
+      "step": 11138
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.15778766166136e-06,
+      "loss": 0.7852,
+      "step": 11139
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.156238155679596e-06,
+      "loss": 0.5859,
+      "step": 11140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.154688695522462e-06,
+      "loss": 1.0,
+      "step": 11141
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.153139281228462e-06,
+      "loss": 0.5742,
+      "step": 11142
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.15158991283611e-06,
+      "loss": 1.2891,
+      "step": 11143
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.150040590383908e-06,
+      "loss": 0.6328,
+      "step": 11144
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.14849131391037e-06,
+      "loss": 0.3105,
+      "step": 11145
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.146942083453995e-06,
+      "loss": 0.293,
+      "step": 11146
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.145392899053289e-06,
+      "loss": 1.125,
+      "step": 11147
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.143843760746755e-06,
+      "loss": 1.1562,
+      "step": 11148
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.142294668572899e-06,
+      "loss": 0.5625,
+      "step": 11149
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.140745622570219e-06,
+      "loss": 0.8516,
+      "step": 11150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.139196622777214e-06,
+      "loss": 0.6914,
+      "step": 11151
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.137647669232382e-06,
+      "loss": 0.457,
+      "step": 11152
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.136098761974223e-06,
+      "loss": 0.5078,
+      "step": 11153
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.13454990104123e-06,
+      "loss": 0.3613,
+      "step": 11154
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.133001086471904e-06,
+      "loss": 0.6602,
+      "step": 11155
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.131452318304733e-06,
+      "loss": 0.4453,
+      "step": 11156
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.129903596578217e-06,
+      "loss": 0.3281,
+      "step": 11157
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.12835492133084e-06,
+      "loss": 1.0938,
+      "step": 11158
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.126806292601095e-06,
+      "loss": 0.3809,
+      "step": 11159
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.12525771042747e-06,
+      "loss": 0.5781,
+      "step": 11160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.123709174848459e-06,
+      "loss": 0.7148,
+      "step": 11161
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.122160685902548e-06,
+      "loss": 0.6562,
+      "step": 11162
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.120612243628215e-06,
+      "loss": 1.0703,
+      "step": 11163
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.119063848063951e-06,
+      "loss": 0.6445,
+      "step": 11164
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.11751549924824e-06,
+      "loss": 0.418,
+      "step": 11165
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.115967197219565e-06,
+      "loss": 0.625,
+      "step": 11166
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.114418942016407e-06,
+      "loss": 1.3906,
+      "step": 11167
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.11287073367724e-06,
+      "loss": 0.9492,
+      "step": 11168
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.11132257224055e-06,
+      "loss": 0.5781,
+      "step": 11169
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.109774457744813e-06,
+      "loss": 0.9727,
+      "step": 11170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.108226390228505e-06,
+      "loss": 0.6523,
+      "step": 11171
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.1066783697301e-06,
+      "loss": 0.9805,
+      "step": 11172
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.105130396288076e-06,
+      "loss": 1.0625,
+      "step": 11173
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.103582469940905e-06,
+      "loss": 1.1953,
+      "step": 11174
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.102034590727057e-06,
+      "loss": 1.4141,
+      "step": 11175
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.100486758685005e-06,
+      "loss": 0.6719,
+      "step": 11176
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.098938973853214e-06,
+      "loss": 1.3203,
+      "step": 11177
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.097391236270161e-06,
+      "loss": 0.1582,
+      "step": 11178
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.095843545974306e-06,
+      "loss": 1.0312,
+      "step": 11179
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.094295903004116e-06,
+      "loss": 0.2695,
+      "step": 11180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.092748307398058e-06,
+      "loss": 1.4375,
+      "step": 11181
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.091200759194595e-06,
+      "loss": 0.4473,
+      "step": 11182
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.089653258432193e-06,
+      "loss": 1.5859,
+      "step": 11183
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.088105805149308e-06,
+      "loss": 1.5391,
+      "step": 11184
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.086558399384398e-06,
+      "loss": 0.3887,
+      "step": 11185
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.085011041175932e-06,
+      "loss": 0.9297,
+      "step": 11186
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.083463730562358e-06,
+      "loss": 0.9062,
+      "step": 11187
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.08191646758214e-06,
+      "loss": 0.7188,
+      "step": 11188
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.080369252273727e-06,
+      "loss": 0.9961,
+      "step": 11189
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.078822084675581e-06,
+      "loss": 0.707,
+      "step": 11190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.077274964826147e-06,
+      "loss": 0.9648,
+      "step": 11191
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.075727892763884e-06,
+      "loss": 1.2969,
+      "step": 11192
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.074180868527233e-06,
+      "loss": 1.7578,
+      "step": 11193
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.072633892154655e-06,
+      "loss": 1.5,
+      "step": 11194
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.071086963684593e-06,
+      "loss": 1.0078,
+      "step": 11195
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.069540083155495e-06,
+      "loss": 0.793,
+      "step": 11196
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.067993250605803e-06,
+      "loss": 0.5469,
+      "step": 11197
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.066446466073967e-06,
+      "loss": 0.9883,
+      "step": 11198
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.06489972959843e-06,
+      "loss": 0.6484,
+      "step": 11199
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.063353041217632e-06,
+      "loss": 1.1484,
+      "step": 11200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.061806400970014e-06,
+      "loss": 0.6094,
+      "step": 11201
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.06025980889402e-06,
+      "loss": 0.2793,
+      "step": 11202
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.058713265028085e-06,
+      "loss": 1.0312,
+      "step": 11203
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.057166769410646e-06,
+      "loss": 1.1172,
+      "step": 11204
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.05562032208014e-06,
+      "loss": 0.6055,
+      "step": 11205
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.054073923075009e-06,
+      "loss": 1.0391,
+      "step": 11206
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.052527572433678e-06,
+      "loss": 0.9336,
+      "step": 11207
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.050981270194585e-06,
+      "loss": 1.0391,
+      "step": 11208
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.049435016396156e-06,
+      "loss": 0.957,
+      "step": 11209
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.047888811076827e-06,
+      "loss": 0.8477,
+      "step": 11210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.04634265427503e-06,
+      "loss": 1.0781,
+      "step": 11211
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.044796546029183e-06,
+      "loss": 1.1484,
+      "step": 11212
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.043250486377717e-06,
+      "loss": 0.8125,
+      "step": 11213
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.041704475359063e-06,
+      "loss": 1.2812,
+      "step": 11214
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.040158513011639e-06,
+      "loss": 0.8438,
+      "step": 11215
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.038612599373872e-06,
+      "loss": 1.2656,
+      "step": 11216
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.037066734484178e-06,
+      "loss": 0.6875,
+      "step": 11217
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.035520918380986e-06,
+      "loss": 0.9531,
+      "step": 11218
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.03397515110271e-06,
+      "loss": 0.8672,
+      "step": 11219
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.032429432687769e-06,
+      "loss": 0.832,
+      "step": 11220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.03088376317458e-06,
+      "loss": 0.6641,
+      "step": 11221
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.029338142601561e-06,
+      "loss": 1.2422,
+      "step": 11222
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.027792571007129e-06,
+      "loss": 1.0859,
+      "step": 11223
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.02624704842969e-06,
+      "loss": 0.1758,
+      "step": 11224
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.024701574907659e-06,
+      "loss": 0.4355,
+      "step": 11225
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.02315615047945e-06,
+      "loss": 1.1328,
+      "step": 11226
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.02161077518347e-06,
+      "loss": 0.2734,
+      "step": 11227
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.020065449058131e-06,
+      "loss": 0.7539,
+      "step": 11228
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.018520172141833e-06,
+      "loss": 0.2773,
+      "step": 11229
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.01697494447299e-06,
+      "loss": 1.0547,
+      "step": 11230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.015429766090003e-06,
+      "loss": 0.4199,
+      "step": 11231
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.013884637031278e-06,
+      "loss": 0.4102,
+      "step": 11232
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.012339557335212e-06,
+      "loss": 0.1797,
+      "step": 11233
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.010794527040213e-06,
+      "loss": 0.2832,
+      "step": 11234
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.009249546184678e-06,
+      "loss": 0.1621,
+      "step": 11235
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.007704614807005e-06,
+      "loss": 0.8398,
+      "step": 11236
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.006159732945593e-06,
+      "loss": 0.4941,
+      "step": 11237
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.004614900638836e-06,
+      "loss": 0.4043,
+      "step": 11238
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.003070117925136e-06,
+      "loss": 0.8984,
+      "step": 11239
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.001525384842877e-06,
+      "loss": 0.3457,
+      "step": 11240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.999980701430454e-06,
+      "loss": 0.8086,
+      "step": 11241
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.998436067726266e-06,
+      "loss": 0.4941,
+      "step": 11242
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.996891483768694e-06,
+      "loss": 1.3516,
+      "step": 11243
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.995346949596134e-06,
+      "loss": 0.8789,
+      "step": 11244
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.993802465246966e-06,
+      "loss": 1.3203,
+      "step": 11245
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.992258030759583e-06,
+      "loss": 1.4609,
+      "step": 11246
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.990713646172367e-06,
+      "loss": 0.8008,
+      "step": 11247
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.989169311523703e-06,
+      "loss": 0.668,
+      "step": 11248
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.987625026851976e-06,
+      "loss": 1.1094,
+      "step": 11249
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.986080792195562e-06,
+      "loss": 1.2188,
+      "step": 11250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.984536607592845e-06,
+      "loss": 0.5977,
+      "step": 11251
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.982992473082202e-06,
+      "loss": 0.7031,
+      "step": 11252
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.981448388702013e-06,
+      "loss": 0.7227,
+      "step": 11253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.979904354490651e-06,
+      "loss": 0.9297,
+      "step": 11254
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.978360370486497e-06,
+      "loss": 1.1406,
+      "step": 11255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.976816436727921e-06,
+      "loss": 0.543,
+      "step": 11256
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.975272553253296e-06,
+      "loss": 0.7344,
+      "step": 11257
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.97372872010099e-06,
+      "loss": 0.9883,
+      "step": 11258
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.972184937309383e-06,
+      "loss": 0.707,
+      "step": 11259
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.970641204916833e-06,
+      "loss": 1.1016,
+      "step": 11260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.96909752296172e-06,
+      "loss": 0.9375,
+      "step": 11261
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.967553891482393e-06,
+      "loss": 1.1719,
+      "step": 11262
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.966010310517234e-06,
+      "loss": 0.4023,
+      "step": 11263
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.9644667801046e-06,
+      "loss": 0.5117,
+      "step": 11264
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.962923300282852e-06,
+      "loss": 1.1172,
+      "step": 11265
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.961379871090354e-06,
+      "loss": 0.7344,
+      "step": 11266
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.959836492565468e-06,
+      "loss": 0.9883,
+      "step": 11267
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.958293164746548e-06,
+      "loss": 0.7617,
+      "step": 11268
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.956749887671957e-06,
+      "loss": 0.8008,
+      "step": 11269
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.955206661380043e-06,
+      "loss": 0.625,
+      "step": 11270
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.953663485909173e-06,
+      "loss": 0.7891,
+      "step": 11271
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.952120361297695e-06,
+      "loss": 1.6328,
+      "step": 11272
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.950577287583959e-06,
+      "loss": 1.2109,
+      "step": 11273
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.949034264806317e-06,
+      "loss": 0.459,
+      "step": 11274
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.947491293003125e-06,
+      "loss": 0.4141,
+      "step": 11275
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.945948372212726e-06,
+      "loss": 0.7773,
+      "step": 11276
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.944405502473471e-06,
+      "loss": 1.1406,
+      "step": 11277
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.9428626838237e-06,
+      "loss": 0.3555,
+      "step": 11278
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.941319916301767e-06,
+      "loss": 0.6289,
+      "step": 11279
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.939777199946009e-06,
+      "loss": 1.0781,
+      "step": 11280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.93823453479477e-06,
+      "loss": 1.4688,
+      "step": 11281
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.936691920886394e-06,
+      "loss": 0.6992,
+      "step": 11282
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.93514935825922e-06,
+      "loss": 1.2891,
+      "step": 11283
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.933606846951584e-06,
+      "loss": 1.5938,
+      "step": 11284
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.932064387001825e-06,
+      "loss": 0.3848,
+      "step": 11285
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.930521978448278e-06,
+      "loss": 0.75,
+      "step": 11286
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.928979621329281e-06,
+      "loss": 0.5312,
+      "step": 11287
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.927437315683166e-06,
+      "loss": 0.4883,
+      "step": 11288
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.925895061548264e-06,
+      "loss": 1.1016,
+      "step": 11289
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.924352858962905e-06,
+      "loss": 0.4902,
+      "step": 11290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.922810707965423e-06,
+      "loss": 0.9258,
+      "step": 11291
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.921268608594142e-06,
+      "loss": 0.4785,
+      "step": 11292
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.919726560887392e-06,
+      "loss": 0.8398,
+      "step": 11293
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.918184564883495e-06,
+      "loss": 1.0312,
+      "step": 11294
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.916642620620784e-06,
+      "loss": 1.1328,
+      "step": 11295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.915100728137575e-06,
+      "loss": 1.1719,
+      "step": 11296
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.91355888747219e-06,
+      "loss": 0.459,
+      "step": 11297
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.912017098662951e-06,
+      "loss": 0.7422,
+      "step": 11298
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.91047536174818e-06,
+      "loss": 1.3203,
+      "step": 11299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.908933676766194e-06,
+      "loss": 0.707,
+      "step": 11300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.907392043755308e-06,
+      "loss": 1.8203,
+      "step": 11301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.905850462753835e-06,
+      "loss": 0.8047,
+      "step": 11302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.904308933800095e-06,
+      "loss": 0.6992,
+      "step": 11303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.902767456932399e-06,
+      "loss": 1.0391,
+      "step": 11304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.901226032189062e-06,
+      "loss": 1.1484,
+      "step": 11305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.899684659608386e-06,
+      "loss": 1.1719,
+      "step": 11306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.898143339228685e-06,
+      "loss": 1.5078,
+      "step": 11307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.896602071088269e-06,
+      "loss": 1.1562,
+      "step": 11308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.89506085522544e-06,
+      "loss": 1.2344,
+      "step": 11309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.893519691678504e-06,
+      "loss": 1.3125,
+      "step": 11310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.891978580485771e-06,
+      "loss": 1.1328,
+      "step": 11311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.890437521685535e-06,
+      "loss": 0.4551,
+      "step": 11312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.888896515316103e-06,
+      "loss": 0.957,
+      "step": 11313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.887355561415769e-06,
+      "loss": 0.8477,
+      "step": 11314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.88581466002284e-06,
+      "loss": 1.3047,
+      "step": 11315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.88427381117561e-06,
+      "loss": 0.625,
+      "step": 11316
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.882733014912371e-06,
+      "loss": 0.625,
+      "step": 11317
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.88119227127142e-06,
+      "loss": 1.0156,
+      "step": 11318
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.879651580291053e-06,
+      "loss": 1.1328,
+      "step": 11319
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.878110942009561e-06,
+      "loss": 0.9922,
+      "step": 11320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.876570356465237e-06,
+      "loss": 0.7656,
+      "step": 11321
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.875029823696362e-06,
+      "loss": 1.4375,
+      "step": 11322
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.873489343741234e-06,
+      "loss": 1.0781,
+      "step": 11323
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.871948916638133e-06,
+      "loss": 1.5156,
+      "step": 11324
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.87040854242535e-06,
+      "loss": 0.9062,
+      "step": 11325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.868868221141166e-06,
+      "loss": 1.3359,
+      "step": 11326
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.867327952823863e-06,
+      "loss": 1.0078,
+      "step": 11327
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.865787737511729e-06,
+      "loss": 0.7812,
+      "step": 11328
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.864247575243036e-06,
+      "loss": 0.8516,
+      "step": 11329
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.862707466056068e-06,
+      "loss": 0.2031,
+      "step": 11330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.861167409989098e-06,
+      "loss": 0.4121,
+      "step": 11331
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.85962740708041e-06,
+      "loss": 0.2227,
+      "step": 11332
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.858087457368275e-06,
+      "loss": 0.9453,
+      "step": 11333
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.856547560890964e-06,
+      "loss": 1.0469,
+      "step": 11334
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.855007717686752e-06,
+      "loss": 0.3867,
+      "step": 11335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.85346792779391e-06,
+      "loss": 0.5742,
+      "step": 11336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.851928191250709e-06,
+      "loss": 0.7617,
+      "step": 11337
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.850388508095416e-06,
+      "loss": 0.2754,
+      "step": 11338
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.848848878366298e-06,
+      "loss": 1.3672,
+      "step": 11339
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.84730930210162e-06,
+      "loss": 0.6133,
+      "step": 11340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.845769779339648e-06,
+      "loss": 0.7383,
+      "step": 11341
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.844230310118644e-06,
+      "loss": 1.1562,
+      "step": 11342
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.84269089447687e-06,
+      "loss": 0.8711,
+      "step": 11343
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.84115153245259e-06,
+      "loss": 1.0391,
+      "step": 11344
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.839612224084057e-06,
+      "loss": 1.1094,
+      "step": 11345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.838072969409532e-06,
+      "loss": 1.1953,
+      "step": 11346
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.836533768467269e-06,
+      "loss": 0.3105,
+      "step": 11347
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.834994621295527e-06,
+      "loss": 0.4766,
+      "step": 11348
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.83345552793256e-06,
+      "loss": 0.8125,
+      "step": 11349
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.831916488416618e-06,
+      "loss": 1.2891,
+      "step": 11350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.830377502785948e-06,
+      "loss": 1.2578,
+      "step": 11351
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.828838571078808e-06,
+      "loss": 0.6172,
+      "step": 11352
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.827299693333443e-06,
+      "loss": 1.3281,
+      "step": 11353
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.8257608695881e-06,
+      "loss": 0.6602,
+      "step": 11354
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.824222099881021e-06,
+      "loss": 0.832,
+      "step": 11355
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.822683384250456e-06,
+      "loss": 1.0859,
+      "step": 11356
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.821144722734648e-06,
+      "loss": 1.3047,
+      "step": 11357
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.819606115371834e-06,
+      "loss": 1.4688,
+      "step": 11358
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.818067562200255e-06,
+      "loss": 0.5586,
+      "step": 11359
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.816529063258155e-06,
+      "loss": 0.668,
+      "step": 11360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.81499061858377e-06,
+      "loss": 0.377,
+      "step": 11361
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.813452228215332e-06,
+      "loss": 0.8945,
+      "step": 11362
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.811913892191078e-06,
+      "loss": 0.7852,
+      "step": 11363
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.810375610549245e-06,
+      "loss": 0.6406,
+      "step": 11364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.808837383328062e-06,
+      "loss": 0.625,
+      "step": 11365
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.807299210565765e-06,
+      "loss": 0.918,
+      "step": 11366
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.805761092300573e-06,
+      "loss": 0.7031,
+      "step": 11367
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.804223028570721e-06,
+      "loss": 1.5547,
+      "step": 11368
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.802685019414437e-06,
+      "loss": 1.0938,
+      "step": 11369
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.801147064869945e-06,
+      "loss": 0.9766,
+      "step": 11370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.799609164975466e-06,
+      "loss": 0.8789,
+      "step": 11371
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.79807131976923e-06,
+      "loss": 0.8164,
+      "step": 11372
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.796533529289453e-06,
+      "loss": 1.125,
+      "step": 11373
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.794995793574355e-06,
+      "loss": 0.9492,
+      "step": 11374
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.793458112662154e-06,
+      "loss": 1.1172,
+      "step": 11375
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.791920486591072e-06,
+      "loss": 0.416,
+      "step": 11376
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.790382915399323e-06,
+      "loss": 0.4434,
+      "step": 11377
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.788845399125119e-06,
+      "loss": 0.5781,
+      "step": 11378
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.787307937806671e-06,
+      "loss": 0.5586,
+      "step": 11379
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.785770531482199e-06,
+      "loss": 1.4609,
+      "step": 11380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.784233180189911e-06,
+      "loss": 0.165,
+      "step": 11381
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.782695883968014e-06,
+      "loss": 0.9062,
+      "step": 11382
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.781158642854713e-06,
+      "loss": 0.6641,
+      "step": 11383
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.779621456888218e-06,
+      "loss": 0.498,
+      "step": 11384
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.778084326106737e-06,
+      "loss": 0.2207,
+      "step": 11385
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.776547250548467e-06,
+      "loss": 0.6055,
+      "step": 11386
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.775010230251614e-06,
+      "loss": 0.8711,
+      "step": 11387
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.77347326525438e-06,
+      "loss": 1.1562,
+      "step": 11388
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.771936355594964e-06,
+      "loss": 0.9648,
+      "step": 11389
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.770399501311562e-06,
+      "loss": 1.5312,
+      "step": 11390
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.768862702442369e-06,
+      "loss": 1.1562,
+      "step": 11391
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.767325959025586e-06,
+      "loss": 0.2734,
+      "step": 11392
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.765789271099404e-06,
+      "loss": 1.9531,
+      "step": 11393
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.764252638702019e-06,
+      "loss": 1.0625,
+      "step": 11394
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.762716061871616e-06,
+      "loss": 0.8164,
+      "step": 11395
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.761179540646389e-06,
+      "loss": 1.1094,
+      "step": 11396
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.759643075064524e-06,
+      "loss": 0.75,
+      "step": 11397
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.758106665164214e-06,
+      "loss": 1.5625,
+      "step": 11398
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.75657031098364e-06,
+      "loss": 1.1406,
+      "step": 11399
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.755034012560983e-06,
+      "loss": 0.5547,
+      "step": 11400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.753497769934431e-06,
+      "loss": 1.4531,
+      "step": 11401
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.751961583142167e-06,
+      "loss": 1.3281,
+      "step": 11402
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.750425452222367e-06,
+      "loss": 0.5586,
+      "step": 11403
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.74888937721321e-06,
+      "loss": 1.0625,
+      "step": 11404
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.747353358152879e-06,
+      "loss": 1.1016,
+      "step": 11405
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.745817395079544e-06,
+      "loss": 0.5352,
+      "step": 11406
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.744281488031381e-06,
+      "loss": 1.2266,
+      "step": 11407
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.742745637046562e-06,
+      "loss": 0.7695,
+      "step": 11408
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.741209842163263e-06,
+      "loss": 1.0781,
+      "step": 11409
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.739674103419654e-06,
+      "loss": 1.1562,
+      "step": 11410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.7381384208539e-06,
+      "loss": 0.7617,
+      "step": 11411
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.736602794504169e-06,
+      "loss": 1.1562,
+      "step": 11412
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.73506722440863e-06,
+      "loss": 0.418,
+      "step": 11413
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.733531710605448e-06,
+      "loss": 1.6016,
+      "step": 11414
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.731996253132788e-06,
+      "loss": 0.2871,
+      "step": 11415
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.730460852028806e-06,
+      "loss": 1.0547,
+      "step": 11416
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.728925507331666e-06,
+      "loss": 0.582,
+      "step": 11417
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.72739021907953e-06,
+      "loss": 1.0625,
+      "step": 11418
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.72585498731055e-06,
+      "loss": 0.8203,
+      "step": 11419
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.724319812062886e-06,
+      "loss": 1.1172,
+      "step": 11420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.722784693374697e-06,
+      "loss": 0.8945,
+      "step": 11421
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.72124963128413e-06,
+      "loss": 1.8828,
+      "step": 11422
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.71971462582934e-06,
+      "loss": 0.3633,
+      "step": 11423
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.718179677048474e-06,
+      "loss": 1.1328,
+      "step": 11424
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.716644784979688e-06,
+      "loss": 1.1875,
+      "step": 11425
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.71510994966113e-06,
+      "loss": 1.125,
+      "step": 11426
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.71357517113094e-06,
+      "loss": 0.9688,
+      "step": 11427
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.712040449427267e-06,
+      "loss": 0.3457,
+      "step": 11428
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.710505784588254e-06,
+      "loss": 1.1562,
+      "step": 11429
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.708971176652044e-06,
+      "loss": 0.4883,
+      "step": 11430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.70743662565678e-06,
+      "loss": 0.5312,
+      "step": 11431
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.705902131640596e-06,
+      "loss": 0.7188,
+      "step": 11432
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.704367694641639e-06,
+      "loss": 0.4883,
+      "step": 11433
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.702833314698036e-06,
+      "loss": 1.4922,
+      "step": 11434
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.701298991847927e-06,
+      "loss": 0.2119,
+      "step": 11435
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.699764726129445e-06,
+      "loss": 0.7461,
+      "step": 11436
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.698230517580723e-06,
+      "loss": 0.7266,
+      "step": 11437
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.696696366239895e-06,
+      "loss": 1.25,
+      "step": 11438
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.695162272145086e-06,
+      "loss": 0.6367,
+      "step": 11439
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.693628235334423e-06,
+      "loss": 0.5938,
+      "step": 11440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.692094255846036e-06,
+      "loss": 0.6055,
+      "step": 11441
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.690560333718052e-06,
+      "loss": 1.125,
+      "step": 11442
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.689026468988594e-06,
+      "loss": 0.1699,
+      "step": 11443
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.68749266169578e-06,
+      "loss": 0.4668,
+      "step": 11444
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.685958911877733e-06,
+      "loss": 1.2031,
+      "step": 11445
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.684425219572577e-06,
+      "loss": 1.7031,
+      "step": 11446
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.682891584818426e-06,
+      "loss": 0.75,
+      "step": 11447
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.681358007653396e-06,
+      "loss": 0.9609,
+      "step": 11448
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.679824488115607e-06,
+      "loss": 0.832,
+      "step": 11449
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.67829102624317e-06,
+      "loss": 0.3027,
+      "step": 11450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.676757622074196e-06,
+      "loss": 0.4766,
+      "step": 11451
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.675224275646796e-06,
+      "loss": 1.4219,
+      "step": 11452
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.673690986999082e-06,
+      "loss": 0.21,
+      "step": 11453
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.672157756169165e-06,
+      "loss": 0.8906,
+      "step": 11454
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.670624583195145e-06,
+      "loss": 1.1016,
+      "step": 11455
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.669091468115128e-06,
+      "loss": 0.6484,
+      "step": 11456
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.667558410967224e-06,
+      "loss": 0.5312,
+      "step": 11457
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.66602541178953e-06,
+      "loss": 0.416,
+      "step": 11458
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.664492470620151e-06,
+      "loss": 0.3711,
+      "step": 11459
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.66295958749718e-06,
+      "loss": 1.0625,
+      "step": 11460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.661426762458722e-06,
+      "loss": 0.5703,
+      "step": 11461
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.659893995542872e-06,
+      "loss": 1.2266,
+      "step": 11462
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.658361286787722e-06,
+      "loss": 1.125,
+      "step": 11463
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.656828636231364e-06,
+      "loss": 0.8633,
+      "step": 11464
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.655296043911899e-06,
+      "loss": 0.6602,
+      "step": 11465
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.653763509867413e-06,
+      "loss": 0.9766,
+      "step": 11466
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.652231034135994e-06,
+      "loss": 0.8516,
+      "step": 11467
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.65069861675573e-06,
+      "loss": 1.2031,
+      "step": 11468
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64916625776471e-06,
+      "loss": 0.9023,
+      "step": 11469
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.647633957201018e-06,
+      "loss": 1.0234,
+      "step": 11470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64610171510274e-06,
+      "loss": 0.8398,
+      "step": 11471
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64456953150795e-06,
+      "loss": 1.2812,
+      "step": 11472
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.643037406454739e-06,
+      "loss": 2.3281,
+      "step": 11473
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.64150533998118e-06,
+      "loss": 0.6406,
+      "step": 11474
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.639973332125352e-06,
+      "loss": 0.2832,
+      "step": 11475
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.638441382925335e-06,
+      "loss": 0.9727,
+      "step": 11476
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.636909492419195e-06,
+      "loss": 0.1953,
+      "step": 11477
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.635377660645014e-06,
+      "loss": 0.3926,
+      "step": 11478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.63384588764086e-06,
+      "loss": 1.0547,
+      "step": 11479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.632314173444805e-06,
+      "loss": 0.3633,
+      "step": 11480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.630782518094914e-06,
+      "loss": 0.2891,
+      "step": 11481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.629250921629264e-06,
+      "loss": 0.2393,
+      "step": 11482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.6277193840859135e-06,
+      "loss": 0.4512,
+      "step": 11483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.626187905502928e-06,
+      "loss": 0.2891,
+      "step": 11484
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.624656485918369e-06,
+      "loss": 0.6406,
+      "step": 11485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.623125125370303e-06,
+      "loss": 0.8828,
+      "step": 11486
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.62159382389679e-06,
+      "loss": 0.9102,
+      "step": 11487
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.6200625815358845e-06,
+      "loss": 0.5234,
+      "step": 11488
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.618531398325644e-06,
+      "loss": 0.8477,
+      "step": 11489
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.617000274304128e-06,
+      "loss": 0.707,
+      "step": 11490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.615469209509389e-06,
+      "loss": 0.5625,
+      "step": 11491
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.613938203979483e-06,
+      "loss": 1.8281,
+      "step": 11492
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.612407257752455e-06,
+      "loss": 0.5547,
+      "step": 11493
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.610876370866359e-06,
+      "loss": 0.9023,
+      "step": 11494
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.6093455433592435e-06,
+      "loss": 0.7617,
+      "step": 11495
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.607814775269156e-06,
+      "loss": 0.8672,
+      "step": 11496
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.606284066634136e-06,
+      "loss": 0.582,
+      "step": 11497
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.604753417492237e-06,
+      "loss": 1.0625,
+      "step": 11498
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.603222827881498e-06,
+      "loss": 0.8984,
+      "step": 11499
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.601692297839958e-06,
+      "loss": 1.5625,
+      "step": 11500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.6001618274056545e-06,
+      "loss": 1.2109,
+      "step": 11501
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.598631416616632e-06,
+      "loss": 1.2266,
+      "step": 11502
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.597101065510923e-06,
+      "loss": 0.6523,
+      "step": 11503
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.595570774126567e-06,
+      "loss": 0.9258,
+      "step": 11504
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.5940405425015896e-06,
+      "loss": 1.1406,
+      "step": 11505
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.592510370674029e-06,
+      "loss": 0.5273,
+      "step": 11506
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.590980258681915e-06,
+      "loss": 1.0078,
+      "step": 11507
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.589450206563277e-06,
+      "loss": 0.9453,
+      "step": 11508
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.587920214356139e-06,
+      "loss": 1.0156,
+      "step": 11509
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.586390282098536e-06,
+      "loss": 0.3418,
+      "step": 11510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.584860409828484e-06,
+      "loss": 0.6445,
+      "step": 11511
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.583330597584009e-06,
+      "loss": 1.4766,
+      "step": 11512
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.581800845403131e-06,
+      "loss": 0.9453,
+      "step": 11513
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.580271153323875e-06,
+      "loss": 0.7812,
+      "step": 11514
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.57874152138426e-06,
+      "loss": 0.4336,
+      "step": 11515
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.577211949622297e-06,
+      "loss": 0.5586,
+      "step": 11516
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.5756824380760045e-06,
+      "loss": 0.9414,
+      "step": 11517
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.5741529867833985e-06,
+      "loss": 1.0234,
+      "step": 11518
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.572623595782491e-06,
+      "loss": 1.4688,
+      "step": 11519
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.571094265111296e-06,
+      "loss": 0.7891,
+      "step": 11520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5695649948078155e-06,
+      "loss": 0.9023,
+      "step": 11521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.568035784910064e-06,
+      "loss": 1.8984,
+      "step": 11522
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5665066354560475e-06,
+      "loss": 1.0703,
+      "step": 11523
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.564977546483772e-06,
+      "loss": 1.1953,
+      "step": 11524
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.563448518031236e-06,
+      "loss": 0.1729,
+      "step": 11525
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5619195501364515e-06,
+      "loss": 0.3262,
+      "step": 11526
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.560390642837412e-06,
+      "loss": 1.0391,
+      "step": 11527
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.558861796172118e-06,
+      "loss": 1.25,
+      "step": 11528
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5573330101785644e-06,
+      "loss": 0.1953,
+      "step": 11529
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5558042848947535e-06,
+      "loss": 1.2344,
+      "step": 11530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.554275620358678e-06,
+      "loss": 0.6211,
+      "step": 11531
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.552747016608331e-06,
+      "loss": 1.2188,
+      "step": 11532
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.551218473681701e-06,
+      "loss": 0.7031,
+      "step": 11533
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.549689991616781e-06,
+      "loss": 1.1484,
+      "step": 11534
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.548161570451561e-06,
+      "loss": 1.6094,
+      "step": 11535
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.546633210224026e-06,
+      "loss": 0.2617,
+      "step": 11536
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.545104910972161e-06,
+      "loss": 0.8555,
+      "step": 11537
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.543576672733955e-06,
+      "loss": 0.6484,
+      "step": 11538
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.542048495547386e-06,
+      "loss": 1.6875,
+      "step": 11539
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5405203794504356e-06,
+      "loss": 0.5547,
+      "step": 11540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.538992324481082e-06,
+      "loss": 0.8164,
+      "step": 11541
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.537464330677309e-06,
+      "loss": 1.3516,
+      "step": 11542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5359363980770896e-06,
+      "loss": 1.7969,
+      "step": 11543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5344085267183975e-06,
+      "loss": 1.5391,
+      "step": 11544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.532880716639206e-06,
+      "loss": 0.7344,
+      "step": 11545
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.53135296787749e-06,
+      "loss": 0.3359,
+      "step": 11546
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.529825280471219e-06,
+      "loss": 0.9219,
+      "step": 11547
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.528297654458363e-06,
+      "loss": 1.1562,
+      "step": 11548
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.526770089876887e-06,
+      "loss": 1.1562,
+      "step": 11549
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.525242586764756e-06,
+      "loss": 0.8477,
+      "step": 11550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5237151451599374e-06,
+      "loss": 0.5352,
+      "step": 11551
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.522187765100393e-06,
+      "loss": 1.625,
+      "step": 11552
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.520660446624086e-06,
+      "loss": 0.3672,
+      "step": 11553
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.519133189768971e-06,
+      "loss": 0.6094,
+      "step": 11554
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.51760599457301e-06,
+      "loss": 1.0469,
+      "step": 11555
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5160788610741605e-06,
+      "loss": 0.8594,
+      "step": 11556
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.514551789310376e-06,
+      "loss": 0.6367,
+      "step": 11557
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5130247793196085e-06,
+      "loss": 1.0469,
+      "step": 11558
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.511497831139817e-06,
+      "loss": 1.4141,
+      "step": 11559
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.509970944808944e-06,
+      "loss": 1.2344,
+      "step": 11560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.5084441203649415e-06,
+      "loss": 0.7656,
+      "step": 11561
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.506917357845757e-06,
+      "loss": 0.6797,
+      "step": 11562
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.505390657289338e-06,
+      "loss": 1.3906,
+      "step": 11563
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.503864018733626e-06,
+      "loss": 1.3281,
+      "step": 11564
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.50233744221657e-06,
+      "loss": 0.9805,
+      "step": 11565
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.500810927776101e-06,
+      "loss": 0.8633,
+      "step": 11566
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.499284475450168e-06,
+      "loss": 0.7734,
+      "step": 11567
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.497758085276705e-06,
+      "loss": 0.6836,
+      "step": 11568
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.49623175729365e-06,
+      "loss": 0.8516,
+      "step": 11569
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.494705491538935e-06,
+      "loss": 1.0078,
+      "step": 11570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.493179288050502e-06,
+      "loss": 0.7227,
+      "step": 11571
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.491653146866275e-06,
+      "loss": 1.7266,
+      "step": 11572
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.490127068024187e-06,
+      "loss": 0.8633,
+      "step": 11573
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.488601051562163e-06,
+      "loss": 0.377,
+      "step": 11574
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.48707509751814e-06,
+      "loss": 0.5273,
+      "step": 11575
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.485549205930037e-06,
+      "loss": 0.9766,
+      "step": 11576
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.484023376835778e-06,
+      "loss": 0.1943,
+      "step": 11577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.482497610273287e-06,
+      "loss": 0.2676,
+      "step": 11578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.480971906280488e-06,
+      "loss": 0.4277,
+      "step": 11579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.479446264895296e-06,
+      "loss": 1.1641,
+      "step": 11580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.477920686155635e-06,
+      "loss": 1.1484,
+      "step": 11581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.476395170099413e-06,
+      "loss": 0.3711,
+      "step": 11582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.474869716764552e-06,
+      "loss": 0.3438,
+      "step": 11583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.473344326188965e-06,
+      "loss": 0.6484,
+      "step": 11584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.47181899841056e-06,
+      "loss": 0.2373,
+      "step": 11585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.470293733467249e-06,
+      "loss": 0.4648,
+      "step": 11586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.468768531396944e-06,
+      "loss": 0.9961,
+      "step": 11587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.467243392237546e-06,
+      "loss": 1.4141,
+      "step": 11588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.4657183160269665e-06,
+      "loss": 0.9297,
+      "step": 11589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.464193302803104e-06,
+      "loss": 1.3125,
+      "step": 11590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.462668352603867e-06,
+      "loss": 0.5742,
+      "step": 11591
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.461143465467156e-06,
+      "loss": 0.5859,
+      "step": 11592
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.4596186414308635e-06,
+      "loss": 0.8516,
+      "step": 11593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.458093880532891e-06,
+      "loss": 0.5703,
+      "step": 11594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.4565691828111375e-06,
+      "loss": 0.9609,
+      "step": 11595
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.455044548303494e-06,
+      "loss": 0.793,
+      "step": 11596
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.453519977047858e-06,
+      "loss": 1.6172,
+      "step": 11597
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.451995469082113e-06,
+      "loss": 0.8984,
+      "step": 11598
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.4504710244441595e-06,
+      "loss": 1.2188,
+      "step": 11599
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.448946643171877e-06,
+      "loss": 0.6211,
+      "step": 11600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.447422325303158e-06,
+      "loss": 0.6055,
+      "step": 11601
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.445898070875881e-06,
+      "loss": 0.5195,
+      "step": 11602
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.444373879927937e-06,
+      "loss": 0.5547,
+      "step": 11603
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.442849752497208e-06,
+      "loss": 0.6328,
+      "step": 11604
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.441325688621568e-06,
+      "loss": 0.6602,
+      "step": 11605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.439801688338897e-06,
+      "loss": 1.2266,
+      "step": 11606
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.438277751687077e-06,
+      "loss": 1.0625,
+      "step": 11607
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.436753878703982e-06,
+      "loss": 0.7266,
+      "step": 11608
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.435230069427488e-06,
+      "loss": 0.9609,
+      "step": 11609
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.43370632389546e-06,
+      "loss": 1.2109,
+      "step": 11610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.432182642145778e-06,
+      "loss": 1.0391,
+      "step": 11611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.4306590242163046e-06,
+      "loss": 0.4238,
+      "step": 11612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.429135470144913e-06,
+      "loss": 1.2656,
+      "step": 11613
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.427611979969463e-06,
+      "loss": 1.0078,
+      "step": 11614
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.4260885537278285e-06,
+      "loss": 0.7695,
+      "step": 11615
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.424565191457865e-06,
+      "loss": 0.9805,
+      "step": 11616
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.423041893197436e-06,
+      "loss": 1.6641,
+      "step": 11617
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.421518658984399e-06,
+      "loss": 0.8008,
+      "step": 11618
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.419995488856619e-06,
+      "loss": 0.4629,
+      "step": 11619
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.418472382851949e-06,
+      "loss": 0.957,
+      "step": 11620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.41694934100824e-06,
+      "loss": 1.2422,
+      "step": 11621
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.415426363363352e-06,
+      "loss": 0.9219,
+      "step": 11622
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.41390344995513e-06,
+      "loss": 0.9375,
+      "step": 11623
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.412380600821431e-06,
+      "loss": 0.4277,
+      "step": 11624
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.410857816000102e-06,
+      "loss": 0.5664,
+      "step": 11625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.4093350955289885e-06,
+      "loss": 0.6133,
+      "step": 11626
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.4078124394459336e-06,
+      "loss": 0.1592,
+      "step": 11627
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.406289847788786e-06,
+      "loss": 0.9336,
+      "step": 11628
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.404767320595388e-06,
+      "loss": 0.8984,
+      "step": 11629
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.403244857903578e-06,
+      "loss": 0.4473,
+      "step": 11630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.401722459751193e-06,
+      "loss": 0.2041,
+      "step": 11631
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.400200126176075e-06,
+      "loss": 1.2422,
+      "step": 11632
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3986778572160565e-06,
+      "loss": 1.2109,
+      "step": 11633
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.397155652908975e-06,
+      "loss": 0.1152,
+      "step": 11634
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.395633513292659e-06,
+      "loss": 1.3984,
+      "step": 11635
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.394111438404944e-06,
+      "loss": 1.125,
+      "step": 11636
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.392589428283658e-06,
+      "loss": 0.5859,
+      "step": 11637
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.39106748296663e-06,
+      "loss": 0.8945,
+      "step": 11638
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.389545602491679e-06,
+      "loss": 0.208,
+      "step": 11639
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3880237868966385e-06,
+      "loss": 0.7109,
+      "step": 11640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.386502036219327e-06,
+      "loss": 0.5039,
+      "step": 11641
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3849803504975705e-06,
+      "loss": 1.125,
+      "step": 11642
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3834587297691815e-06,
+      "loss": 1.9688,
+      "step": 11643
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3819371740719825e-06,
+      "loss": 0.9961,
+      "step": 11644
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.380415683443791e-06,
+      "loss": 0.4785,
+      "step": 11645
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.37889425792242e-06,
+      "loss": 0.7734,
+      "step": 11646
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.377372897545681e-06,
+      "loss": 0.375,
+      "step": 11647
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.375851602351393e-06,
+      "loss": 0.7031,
+      "step": 11648
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.374330372377359e-06,
+      "loss": 0.4902,
+      "step": 11649
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.372809207661389e-06,
+      "loss": 0.6484,
+      "step": 11650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.371288108241288e-06,
+      "loss": 1.3438,
+      "step": 11651
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3697670741548655e-06,
+      "loss": 0.7344,
+      "step": 11652
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.368246105439927e-06,
+      "loss": 1.4375,
+      "step": 11653
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.366725202134267e-06,
+      "loss": 0.7852,
+      "step": 11654
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3652043642756884e-06,
+      "loss": 0.8867,
+      "step": 11655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.363683591901993e-06,
+      "loss": 0.6914,
+      "step": 11656
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.362162885050975e-06,
+      "loss": 0.3906,
+      "step": 11657
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.360642243760434e-06,
+      "loss": 0.7266,
+      "step": 11658
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.359121668068156e-06,
+      "loss": 0.6836,
+      "step": 11659
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.35760115801194e-06,
+      "loss": 0.7852,
+      "step": 11660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.356080713629575e-06,
+      "loss": 0.4395,
+      "step": 11661
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.354560334958848e-06,
+      "loss": 0.4102,
+      "step": 11662
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.353040022037545e-06,
+      "loss": 0.8711,
+      "step": 11663
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.35151977490346e-06,
+      "loss": 1.2031,
+      "step": 11664
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.34999959359437e-06,
+      "loss": 0.9727,
+      "step": 11665
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3484794781480575e-06,
+      "loss": 0.5273,
+      "step": 11666
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.346959428602303e-06,
+      "loss": 0.4746,
+      "step": 11667
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.345439444994889e-06,
+      "loss": 0.707,
+      "step": 11668
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.34391952736359e-06,
+      "loss": 0.8438,
+      "step": 11669
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.342399675746186e-06,
+      "loss": 0.6445,
+      "step": 11670
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.340879890180443e-06,
+      "loss": 0.7695,
+      "step": 11671
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.339360170704142e-06,
+      "loss": 0.5508,
+      "step": 11672
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.3378405173550506e-06,
+      "loss": 0.4082,
+      "step": 11673
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.336320930170937e-06,
+      "loss": 0.9062,
+      "step": 11674
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.334801409189568e-06,
+      "loss": 1.375,
+      "step": 11675
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.333281954448717e-06,
+      "loss": 1.1094,
+      "step": 11676
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.331762565986139e-06,
+      "loss": 0.3965,
+      "step": 11677
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.330243243839601e-06,
+      "loss": 0.6719,
+      "step": 11678
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.328723988046862e-06,
+      "loss": 0.4785,
+      "step": 11679
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.327204798645684e-06,
+      "loss": 0.6055,
+      "step": 11680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.325685675673828e-06,
+      "loss": 1.1562,
+      "step": 11681
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.324166619169042e-06,
+      "loss": 0.7812,
+      "step": 11682
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.322647629169082e-06,
+      "loss": 0.1953,
+      "step": 11683
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.321128705711706e-06,
+      "loss": 0.5117,
+      "step": 11684
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.3196098488346615e-06,
+      "loss": 0.3711,
+      "step": 11685
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.3180910585757e-06,
+      "loss": 0.3066,
+      "step": 11686
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.316572334972566e-06,
+      "loss": 0.2715,
+      "step": 11687
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.315053678063009e-06,
+      "loss": 0.3477,
+      "step": 11688
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.313535087884771e-06,
+      "loss": 0.7891,
+      "step": 11689
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.312016564475597e-06,
+      "loss": 1.1875,
+      "step": 11690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.310498107873224e-06,
+      "loss": 0.7383,
+      "step": 11691
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.3089797181154e-06,
+      "loss": 0.7734,
+      "step": 11692
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.307461395239854e-06,
+      "loss": 1.1328,
+      "step": 11693
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.3059431392843275e-06,
+      "loss": 0.4219,
+      "step": 11694
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.30442495028655e-06,
+      "loss": 0.9883,
+      "step": 11695
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.302906828284261e-06,
+      "loss": 1.1406,
+      "step": 11696
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.30138877331519e-06,
+      "loss": 1.3047,
+      "step": 11697
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.2998707854170625e-06,
+      "loss": 1.2422,
+      "step": 11698
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.2983528646276095e-06,
+      "loss": 0.3672,
+      "step": 11699
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.296835010984553e-06,
+      "loss": 1.5234,
+      "step": 11700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.295317224525625e-06,
+      "loss": 0.6328,
+      "step": 11701
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.293799505288543e-06,
+      "loss": 1.0156,
+      "step": 11702
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.292281853311033e-06,
+      "loss": 1.0234,
+      "step": 11703
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.290764268630806e-06,
+      "loss": 1.4062,
+      "step": 11704
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.289246751285588e-06,
+      "loss": 0.9805,
+      "step": 11705
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.287729301313091e-06,
+      "loss": 1.2656,
+      "step": 11706
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.286211918751033e-06,
+      "loss": 1.1484,
+      "step": 11707
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.284694603637121e-06,
+      "loss": 0.8633,
+      "step": 11708
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.283177356009075e-06,
+      "loss": 0.8242,
+      "step": 11709
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.281660175904595e-06,
+      "loss": 1.1641,
+      "step": 11710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.280143063361394e-06,
+      "loss": 1.5625,
+      "step": 11711
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.278626018417177e-06,
+      "loss": 0.5273,
+      "step": 11712
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.277109041109649e-06,
+      "loss": 0.6055,
+      "step": 11713
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.275592131476514e-06,
+      "loss": 1.0859,
+      "step": 11714
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.27407528955547e-06,
+      "loss": 1.0547,
+      "step": 11715
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.272558515384216e-06,
+      "loss": 0.6719,
+      "step": 11716
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.271041809000454e-06,
+      "loss": 1.1172,
+      "step": 11717
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.269525170441877e-06,
+      "loss": 1.1562,
+      "step": 11718
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.268008599746181e-06,
+      "loss": 0.8398,
+      "step": 11719
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.266492096951055e-06,
+      "loss": 0.7461,
+      "step": 11720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.264975662094196e-06,
+      "loss": 0.7383,
+      "step": 11721
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.263459295213287e-06,
+      "loss": 1.0547,
+      "step": 11722
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.261942996346021e-06,
+      "loss": 0.8359,
+      "step": 11723
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.260426765530078e-06,
+      "loss": 1.0938,
+      "step": 11724
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.258910602803148e-06,
+      "loss": 1.1016,
+      "step": 11725
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.257394508202912e-06,
+      "loss": 0.8086,
+      "step": 11726
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.255878481767049e-06,
+      "loss": 0.2773,
+      "step": 11727
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.254362523533236e-06,
+      "loss": 1.125,
+      "step": 11728
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.252846633539157e-06,
+      "loss": 0.2295,
+      "step": 11729
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.2513308118224855e-06,
+      "loss": 0.5977,
+      "step": 11730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.249815058420891e-06,
+      "loss": 2.0,
+      "step": 11731
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.24829937337205e-06,
+      "loss": 1.0781,
+      "step": 11732
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.246783756713631e-06,
+      "loss": 0.9531,
+      "step": 11733
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.245268208483307e-06,
+      "loss": 0.2949,
+      "step": 11734
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.24375272871874e-06,
+      "loss": 0.5859,
+      "step": 11735
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.242237317457599e-06,
+      "loss": 1.4453,
+      "step": 11736
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.240721974737548e-06,
+      "loss": 1.1406,
+      "step": 11737
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.239206700596247e-06,
+      "loss": 0.4727,
+      "step": 11738
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.237691495071359e-06,
+      "loss": 1.3594,
+      "step": 11739
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.236176358200537e-06,
+      "loss": 1.5312,
+      "step": 11740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.2346612900214455e-06,
+      "loss": 0.8945,
+      "step": 11741
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.233146290571737e-06,
+      "loss": 0.8672,
+      "step": 11742
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.231631359889064e-06,
+      "loss": 0.8633,
+      "step": 11743
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.2301164980110774e-06,
+      "loss": 0.6914,
+      "step": 11744
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.228601704975429e-06,
+      "loss": 0.4277,
+      "step": 11745
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.227086980819768e-06,
+      "loss": 0.8555,
+      "step": 11746
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.225572325581743e-06,
+      "loss": 1.1797,
+      "step": 11747
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.2240577392989934e-06,
+      "loss": 0.7109,
+      "step": 11748
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.222543222009166e-06,
+      "loss": 0.832,
+      "step": 11749
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.221028773749903e-06,
+      "loss": 1.3359,
+      "step": 11750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.219514394558842e-06,
+      "loss": 1.5625,
+      "step": 11751
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.218000084473622e-06,
+      "loss": 0.793,
+      "step": 11752
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.216485843531883e-06,
+      "loss": 0.8125,
+      "step": 11753
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.214971671771256e-06,
+      "loss": 0.8477,
+      "step": 11754
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.213457569229374e-06,
+      "loss": 0.8633,
+      "step": 11755
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.211943535943866e-06,
+      "loss": 1.6641,
+      "step": 11756
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.2104295719523685e-06,
+      "loss": 0.4258,
+      "step": 11757
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.208915677292507e-06,
+      "loss": 0.5156,
+      "step": 11758
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.207401852001904e-06,
+      "loss": 0.7227,
+      "step": 11759
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.205888096118184e-06,
+      "loss": 0.7812,
+      "step": 11760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.204374409678974e-06,
+      "loss": 0.6602,
+      "step": 11761
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.202860792721893e-06,
+      "loss": 0.6953,
+      "step": 11762
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.2013472452845626e-06,
+      "loss": 0.8789,
+      "step": 11763
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.199833767404594e-06,
+      "loss": 1.0938,
+      "step": 11764
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.198320359119607e-06,
+      "loss": 0.875,
+      "step": 11765
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.196807020467216e-06,
+      "loss": 0.6211,
+      "step": 11766
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.195293751485034e-06,
+      "loss": 1.0781,
+      "step": 11767
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1937805522106675e-06,
+      "loss": 0.6875,
+      "step": 11768
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.192267422681731e-06,
+      "loss": 0.7773,
+      "step": 11769
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.19075436293583e-06,
+      "loss": 1.0234,
+      "step": 11770
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.189241373010567e-06,
+      "loss": 0.9453,
+      "step": 11771
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.187728452943547e-06,
+      "loss": 0.4082,
+      "step": 11772
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.186215602772371e-06,
+      "loss": 1.0156,
+      "step": 11773
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1847028225346416e-06,
+      "loss": 0.5703,
+      "step": 11774
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1831901122679595e-06,
+      "loss": 0.4043,
+      "step": 11775
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.181677472009916e-06,
+      "loss": 0.6914,
+      "step": 11776
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.180164901798106e-06,
+      "loss": 0.6055,
+      "step": 11777
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.178652401670125e-06,
+      "loss": 0.2109,
+      "step": 11778
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.177139971663567e-06,
+      "loss": 0.293,
+      "step": 11779
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.17562761181602e-06,
+      "loss": 0.6406,
+      "step": 11780
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.174115322165067e-06,
+      "loss": 0.6094,
+      "step": 11781
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.172603102748301e-06,
+      "loss": 0.957,
+      "step": 11782
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.171090953603304e-06,
+      "loss": 0.8672,
+      "step": 11783
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.169578874767658e-06,
+      "loss": 0.5938,
+      "step": 11784
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.168066866278942e-06,
+      "loss": 0.7578,
+      "step": 11785
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.166554928174744e-06,
+      "loss": 1.1797,
+      "step": 11786
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.165043060492631e-06,
+      "loss": 1.2188,
+      "step": 11787
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.163531263270186e-06,
+      "loss": 0.9961,
+      "step": 11788
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1620195365449775e-06,
+      "loss": 0.5078,
+      "step": 11789
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.160507880354582e-06,
+      "loss": 1.0,
+      "step": 11790
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1589962947365696e-06,
+      "loss": 1.2344,
+      "step": 11791
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.157484779728507e-06,
+      "loss": 0.9961,
+      "step": 11792
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1559733353679606e-06,
+      "loss": 1.2031,
+      "step": 11793
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.154461961692498e-06,
+      "loss": 1.0547,
+      "step": 11794
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.152950658739684e-06,
+      "loss": 1.4453,
+      "step": 11795
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.15143942654708e-06,
+      "loss": 0.8047,
+      "step": 11796
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.149928265152238e-06,
+      "loss": 1.2734,
+      "step": 11797
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.148417174592726e-06,
+      "loss": 1.1016,
+      "step": 11798
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.146906154906097e-06,
+      "loss": 1.4141,
+      "step": 11799
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.145395206129906e-06,
+      "loss": 1.2578,
+      "step": 11800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.143884328301702e-06,
+      "loss": 0.5703,
+      "step": 11801
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1423735214590465e-06,
+      "loss": 0.5625,
+      "step": 11802
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.140862785639477e-06,
+      "loss": 0.8516,
+      "step": 11803
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.13935212088055e-06,
+      "loss": 0.4355,
+      "step": 11804
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.137841527219802e-06,
+      "loss": 0.7031,
+      "step": 11805
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.136331004694787e-06,
+      "loss": 0.5508,
+      "step": 11806
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.134820553343043e-06,
+      "loss": 0.4902,
+      "step": 11807
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.133310173202112e-06,
+      "loss": 0.4199,
+      "step": 11808
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.131799864309527e-06,
+      "loss": 0.8828,
+      "step": 11809
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.130289626702831e-06,
+      "loss": 1.7031,
+      "step": 11810
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.128779460419559e-06,
+      "loss": 1.3281,
+      "step": 11811
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.127269365497244e-06,
+      "loss": 1.1484,
+      "step": 11812
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.125759341973413e-06,
+      "loss": 1.0234,
+      "step": 11813
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.124249389885605e-06,
+      "loss": 1.0781,
+      "step": 11814
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1227395092713416e-06,
+      "loss": 1.0938,
+      "step": 11815
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.12122970016815e-06,
+      "loss": 0.9648,
+      "step": 11816
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.119719962613553e-06,
+      "loss": 0.5078,
+      "step": 11817
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.118210296645078e-06,
+      "loss": 0.5117,
+      "step": 11818
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.116700702300246e-06,
+      "loss": 1.2734,
+      "step": 11819
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.115191179616573e-06,
+      "loss": 0.8945,
+      "step": 11820
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.113681728631575e-06,
+      "loss": 0.9023,
+      "step": 11821
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.112172349382773e-06,
+      "loss": 0.8906,
+      "step": 11822
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1106630419076776e-06,
+      "loss": 1.0703,
+      "step": 11823
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.109153806243804e-06,
+      "loss": 0.7539,
+      "step": 11824
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.107644642428655e-06,
+      "loss": 0.2598,
+      "step": 11825
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.106135550499748e-06,
+      "loss": 0.25,
+      "step": 11826
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.104626530494585e-06,
+      "loss": 1.0469,
+      "step": 11827
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.103117582450673e-06,
+      "loss": 0.4707,
+      "step": 11828
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.101608706405512e-06,
+      "loss": 0.6016,
+      "step": 11829
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.100099902396609e-06,
+      "loss": 0.5742,
+      "step": 11830
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.098591170461459e-06,
+      "loss": 1.5156,
+      "step": 11831
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.097082510637563e-06,
+      "loss": 0.2949,
+      "step": 11832
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.095573922962411e-06,
+      "loss": 0.9062,
+      "step": 11833
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.094065407473504e-06,
+      "loss": 0.5039,
+      "step": 11834
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.092556964208334e-06,
+      "loss": 1.0312,
+      "step": 11835
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.091048593204388e-06,
+      "loss": 0.377,
+      "step": 11836
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.089540294499154e-06,
+      "loss": 0.6094,
+      "step": 11837
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.088032068130124e-06,
+      "loss": 0.1206,
+      "step": 11838
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.08652391413478e-06,
+      "loss": 1.6719,
+      "step": 11839
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.085015832550608e-06,
+      "loss": 0.668,
+      "step": 11840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.083507823415084e-06,
+      "loss": 0.2559,
+      "step": 11841
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.081999886765698e-06,
+      "loss": 1.4609,
+      "step": 11842
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.080492022639918e-06,
+      "loss": 0.9414,
+      "step": 11843
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.078984231075225e-06,
+      "loss": 0.7031,
+      "step": 11844
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.077476512109093e-06,
+      "loss": 0.2363,
+      "step": 11845
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0759688657789905e-06,
+      "loss": 0.918,
+      "step": 11846
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.074461292122398e-06,
+      "loss": 0.9297,
+      "step": 11847
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.072953791176776e-06,
+      "loss": 0.7031,
+      "step": 11848
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0714463629795945e-06,
+      "loss": 0.9453,
+      "step": 11849
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.069939007568318e-06,
+      "loss": 1.1797,
+      "step": 11850
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.068431724980411e-06,
+      "loss": 1.0703,
+      "step": 11851
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.066924515253339e-06,
+      "loss": 0.9375,
+      "step": 11852
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.065417378424556e-06,
+      "loss": 1.3984,
+      "step": 11853
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.063910314531519e-06,
+      "loss": 0.6367,
+      "step": 11854
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.062403323611691e-06,
+      "loss": 0.7695,
+      "step": 11855
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.060896405702524e-06,
+      "loss": 1.0703,
+      "step": 11856
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.059389560841472e-06,
+      "loss": 0.6758,
+      "step": 11857
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0578827890659796e-06,
+      "loss": 1.2188,
+      "step": 11858
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.056376090413504e-06,
+      "loss": 0.3711,
+      "step": 11859
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0548694649214865e-06,
+      "loss": 0.5938,
+      "step": 11860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.053362912627376e-06,
+      "loss": 0.6484,
+      "step": 11861
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.051856433568613e-06,
+      "loss": 1.0859,
+      "step": 11862
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.050350027782646e-06,
+      "loss": 0.7148,
+      "step": 11863
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.048843695306908e-06,
+      "loss": 0.7383,
+      "step": 11864
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.047337436178842e-06,
+      "loss": 0.6641,
+      "step": 11865
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.045831250435877e-06,
+      "loss": 1.0156,
+      "step": 11866
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0443251381154565e-06,
+      "loss": 1.3203,
+      "step": 11867
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.042819099255012e-06,
+      "loss": 1.2969,
+      "step": 11868
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.041313133891967e-06,
+      "loss": 0.6836,
+      "step": 11869
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.039807242063757e-06,
+      "loss": 0.7227,
+      "step": 11870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0383014238078075e-06,
+      "loss": 0.8477,
+      "step": 11871
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.036795679161543e-06,
+      "loss": 1.1641,
+      "step": 11872
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.03529000816239e-06,
+      "loss": 0.2207,
+      "step": 11873
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.033784410847766e-06,
+      "loss": 0.4258,
+      "step": 11874
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.032278887255097e-06,
+      "loss": 0.9414,
+      "step": 11875
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.030773437421795e-06,
+      "loss": 1.625,
+      "step": 11876
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0292680613852796e-06,
+      "loss": 1.3203,
+      "step": 11877
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.027762759182961e-06,
+      "loss": 1.3203,
+      "step": 11878
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.026257530852258e-06,
+      "loss": 1.6094,
+      "step": 11879
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.024752376430582e-06,
+      "loss": 0.4805,
+      "step": 11880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.023247295955335e-06,
+      "loss": 1.3281,
+      "step": 11881
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.021742289463925e-06,
+      "loss": 1.0234,
+      "step": 11882
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0202373569937625e-06,
+      "loss": 0.3867,
+      "step": 11883
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.018732498582247e-06,
+      "loss": 1.3516,
+      "step": 11884
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.017227714266783e-06,
+      "loss": 1.0312,
+      "step": 11885
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0157230040847664e-06,
+      "loss": 0.9375,
+      "step": 11886
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.014218368073597e-06,
+      "loss": 0.7461,
+      "step": 11887
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.012713806270672e-06,
+      "loss": 0.1157,
+      "step": 11888
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.0112093187133856e-06,
+      "loss": 0.1484,
+      "step": 11889
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.009704905439125e-06,
+      "loss": 0.4062,
+      "step": 11890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.00820056648529e-06,
+      "loss": 0.3242,
+      "step": 11891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.006696301889261e-06,
+      "loss": 0.8242,
+      "step": 11892
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.005192111688428e-06,
+      "loss": 1.1094,
+      "step": 11893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.003687995920174e-06,
+      "loss": 0.957,
+      "step": 11894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.002183954621886e-06,
+      "loss": 1.7891,
+      "step": 11895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.000679987830945e-06,
+      "loss": 1.7656,
+      "step": 11896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.999176095584728e-06,
+      "loss": 0.3477,
+      "step": 11897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.997672277920609e-06,
+      "loss": 1.2734,
+      "step": 11898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.99616853487597e-06,
+      "loss": 0.9062,
+      "step": 11899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.994664866488184e-06,
+      "loss": 1.0234,
+      "step": 11900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.993161272794622e-06,
+      "loss": 1.0781,
+      "step": 11901
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.99165775383265e-06,
+      "loss": 1.0234,
+      "step": 11902
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.990154309639642e-06,
+      "loss": 1.0234,
+      "step": 11903
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.988650940252963e-06,
+      "loss": 0.7461,
+      "step": 11904
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.987147645709977e-06,
+      "loss": 1.0,
+      "step": 11905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.985644426048044e-06,
+      "loss": 1.0938,
+      "step": 11906
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.984141281304529e-06,
+      "loss": 0.9297,
+      "step": 11907
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.982638211516792e-06,
+      "loss": 0.5977,
+      "step": 11908
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.981135216722187e-06,
+      "loss": 0.8125,
+      "step": 11909
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.979632296958066e-06,
+      "loss": 1.0859,
+      "step": 11910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.978129452261788e-06,
+      "loss": 0.8672,
+      "step": 11911
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.976626682670704e-06,
+      "loss": 0.8203,
+      "step": 11912
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.975123988222165e-06,
+      "loss": 0.9023,
+      "step": 11913
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.97362136895351e-06,
+      "loss": 0.8203,
+      "step": 11914
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.9721188249020946e-06,
+      "loss": 1.6641,
+      "step": 11915
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.9706163561052585e-06,
+      "loss": 0.7617,
+      "step": 11916
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.969113962600344e-06,
+      "loss": 0.8672,
+      "step": 11917
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9676116444246925e-06,
+      "loss": 1.0859,
+      "step": 11918
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.966109401615644e-06,
+      "loss": 1.1094,
+      "step": 11919
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.964607234210532e-06,
+      "loss": 0.8555,
+      "step": 11920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.963105142246691e-06,
+      "loss": 0.918,
+      "step": 11921
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.961603125761456e-06,
+      "loss": 0.4805,
+      "step": 11922
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.960101184792154e-06,
+      "loss": 1.2734,
+      "step": 11923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.958599319376123e-06,
+      "loss": 0.8008,
+      "step": 11924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.95709752955068e-06,
+      "loss": 0.8398,
+      "step": 11925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.955595815353156e-06,
+      "loss": 0.9766,
+      "step": 11926
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.95409417682087e-06,
+      "loss": 0.3262,
+      "step": 11927
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.952592613991149e-06,
+      "loss": 0.2637,
+      "step": 11928
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.951091126901313e-06,
+      "loss": 0.4355,
+      "step": 11929
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.949589715588673e-06,
+      "loss": 0.3398,
+      "step": 11930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.948088380090547e-06,
+      "loss": 0.2637,
+      "step": 11931
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.946587120444252e-06,
+      "loss": 1.1719,
+      "step": 11932
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.945085936687102e-06,
+      "loss": 0.2109,
+      "step": 11933
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.943584828856403e-06,
+      "loss": 0.7266,
+      "step": 11934
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9420837969894625e-06,
+      "loss": 0.3281,
+      "step": 11935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.94058284112359e-06,
+      "loss": 0.582,
+      "step": 11936
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.939081961296088e-06,
+      "loss": 0.5898,
+      "step": 11937
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.937581157544262e-06,
+      "loss": 0.9805,
+      "step": 11938
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9360804299054075e-06,
+      "loss": 0.6055,
+      "step": 11939
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.93457977841683e-06,
+      "loss": 0.4824,
+      "step": 11940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.933079203115825e-06,
+      "loss": 1.3281,
+      "step": 11941
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.931578704039683e-06,
+      "loss": 1.25,
+      "step": 11942
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.930078281225701e-06,
+      "loss": 0.668,
+      "step": 11943
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.928577934711168e-06,
+      "loss": 1.6875,
+      "step": 11944
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.927077664533377e-06,
+      "loss": 1.5547,
+      "step": 11945
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.925577470729614e-06,
+      "loss": 0.2871,
+      "step": 11946
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.924077353337162e-06,
+      "loss": 1.0938,
+      "step": 11947
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.922577312393308e-06,
+      "loss": 0.9844,
+      "step": 11948
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9210773479353325e-06,
+      "loss": 1.2578,
+      "step": 11949
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.919577460000515e-06,
+      "loss": 1.25,
+      "step": 11950
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.918077648626133e-06,
+      "loss": 0.5938,
+      "step": 11951
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.916577913849467e-06,
+      "loss": 0.5391,
+      "step": 11952
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.915078255707786e-06,
+      "loss": 1.2734,
+      "step": 11953
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.913578674238365e-06,
+      "loss": 1.7891,
+      "step": 11954
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9120791694784715e-06,
+      "loss": 1.3438,
+      "step": 11955
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.910579741465377e-06,
+      "loss": 0.3164,
+      "step": 11956
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.909080390236351e-06,
+      "loss": 0.9727,
+      "step": 11957
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.907581115828652e-06,
+      "loss": 0.8672,
+      "step": 11958
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9060819182795426e-06,
+      "loss": 1.5859,
+      "step": 11959
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9045827976262895e-06,
+      "loss": 1.1016,
+      "step": 11960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.9030837539061455e-06,
+      "loss": 1.0391,
+      "step": 11961
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.901584787156376e-06,
+      "loss": 1.2734,
+      "step": 11962
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.900085897414225e-06,
+      "loss": 1.3672,
+      "step": 11963
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.898587084716955e-06,
+      "loss": 0.8281,
+      "step": 11964
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.897088349101811e-06,
+      "loss": 0.9141,
+      "step": 11965
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.895589690606047e-06,
+      "loss": 0.9258,
+      "step": 11966
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.894091109266907e-06,
+      "loss": 0.9961,
+      "step": 11967
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.892592605121642e-06,
+      "loss": 0.9648,
+      "step": 11968
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.891094178207489e-06,
+      "loss": 0.8477,
+      "step": 11969
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.889595828561693e-06,
+      "loss": 0.625,
+      "step": 11970
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.888097556221492e-06,
+      "loss": 0.4902,
+      "step": 11971
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.886599361224124e-06,
+      "loss": 1.3047,
+      "step": 11972
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.885101243606832e-06,
+      "loss": 1.0391,
+      "step": 11973
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.883603203406841e-06,
+      "loss": 1.8438,
+      "step": 11974
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.882105240661383e-06,
+      "loss": 0.3535,
+      "step": 11975
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.880607355407696e-06,
+      "loss": 0.2266,
+      "step": 11976
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.879109547683001e-06,
+      "loss": 1.4219,
+      "step": 11977
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.877611817524528e-06,
+      "loss": 1.4297,
+      "step": 11978
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.876114164969498e-06,
+      "loss": 0.4102,
+      "step": 11979
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.874616590055141e-06,
+      "loss": 0.1924,
+      "step": 11980
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.873119092818671e-06,
+      "loss": 0.9414,
+      "step": 11981
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.871621673297307e-06,
+      "loss": 0.1914,
+      "step": 11982
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.870124331528265e-06,
+      "loss": 1.8203,
+      "step": 11983
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.868627067548763e-06,
+      "loss": 1.0781,
+      "step": 11984
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.867129881396014e-06,
+      "loss": 0.4023,
+      "step": 11985
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.865632773107226e-06,
+      "loss": 1.3281,
+      "step": 11986
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.8641357427196065e-06,
+      "loss": 0.3359,
+      "step": 11987
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.8626387902703675e-06,
+      "loss": 1.125,
+      "step": 11988
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.8611419157967115e-06,
+      "loss": 1.3281,
+      "step": 11989
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.8596451193358446e-06,
+      "loss": 0.2275,
+      "step": 11990
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.85814840092496e-06,
+      "loss": 0.2158,
+      "step": 11991
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.856651760601265e-06,
+      "loss": 1.2891,
+      "step": 11992
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.855155198401953e-06,
+      "loss": 0.9609,
+      "step": 11993
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.853658714364222e-06,
+      "loss": 0.8164,
+      "step": 11994
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.852162308525264e-06,
+      "loss": 1.5859,
+      "step": 11995
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.850665980922267e-06,
+      "loss": 2.125,
+      "step": 11996
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.849169731592427e-06,
+      "loss": 0.9414,
+      "step": 11997
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.8476735605729275e-06,
+      "loss": 1.0234,
+      "step": 11998
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.846177467900956e-06,
+      "loss": 0.8984,
+      "step": 11999
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.844681453613692e-06,
+      "loss": 0.5117,
+      "step": 12000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.8431855177483255e-06,
+      "loss": 0.6523,
+      "step": 12001
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.841689660342029e-06,
+      "loss": 1.0703,
+      "step": 12002
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.840193881431982e-06,
+      "loss": 1.7656,
+      "step": 12003
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.838698181055361e-06,
+      "loss": 0.8281,
+      "step": 12004
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.837202559249341e-06,
+      "loss": 0.9961,
+      "step": 12005
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.835707016051095e-06,
+      "loss": 0.3301,
+      "step": 12006
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.834211551497789e-06,
+      "loss": 0.8438,
+      "step": 12007
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.832716165626592e-06,
+      "loss": 0.5977,
+      "step": 12008
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.831220858474672e-06,
+      "loss": 0.582,
+      "step": 12009
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.829725630079193e-06,
+      "loss": 1.4688,
+      "step": 12010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.828230480477318e-06,
+      "loss": 1.1094,
+      "step": 12011
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.826735409706203e-06,
+      "loss": 0.4512,
+      "step": 12012
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.825240417803014e-06,
+      "loss": 0.3652,
+      "step": 12013
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.823745504804899e-06,
+      "loss": 0.4043,
+      "step": 12014
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.8222506707490176e-06,
+      "loss": 1.1797,
+      "step": 12015
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.820755915672519e-06,
+      "loss": 0.5781,
+      "step": 12016
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.819261239612555e-06,
+      "loss": 0.5977,
+      "step": 12017
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.817766642606278e-06,
+      "loss": 0.7266,
+      "step": 12018
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.81627212469083e-06,
+      "loss": 1.0078,
+      "step": 12019
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.814777685903353e-06,
+      "loss": 0.8789,
+      "step": 12020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.813283326280995e-06,
+      "loss": 0.6641,
+      "step": 12021
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.811789045860895e-06,
+      "loss": 1.1562,
+      "step": 12022
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.810294844680194e-06,
+      "loss": 2.3281,
+      "step": 12023
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.808800722776021e-06,
+      "loss": 1.6875,
+      "step": 12024
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.807306680185519e-06,
+      "loss": 0.4121,
+      "step": 12025
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.8058127169458166e-06,
+      "loss": 1.0391,
+      "step": 12026
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.8043188330940455e-06,
+      "loss": 0.1099,
+      "step": 12027
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.8028250286673324e-06,
+      "loss": 0.9062,
+      "step": 12028
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.8013313037028105e-06,
+      "loss": 1.1641,
+      "step": 12029
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.799837658237598e-06,
+      "loss": 0.793,
+      "step": 12030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.7983440923088215e-06,
+      "loss": 0.5938,
+      "step": 12031
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.796850605953598e-06,
+      "loss": 0.1152,
+      "step": 12032
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.795357199209049e-06,
+      "loss": 0.9531,
+      "step": 12033
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.793863872112296e-06,
+      "loss": 0.4961,
+      "step": 12034
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.792370624700446e-06,
+      "loss": 1.0625,
+      "step": 12035
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.790877457010614e-06,
+      "loss": 0.9297,
+      "step": 12036
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.789384369079914e-06,
+      "loss": 0.3398,
+      "step": 12037
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.787891360945453e-06,
+      "loss": 0.4277,
+      "step": 12038
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.786398432644342e-06,
+      "loss": 0.9023,
+      "step": 12039
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.784905584213677e-06,
+      "loss": 0.9922,
+      "step": 12040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.78341281569057e-06,
+      "loss": 1.6484,
+      "step": 12041
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.781920127112117e-06,
+      "loss": 1.4609,
+      "step": 12042
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.78042751851542e-06,
+      "loss": 1.2734,
+      "step": 12043
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.7789349899375725e-06,
+      "loss": 0.5898,
+      "step": 12044
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.777442541415673e-06,
+      "loss": 0.4805,
+      "step": 12045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.775950172986817e-06,
+      "loss": 0.4902,
+      "step": 12046
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.774457884688089e-06,
+      "loss": 1.1797,
+      "step": 12047
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.77296567655658e-06,
+      "loss": 0.9102,
+      "step": 12048
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.7714735486293815e-06,
+      "loss": 0.4922,
+      "step": 12049
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.769981500943574e-06,
+      "loss": 0.9805,
+      "step": 12050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.768489533536245e-06,
+      "loss": 1.0469,
+      "step": 12051
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.766997646444469e-06,
+      "loss": 1.6016,
+      "step": 12052
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.765505839705331e-06,
+      "loss": 0.5859,
+      "step": 12053
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.764014113355908e-06,
+      "loss": 1.6797,
+      "step": 12054
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.762522467433272e-06,
+      "loss": 0.7656,
+      "step": 12055
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.761030901974495e-06,
+      "loss": 0.8281,
+      "step": 12056
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.7595394170166565e-06,
+      "loss": 0.4766,
+      "step": 12057
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.758048012596817e-06,
+      "loss": 0.6875,
+      "step": 12058
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.756556688752047e-06,
+      "loss": 0.7695,
+      "step": 12059
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.755065445519411e-06,
+      "loss": 1.3828,
+      "step": 12060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.753574282935973e-06,
+      "loss": 1.5625,
+      "step": 12061
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.752083201038797e-06,
+      "loss": 0.3945,
+      "step": 12062
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.750592199864936e-06,
+      "loss": 1.2109,
+      "step": 12063
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.749101279451448e-06,
+      "loss": 0.8945,
+      "step": 12064
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.747610439835393e-06,
+      "loss": 0.5391,
+      "step": 12065
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.74611968105382e-06,
+      "loss": 0.7773,
+      "step": 12066
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.744629003143786e-06,
+      "loss": 1.0234,
+      "step": 12067
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.743138406142329e-06,
+      "loss": 1.1328,
+      "step": 12068
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.7416478900865065e-06,
+      "loss": 0.6211,
+      "step": 12069
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.740157455013358e-06,
+      "loss": 0.8359,
+      "step": 12070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.738667100959931e-06,
+      "loss": 0.8047,
+      "step": 12071
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.737176827963266e-06,
+      "loss": 0.5195,
+      "step": 12072
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.735686636060394e-06,
+      "loss": 0.4746,
+      "step": 12073
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.734196525288361e-06,
+      "loss": 0.8477,
+      "step": 12074
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.732706495684199e-06,
+      "loss": 1.6406,
+      "step": 12075
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.731216547284942e-06,
+      "loss": 1.5547,
+      "step": 12076
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.729726680127618e-06,
+      "loss": 0.4805,
+      "step": 12077
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.728236894249259e-06,
+      "loss": 1.5,
+      "step": 12078
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.726747189686894e-06,
+      "loss": 0.7812,
+      "step": 12079
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.725257566477543e-06,
+      "loss": 0.1523,
+      "step": 12080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.723768024658229e-06,
+      "loss": 0.75,
+      "step": 12081
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.722278564265976e-06,
+      "loss": 0.8828,
+      "step": 12082
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.720789185337804e-06,
+      "loss": 1.0859,
+      "step": 12083
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.719299887910728e-06,
+      "loss": 1.1875,
+      "step": 12084
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.717810672021757e-06,
+      "loss": 0.4277,
+      "step": 12085
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.716321537707913e-06,
+      "loss": 0.332,
+      "step": 12086
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.714832485006202e-06,
+      "loss": 0.5547,
+      "step": 12087
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.713343513953634e-06,
+      "loss": 0.4609,
+      "step": 12088
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.711854624587213e-06,
+      "loss": 0.7656,
+      "step": 12089
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.71036581694395e-06,
+      "loss": 0.4844,
+      "step": 12090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.708877091060842e-06,
+      "loss": 0.7773,
+      "step": 12091
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.707388446974891e-06,
+      "loss": 0.1157,
+      "step": 12092
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.705899884723093e-06,
+      "loss": 1.1641,
+      "step": 12093
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.70441140434245e-06,
+      "loss": 0.4219,
+      "step": 12094
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.702923005869957e-06,
+      "loss": 0.3672,
+      "step": 12095
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.701434689342599e-06,
+      "loss": 0.4863,
+      "step": 12096
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.69994645479737e-06,
+      "loss": 0.9453,
+      "step": 12097
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.698458302271262e-06,
+      "loss": 1.2422,
+      "step": 12098
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.696970231801257e-06,
+      "loss": 0.418,
+      "step": 12099
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.695482243424344e-06,
+      "loss": 1.3125,
+      "step": 12100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.693994337177497e-06,
+      "loss": 0.4336,
+      "step": 12101
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.692506513097704e-06,
+      "loss": 0.4922,
+      "step": 12102
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.69101877122194e-06,
+      "loss": 0.6836,
+      "step": 12103
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.689531111587182e-06,
+      "loss": 1.0781,
+      "step": 12104
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.688043534230401e-06,
+      "loss": 0.4863,
+      "step": 12105
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.686556039188578e-06,
+      "loss": 0.3828,
+      "step": 12106
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.685068626498674e-06,
+      "loss": 1.0234,
+      "step": 12107
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.683581296197659e-06,
+      "loss": 1.5234,
+      "step": 12108
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.682094048322498e-06,
+      "loss": 1.1719,
+      "step": 12109
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.680606882910159e-06,
+      "loss": 0.5938,
+      "step": 12110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.6791197999975995e-06,
+      "loss": 1.3125,
+      "step": 12111
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.677632799621786e-06,
+      "loss": 0.6797,
+      "step": 12112
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.676145881819666e-06,
+      "loss": 0.5078,
+      "step": 12113
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.6746590466282025e-06,
+      "loss": 0.4531,
+      "step": 12114
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.673172294084349e-06,
+      "loss": 0.5898,
+      "step": 12115
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.671685624225053e-06,
+      "loss": 0.9883,
+      "step": 12116
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.670199037087266e-06,
+      "loss": 0.8008,
+      "step": 12117
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.668712532707939e-06,
+      "loss": 1.1484,
+      "step": 12118
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.667226111124011e-06,
+      "loss": 0.9102,
+      "step": 12119
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.6657397723724305e-06,
+      "loss": 0.4219,
+      "step": 12120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.664253516490133e-06,
+      "loss": 0.7773,
+      "step": 12121
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.662767343514065e-06,
+      "loss": 0.7539,
+      "step": 12122
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.66128125348116e-06,
+      "loss": 1.125,
+      "step": 12123
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.659795246428353e-06,
+      "loss": 0.8711,
+      "step": 12124
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.658309322392575e-06,
+      "loss": 0.7344,
+      "step": 12125
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.656823481410761e-06,
+      "loss": 1.5938,
+      "step": 12126
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.655337723519838e-06,
+      "loss": 1.5625,
+      "step": 12127
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.653852048756735e-06,
+      "loss": 0.7539,
+      "step": 12128
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.652366457158371e-06,
+      "loss": 0.8945,
+      "step": 12129
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.650880948761674e-06,
+      "loss": 0.8516,
+      "step": 12130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.649395523603564e-06,
+      "loss": 1.4922,
+      "step": 12131
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.647910181720958e-06,
+      "loss": 0.3633,
+      "step": 12132
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.646424923150771e-06,
+      "loss": 1.5078,
+      "step": 12133
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.6449397479299236e-06,
+      "loss": 0.2412,
+      "step": 12134
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.643454656095322e-06,
+      "loss": 0.7422,
+      "step": 12135
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.64196964768388e-06,
+      "loss": 0.4824,
+      "step": 12136
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.640484722732501e-06,
+      "loss": 1.3516,
+      "step": 12137
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.638999881278097e-06,
+      "loss": 0.457,
+      "step": 12138
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.637515123357572e-06,
+      "loss": 0.2412,
+      "step": 12139
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.636030449007824e-06,
+      "loss": 0.4824,
+      "step": 12140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.6345458582657505e-06,
+      "loss": 1.2031,
+      "step": 12141
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.633061351168257e-06,
+      "loss": 1.1406,
+      "step": 12142
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.631576927752236e-06,
+      "loss": 0.8125,
+      "step": 12143
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.630092588054582e-06,
+      "loss": 1.0781,
+      "step": 12144
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.628608332112184e-06,
+      "loss": 1.3281,
+      "step": 12145
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.627124159961931e-06,
+      "loss": 0.8164,
+      "step": 12146
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.6256400716407134e-06,
+      "loss": 0.6797,
+      "step": 12147
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.624156067185416e-06,
+      "loss": 0.7734,
+      "step": 12148
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.622672146632922e-06,
+      "loss": 1.1875,
+      "step": 12149
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.62118831002011e-06,
+      "loss": 0.4043,
+      "step": 12150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.619704557383865e-06,
+      "loss": 0.9609,
+      "step": 12151
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.618220888761057e-06,
+      "loss": 1.4297,
+      "step": 12152
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.616737304188566e-06,
+      "loss": 0.6992,
+      "step": 12153
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.615253803703261e-06,
+      "loss": 0.4785,
+      "step": 12154
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.613770387342015e-06,
+      "loss": 1.0156,
+      "step": 12155
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.6122870551417e-06,
+      "loss": 0.7852,
+      "step": 12156
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.610803807139177e-06,
+      "loss": 1.0078,
+      "step": 12157
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.609320643371309e-06,
+      "loss": 0.6016,
+      "step": 12158
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.6078375638749635e-06,
+      "loss": 0.7422,
+      "step": 12159
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.606354568687e-06,
+      "loss": 1.4922,
+      "step": 12160
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.6048716578442785e-06,
+      "loss": 0.9414,
+      "step": 12161
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.603388831383646e-06,
+      "loss": 1.6797,
+      "step": 12162
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.6019060893419665e-06,
+      "loss": 0.4902,
+      "step": 12163
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.6004234317560865e-06,
+      "loss": 0.3594,
+      "step": 12164
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.5989408586628586e-06,
+      "loss": 0.5352,
+      "step": 12165
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.597458370099128e-06,
+      "loss": 0.6328,
+      "step": 12166
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.595975966101744e-06,
+      "loss": 1.125,
+      "step": 12167
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.594493646707547e-06,
+      "loss": 0.7773,
+      "step": 12168
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.593011411953378e-06,
+      "loss": 0.8398,
+      "step": 12169
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.5915292618760765e-06,
+      "loss": 0.6016,
+      "step": 12170
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.590047196512481e-06,
+      "loss": 0.707,
+      "step": 12171
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.588565215899429e-06,
+      "loss": 1.8438,
+      "step": 12172
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.587083320073748e-06,
+      "loss": 0.8828,
+      "step": 12173
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.58560150907227e-06,
+      "loss": 1.5703,
+      "step": 12174
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.584119782931828e-06,
+      "loss": 0.6836,
+      "step": 12175
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.582638141689244e-06,
+      "loss": 0.4375,
+      "step": 12176
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.581156585381347e-06,
+      "loss": 0.3496,
+      "step": 12177
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.579675114044952e-06,
+      "loss": 0.4219,
+      "step": 12178
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.578193727716887e-06,
+      "loss": 0.4785,
+      "step": 12179
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.576712426433966e-06,
+      "loss": 0.3242,
+      "step": 12180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.575231210233008e-06,
+      "loss": 0.9414,
+      "step": 12181
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.573750079150822e-06,
+      "loss": 0.4395,
+      "step": 12182
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.572269033224225e-06,
+      "loss": 1.0625,
+      "step": 12183
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.570788072490029e-06,
+      "loss": 0.6133,
+      "step": 12184
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.5693071969850335e-06,
+      "loss": 0.7852,
+      "step": 12185
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.567826406746045e-06,
+      "loss": 0.6758,
+      "step": 12186
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.566345701809873e-06,
+      "loss": 1.0312,
+      "step": 12187
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.564865082213316e-06,
+      "loss": 1.2266,
+      "step": 12188
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.563384547993174e-06,
+      "loss": 0.1484,
+      "step": 12189
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.561904099186238e-06,
+      "loss": 0.707,
+      "step": 12190
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.560423735829311e-06,
+      "loss": 1.2031,
+      "step": 12191
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.558943457959181e-06,
+      "loss": 1.2812,
+      "step": 12192
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.557463265612641e-06,
+      "loss": 0.9375,
+      "step": 12193
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.555983158826475e-06,
+      "loss": 1.0547,
+      "step": 12194
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.554503137637479e-06,
+      "loss": 0.8516,
+      "step": 12195
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.553023202082428e-06,
+      "loss": 1.5781,
+      "step": 12196
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.551543352198108e-06,
+      "loss": 1.4219,
+      "step": 12197
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.550063588021296e-06,
+      "loss": 0.7109,
+      "step": 12198
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.548583909588773e-06,
+      "loss": 0.9375,
+      "step": 12199
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.547104316937317e-06,
+      "loss": 0.6055,
+      "step": 12200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.545624810103697e-06,
+      "loss": 1.0469,
+      "step": 12201
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.544145389124683e-06,
+      "loss": 0.5625,
+      "step": 12202
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.54266605403705e-06,
+      "loss": 0.3105,
+      "step": 12203
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.541186804877562e-06,
+      "loss": 1.3359,
+      "step": 12204
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.5397076416829876e-06,
+      "loss": 1.1641,
+      "step": 12205
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.538228564490081e-06,
+      "loss": 0.5391,
+      "step": 12206
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.536749573335612e-06,
+      "loss": 1.2656,
+      "step": 12207
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.535270668256337e-06,
+      "loss": 1.0391,
+      "step": 12208
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.53379184928901e-06,
+      "loss": 1.0547,
+      "step": 12209
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.532313116470386e-06,
+      "loss": 0.6719,
+      "step": 12210
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.530834469837221e-06,
+      "loss": 1.6094,
+      "step": 12211
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.529355909426261e-06,
+      "loss": 0.6367,
+      "step": 12212
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.527877435274255e-06,
+      "loss": 1.2109,
+      "step": 12213
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.526399047417946e-06,
+      "loss": 0.7383,
+      "step": 12214
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.524920745894083e-06,
+      "loss": 0.7773,
+      "step": 12215
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.523442530739408e-06,
+      "loss": 0.9883,
+      "step": 12216
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.521964401990657e-06,
+      "loss": 0.8711,
+      "step": 12217
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.520486359684566e-06,
+      "loss": 0.9883,
+      "step": 12218
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.519008403857871e-06,
+      "loss": 0.5781,
+      "step": 12219
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.517530534547306e-06,
+      "loss": 0.668,
+      "step": 12220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.516052751789604e-06,
+      "loss": 0.9258,
+      "step": 12221
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.51457505562149e-06,
+      "loss": 1.5234,
+      "step": 12222
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.51309744607969e-06,
+      "loss": 0.3418,
+      "step": 12223
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.511619923200932e-06,
+      "loss": 0.9258,
+      "step": 12224
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.510142487021937e-06,
+      "loss": 1.1797,
+      "step": 12225
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.5086651375794215e-06,
+      "loss": 0.3848,
+      "step": 12226
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.5071878749101055e-06,
+      "loss": 0.8945,
+      "step": 12227
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.505710699050711e-06,
+      "loss": 1.1094,
+      "step": 12228
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.504233610037942e-06,
+      "loss": 0.9844,
+      "step": 12229
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.502756607908514e-06,
+      "loss": 0.6914,
+      "step": 12230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.501279692699134e-06,
+      "loss": 0.4238,
+      "step": 12231
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.499802864446513e-06,
+      "loss": 0.9102,
+      "step": 12232
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.498326123187355e-06,
+      "loss": 1.0078,
+      "step": 12233
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.4968494689583615e-06,
+      "loss": 0.6367,
+      "step": 12234
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.49537290179623e-06,
+      "loss": 0.3633,
+      "step": 12235
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.493896421737664e-06,
+      "loss": 1.0859,
+      "step": 12236
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.492420028819359e-06,
+      "loss": 0.4102,
+      "step": 12237
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4909437230780095e-06,
+      "loss": 0.8047,
+      "step": 12238
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4894675045503005e-06,
+      "loss": 1.5469,
+      "step": 12239
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4879913732729304e-06,
+      "loss": 1.1328,
+      "step": 12240
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.486515329282583e-06,
+      "loss": 1.6094,
+      "step": 12241
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.485039372615945e-06,
+      "loss": 0.8672,
+      "step": 12242
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.483563503309698e-06,
+      "loss": 0.1445,
+      "step": 12243
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4820877214005254e-06,
+      "loss": 1.3594,
+      "step": 12244
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.480612026925104e-06,
+      "loss": 0.8906,
+      "step": 12245
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.479136419920111e-06,
+      "loss": 0.8008,
+      "step": 12246
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.477660900422219e-06,
+      "loss": 1.1562,
+      "step": 12247
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4761854684681044e-06,
+      "loss": 0.6953,
+      "step": 12248
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4747101240944365e-06,
+      "loss": 0.7695,
+      "step": 12249
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.473234867337883e-06,
+      "loss": 0.7617,
+      "step": 12250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.471759698235105e-06,
+      "loss": 1.7734,
+      "step": 12251
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.470284616822773e-06,
+      "loss": 0.9492,
+      "step": 12252
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.468809623137546e-06,
+      "loss": 0.9336,
+      "step": 12253
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.467334717216083e-06,
+      "loss": 0.6914,
+      "step": 12254
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4658598990950385e-06,
+      "loss": 0.5508,
+      "step": 12255
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.464385168811075e-06,
+      "loss": 0.957,
+      "step": 12256
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.462910526400837e-06,
+      "loss": 0.7852,
+      "step": 12257
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.46143597190098e-06,
+      "loss": 1.4062,
+      "step": 12258
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.459961505348148e-06,
+      "loss": 0.7578,
+      "step": 12259
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.458487126778992e-06,
+      "loss": 0.6445,
+      "step": 12260
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.457012836230156e-06,
+      "loss": 1.7109,
+      "step": 12261
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.455538633738278e-06,
+      "loss": 0.4434,
+      "step": 12262
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4540645193399985e-06,
+      "loss": 0.9805,
+      "step": 12263
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.452590493071957e-06,
+      "loss": 0.75,
+      "step": 12264
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.451116554970789e-06,
+      "loss": 0.5352,
+      "step": 12265
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.449642705073127e-06,
+      "loss": 0.957,
+      "step": 12266
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4481689434155985e-06,
+      "loss": 0.9453,
+      "step": 12267
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.446695270034835e-06,
+      "loss": 0.9531,
+      "step": 12268
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.445221684967464e-06,
+      "loss": 0.7188,
+      "step": 12269
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.44374818825011e-06,
+      "loss": 0.7891,
+      "step": 12270
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4422747799193884e-06,
+      "loss": 0.8828,
+      "step": 12271
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.440801460011931e-06,
+      "loss": 0.9883,
+      "step": 12272
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.439328228564347e-06,
+      "loss": 1.0938,
+      "step": 12273
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.437855085613254e-06,
+      "loss": 1.3516,
+      "step": 12274
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.436382031195263e-06,
+      "loss": 0.2734,
+      "step": 12275
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.434909065346989e-06,
+      "loss": 0.9102,
+      "step": 12276
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.433436188105042e-06,
+      "loss": 0.4316,
+      "step": 12277
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.431963399506023e-06,
+      "loss": 0.4219,
+      "step": 12278
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.430490699586538e-06,
+      "loss": 0.5156,
+      "step": 12279
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.429018088383193e-06,
+      "loss": 0.543,
+      "step": 12280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.427545565932586e-06,
+      "loss": 1.2109,
+      "step": 12281
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4260731322713145e-06,
+      "loss": 0.5977,
+      "step": 12282
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.424600787435972e-06,
+      "loss": 0.7539,
+      "step": 12283
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.423128531463158e-06,
+      "loss": 0.8398,
+      "step": 12284
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.421656364389459e-06,
+      "loss": 0.918,
+      "step": 12285
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.420184286251466e-06,
+      "loss": 0.7109,
+      "step": 12286
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.418712297085763e-06,
+      "loss": 0.334,
+      "step": 12287
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.417240396928937e-06,
+      "loss": 0.9141,
+      "step": 12288
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.415768585817574e-06,
+      "loss": 1.7578,
+      "step": 12289
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.414296863788247e-06,
+      "loss": 0.8047,
+      "step": 12290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.412825230877536e-06,
+      "loss": 0.4414,
+      "step": 12291
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.41135368712202e-06,
+      "loss": 1.25,
+      "step": 12292
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.409882232558271e-06,
+      "loss": 1.0156,
+      "step": 12293
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.408410867222862e-06,
+      "loss": 1.3047,
+      "step": 12294
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.406939591152358e-06,
+      "loss": 1.4062,
+      "step": 12295
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.405468404383326e-06,
+      "loss": 0.7266,
+      "step": 12296
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.403997306952335e-06,
+      "loss": 0.6211,
+      "step": 12297
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.4025262988959444e-06,
+      "loss": 1.2266,
+      "step": 12298
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.401055380250718e-06,
+      "loss": 0.4531,
+      "step": 12299
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.399584551053207e-06,
+      "loss": 0.2129,
+      "step": 12300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.3981138113399745e-06,
+      "loss": 0.7461,
+      "step": 12301
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.3966431611475684e-06,
+      "loss": 0.4492,
+      "step": 12302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.395172600512545e-06,
+      "loss": 0.3848,
+      "step": 12303
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.393702129471448e-06,
+      "loss": 0.3887,
+      "step": 12304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.392231748060831e-06,
+      "loss": 0.4375,
+      "step": 12305
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.390761456317234e-06,
+      "loss": 1.2969,
+      "step": 12306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.389291254277199e-06,
+      "loss": 0.3457,
+      "step": 12307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.387821141977266e-06,
+      "loss": 0.7695,
+      "step": 12308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.386351119453977e-06,
+      "loss": 0.6484,
+      "step": 12309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.384881186743868e-06,
+      "loss": 1.3594,
+      "step": 12310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.383411343883468e-06,
+      "loss": 0.7305,
+      "step": 12311
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.381941590909308e-06,
+      "loss": 1.0078,
+      "step": 12312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.380471927857921e-06,
+      "loss": 0.6602,
+      "step": 12313
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.3790023547658305e-06,
+      "loss": 1.1094,
+      "step": 12314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.3775328716695675e-06,
+      "loss": 0.707,
+      "step": 12315
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.376063478605641e-06,
+      "loss": 0.707,
+      "step": 12316
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.374594175610587e-06,
+      "loss": 0.8125,
+      "step": 12317
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.373124962720914e-06,
+      "loss": 1.125,
+      "step": 12318
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.371655839973139e-06,
+      "loss": 0.9844,
+      "step": 12319
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.370186807403774e-06,
+      "loss": 0.8164,
+      "step": 12320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.368717865049334e-06,
+      "loss": 0.5977,
+      "step": 12321
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.367249012946329e-06,
+      "loss": 0.9453,
+      "step": 12322
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.365780251131262e-06,
+      "loss": 0.8164,
+      "step": 12323
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.364311579640634e-06,
+      "loss": 0.3711,
+      "step": 12324
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.362842998510954e-06,
+      "loss": 0.8281,
+      "step": 12325
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3613745077787195e-06,
+      "loss": 0.1953,
+      "step": 12326
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.359906107480429e-06,
+      "loss": 0.2598,
+      "step": 12327
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.358437797652573e-06,
+      "loss": 1.1016,
+      "step": 12328
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.356969578331651e-06,
+      "loss": 1.3203,
+      "step": 12329
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.355501449554152e-06,
+      "loss": 0.5312,
+      "step": 12330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.354033411356563e-06,
+      "loss": 0.3379,
+      "step": 12331
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.352565463775371e-06,
+      "loss": 0.3809,
+      "step": 12332
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.351097606847065e-06,
+      "loss": 1.8203,
+      "step": 12333
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3496298406081205e-06,
+      "loss": 1.3672,
+      "step": 12334
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.34816216509502e-06,
+      "loss": 0.6797,
+      "step": 12335
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.346694580344237e-06,
+      "loss": 0.9688,
+      "step": 12336
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.345227086392252e-06,
+      "loss": 0.3105,
+      "step": 12337
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.343759683275539e-06,
+      "loss": 0.5469,
+      "step": 12338
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3422923710305624e-06,
+      "loss": 1.2422,
+      "step": 12339
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.340825149693794e-06,
+      "loss": 1.3594,
+      "step": 12340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3393580193017005e-06,
+      "loss": 0.1982,
+      "step": 12341
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3378909798907436e-06,
+      "loss": 1.0781,
+      "step": 12342
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.33642403149739e-06,
+      "loss": 1.2422,
+      "step": 12343
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.334957174158092e-06,
+      "loss": 0.5352,
+      "step": 12344
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.333490407909311e-06,
+      "loss": 0.9805,
+      "step": 12345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3320237327875e-06,
+      "loss": 1.1406,
+      "step": 12346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.330557148829114e-06,
+      "loss": 0.4492,
+      "step": 12347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.329090656070599e-06,
+      "loss": 1.4453,
+      "step": 12348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.32762425454841e-06,
+      "loss": 0.9805,
+      "step": 12349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.326157944298987e-06,
+      "loss": 1.1484,
+      "step": 12350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.324691725358774e-06,
+      "loss": 0.6797,
+      "step": 12351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.323225597764212e-06,
+      "loss": 0.8633,
+      "step": 12352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.321759561551743e-06,
+      "loss": 1.0391,
+      "step": 12353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.320293616757802e-06,
+      "loss": 0.5898,
+      "step": 12354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.318827763418825e-06,
+      "loss": 0.7656,
+      "step": 12355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3173620015712386e-06,
+      "loss": 0.7539,
+      "step": 12356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3158963312514785e-06,
+      "loss": 0.9844,
+      "step": 12357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3144307524959715e-06,
+      "loss": 0.5742,
+      "step": 12358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.31296526534114e-06,
+      "loss": 0.5547,
+      "step": 12359
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.311499869823407e-06,
+      "loss": 1.3281,
+      "step": 12360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.310034565979199e-06,
+      "loss": 1.3672,
+      "step": 12361
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.308569353844929e-06,
+      "loss": 0.8008,
+      "step": 12362
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3071042334570155e-06,
+      "loss": 0.6055,
+      "step": 12363
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3056392048518675e-06,
+      "loss": 1.375,
+      "step": 12364
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.304174268065903e-06,
+      "loss": 0.7852,
+      "step": 12365
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.3027094231355335e-06,
+      "loss": 0.6094,
+      "step": 12366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.301244670097158e-06,
+      "loss": 0.8125,
+      "step": 12367
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.299780008987184e-06,
+      "loss": 0.6445,
+      "step": 12368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.298315439842014e-06,
+      "loss": 1.1172,
+      "step": 12369
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.296850962698051e-06,
+      "loss": 0.8359,
+      "step": 12370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.295386577591693e-06,
+      "loss": 0.7891,
+      "step": 12371
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.293922284559333e-06,
+      "loss": 1.3438,
+      "step": 12372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.292458083637361e-06,
+      "loss": 0.75,
+      "step": 12373
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.290993974862175e-06,
+      "loss": 0.8203,
+      "step": 12374
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.289529958270162e-06,
+      "loss": 0.3535,
+      "step": 12375
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.288066033897708e-06,
+      "loss": 0.6211,
+      "step": 12376
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.286602201781193e-06,
+      "loss": 0.2441,
+      "step": 12377
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.285138461957005e-06,
+      "loss": 0.4023,
+      "step": 12378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.28367481446152e-06,
+      "loss": 1.7344,
+      "step": 12379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.282211259331117e-06,
+      "loss": 0.9922,
+      "step": 12380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.280747796602169e-06,
+      "loss": 0.6797,
+      "step": 12381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.2792844263110544e-06,
+      "loss": 0.1553,
+      "step": 12382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.2778211484941365e-06,
+      "loss": 0.7969,
+      "step": 12383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.276357963187786e-06,
+      "loss": 0.9375,
+      "step": 12384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.2748948704283675e-06,
+      "loss": 0.7266,
+      "step": 12385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.273431870252248e-06,
+      "loss": 1.0078,
+      "step": 12386
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.271968962695787e-06,
+      "loss": 0.7305,
+      "step": 12387
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.270506147795345e-06,
+      "loss": 0.8398,
+      "step": 12388
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.269043425587273e-06,
+      "loss": 0.6016,
+      "step": 12389
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.267580796107931e-06,
+      "loss": 2.0781,
+      "step": 12390
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.26611825939367e-06,
+      "loss": 1.4922,
+      "step": 12391
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.2646558154808376e-06,
+      "loss": 0.6914,
+      "step": 12392
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.263193464405781e-06,
+      "loss": 1.2344,
+      "step": 12393
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.261731206204851e-06,
+      "loss": 1.0391,
+      "step": 12394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.260269040914384e-06,
+      "loss": 0.668,
+      "step": 12395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.258806968570722e-06,
+      "loss": 1.1562,
+      "step": 12396
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.257344989210202e-06,
+      "loss": 1.2031,
+      "step": 12397
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.255883102869165e-06,
+      "loss": 0.5703,
+      "step": 12398
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.254421309583941e-06,
+      "loss": 0.9297,
+      "step": 12399
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.25295960939086e-06,
+      "loss": 0.7656,
+      "step": 12400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.25149800232625e-06,
+      "loss": 0.7266,
+      "step": 12401
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.250036488426443e-06,
+      "loss": 1.5078,
+      "step": 12402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.24857506772776e-06,
+      "loss": 0.8711,
+      "step": 12403
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.247113740266526e-06,
+      "loss": 1.1094,
+      "step": 12404
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2456525060790515e-06,
+      "loss": 1.0078,
+      "step": 12405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2441913652016635e-06,
+      "loss": 0.7148,
+      "step": 12406
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.242730317670673e-06,
+      "loss": 0.6445,
+      "step": 12407
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2412693635223955e-06,
+      "loss": 0.3145,
+      "step": 12408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.239808502793134e-06,
+      "loss": 0.7305,
+      "step": 12409
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.238347735519208e-06,
+      "loss": 0.9727,
+      "step": 12410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2368870617369135e-06,
+      "loss": 0.8047,
+      "step": 12411
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.235426481482559e-06,
+      "loss": 0.5312,
+      "step": 12412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.233965994792442e-06,
+      "loss": 0.9062,
+      "step": 12413
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.232505601702863e-06,
+      "loss": 1.4453,
+      "step": 12414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.231045302250122e-06,
+      "loss": 0.9961,
+      "step": 12415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.229585096470509e-06,
+      "loss": 0.9297,
+      "step": 12416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2281249844003126e-06,
+      "loss": 0.4473,
+      "step": 12417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.226664966075828e-06,
+      "loss": 0.8594,
+      "step": 12418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.225205041533342e-06,
+      "loss": 1.1406,
+      "step": 12419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.223745210809138e-06,
+      "loss": 0.7617,
+      "step": 12420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.222285473939496e-06,
+      "loss": 0.7344,
+      "step": 12421
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2208258309607e-06,
+      "loss": 0.8516,
+      "step": 12422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.219366281909027e-06,
+      "loss": 0.8555,
+      "step": 12423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2179068268207506e-06,
+      "loss": 0.3848,
+      "step": 12424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.216447465732143e-06,
+      "loss": 1.1797,
+      "step": 12425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.214988198679479e-06,
+      "loss": 1.3047,
+      "step": 12426
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.213529025699028e-06,
+      "loss": 1.0625,
+      "step": 12427
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.212069946827049e-06,
+      "loss": 1.6016,
+      "step": 12428
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.21061096209981e-06,
+      "loss": 1.0469,
+      "step": 12429
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2091520715535725e-06,
+      "loss": 0.4766,
+      "step": 12430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.207693275224597e-06,
+      "loss": 1.7031,
+      "step": 12431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2062345731491396e-06,
+      "loss": 0.3691,
+      "step": 12432
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.20477596536345e-06,
+      "loss": 1.0234,
+      "step": 12433
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.203317451903785e-06,
+      "loss": 1.0312,
+      "step": 12434
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.201859032806394e-06,
+      "loss": 0.4316,
+      "step": 12435
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.2004007081075245e-06,
+      "loss": 1.2344,
+      "step": 12436
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1989424778434185e-06,
+      "loss": 0.7734,
+      "step": 12437
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.197484342050324e-06,
+      "loss": 0.7812,
+      "step": 12438
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.196026300764476e-06,
+      "loss": 0.9297,
+      "step": 12439
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1945683540221145e-06,
+      "loss": 0.1118,
+      "step": 12440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.193110501859476e-06,
+      "loss": 1.0938,
+      "step": 12441
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1916527443127915e-06,
+      "loss": 0.6289,
+      "step": 12442
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.190195081418298e-06,
+      "loss": 0.9023,
+      "step": 12443
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.188737513212215e-06,
+      "loss": 1.0547,
+      "step": 12444
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1872800397307744e-06,
+      "loss": 0.7773,
+      "step": 12445
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.185822661010197e-06,
+      "loss": 1.1562,
+      "step": 12446
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.18436537708671e-06,
+      "loss": 1.2344,
+      "step": 12447
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.182908187996528e-06,
+      "loss": 0.8945,
+      "step": 12448
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.181451093775867e-06,
+      "loss": 0.8203,
+      "step": 12449
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.179994094460942e-06,
+      "loss": 0.4512,
+      "step": 12450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1785371900879676e-06,
+      "loss": 0.9102,
+      "step": 12451
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1770803806931526e-06,
+      "loss": 0.6016,
+      "step": 12452
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.175623666312703e-06,
+      "loss": 0.4336,
+      "step": 12453
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.174167046982823e-06,
+      "loss": 0.4219,
+      "step": 12454
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.17271052273972e-06,
+      "loss": 0.334,
+      "step": 12455
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.17125409361959e-06,
+      "loss": 1.0078,
+      "step": 12456
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.169797759658632e-06,
+      "loss": 0.2695,
+      "step": 12457
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.16834152089304e-06,
+      "loss": 0.9688,
+      "step": 12458
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.166885377359009e-06,
+      "loss": 0.8047,
+      "step": 12459
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.165429329092733e-06,
+      "loss": 0.4902,
+      "step": 12460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1639733761303945e-06,
+      "loss": 0.4336,
+      "step": 12461
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.16251751850818e-06,
+      "loss": 0.5234,
+      "step": 12462
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.161061756262277e-06,
+      "loss": 0.8516,
+      "step": 12463
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1596060894288645e-06,
+      "loss": 0.9258,
+      "step": 12464
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1581505180441245e-06,
+      "loss": 0.7578,
+      "step": 12465
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.156695042144227e-06,
+      "loss": 1.0703,
+      "step": 12466
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.155239661765353e-06,
+      "loss": 1.2656,
+      "step": 12467
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.153784376943671e-06,
+      "loss": 1.0703,
+      "step": 12468
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.152329187715351e-06,
+      "loss": 1.2188,
+      "step": 12469
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.150874094116559e-06,
+      "loss": 0.793,
+      "step": 12470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.1494190961834646e-06,
+      "loss": 0.8516,
+      "step": 12471
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.147964193952224e-06,
+      "loss": 0.8711,
+      "step": 12472
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.146509387459e-06,
+      "loss": 0.9102,
+      "step": 12473
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.145054676739948e-06,
+      "loss": 0.1094,
+      "step": 12474
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.1436000618312265e-06,
+      "loss": 0.9648,
+      "step": 12475
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.142145542768989e-06,
+      "loss": 0.6992,
+      "step": 12476
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.140691119589382e-06,
+      "loss": 1.2969,
+      "step": 12477
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.139236792328553e-06,
+      "loss": 0.7188,
+      "step": 12478
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.137782561022652e-06,
+      "loss": 0.543,
+      "step": 12479
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.136328425707821e-06,
+      "loss": 0.1602,
+      "step": 12480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.134874386420201e-06,
+      "loss": 0.9297,
+      "step": 12481
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.133420443195926e-06,
+      "loss": 1.25,
+      "step": 12482
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.131966596071138e-06,
+      "loss": 0.2305,
+      "step": 12483
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.130512845081968e-06,
+      "loss": 0.7812,
+      "step": 12484
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.1290591902645494e-06,
+      "loss": 0.2324,
+      "step": 12485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.127605631655007e-06,
+      "loss": 0.2344,
+      "step": 12486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.126152169289471e-06,
+      "loss": 0.9141,
+      "step": 12487
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.124698803204068e-06,
+      "loss": 0.8242,
+      "step": 12488
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.123245533434912e-06,
+      "loss": 0.5547,
+      "step": 12489
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.121792360018126e-06,
+      "loss": 0.8047,
+      "step": 12490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.120339282989831e-06,
+      "loss": 1.2422,
+      "step": 12491
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.1188863023861365e-06,
+      "loss": 0.9141,
+      "step": 12492
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.117433418243159e-06,
+      "loss": 0.9023,
+      "step": 12493
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.115980630597003e-06,
+      "loss": 1.1172,
+      "step": 12494
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.114527939483779e-06,
+      "loss": 0.8828,
+      "step": 12495
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.113075344939592e-06,
+      "loss": 0.6914,
+      "step": 12496
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.111622847000544e-06,
+      "loss": 0.7734,
+      "step": 12497
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.110170445702736e-06,
+      "loss": 0.3008,
+      "step": 12498
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.108718141082267e-06,
+      "loss": 0.2891,
+      "step": 12499
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.107265933175229e-06,
+      "loss": 1.0391,
+      "step": 12500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.105813822017718e-06,
+      "loss": 1.2109,
+      "step": 12501
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.10436180764582e-06,
+      "loss": 0.9258,
+      "step": 12502
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.102909890095629e-06,
+      "loss": 0.707,
+      "step": 12503
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.10145806940323e-06,
+      "loss": 1.0078,
+      "step": 12504
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.100006345604703e-06,
+      "loss": 1.0703,
+      "step": 12505
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.098554718736127e-06,
+      "loss": 0.3047,
+      "step": 12506
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.097103188833589e-06,
+      "loss": 1.0469,
+      "step": 12507
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.095651755933159e-06,
+      "loss": 0.2891,
+      "step": 12508
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.094200420070913e-06,
+      "loss": 1.1484,
+      "step": 12509
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.092749181282917e-06,
+      "loss": 1.3672,
+      "step": 12510
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.091298039605248e-06,
+      "loss": 1.2656,
+      "step": 12511
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.089846995073968e-06,
+      "loss": 0.625,
+      "step": 12512
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.088396047725142e-06,
+      "loss": 0.9844,
+      "step": 12513
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.086945197594829e-06,
+      "loss": 0.7734,
+      "step": 12514
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.085494444719096e-06,
+      "loss": 1.1094,
+      "step": 12515
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.084043789133991e-06,
+      "loss": 1.0625,
+      "step": 12516
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.082593230875573e-06,
+      "loss": 0.416,
+      "step": 12517
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.081142769979892e-06,
+      "loss": 0.8125,
+      "step": 12518
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.079692406482998e-06,
+      "loss": 0.668,
+      "step": 12519
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.07824214042094e-06,
+      "loss": 0.7422,
+      "step": 12520
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.076791971829762e-06,
+      "loss": 0.8945,
+      "step": 12521
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.075341900745505e-06,
+      "loss": 1.2109,
+      "step": 12522
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.073891927204207e-06,
+      "loss": 0.875,
+      "step": 12523
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.072442051241912e-06,
+      "loss": 1.125,
+      "step": 12524
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.070992272894649e-06,
+      "loss": 0.3887,
+      "step": 12525
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.069542592198455e-06,
+      "loss": 0.9961,
+      "step": 12526
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.068093009189355e-06,
+      "loss": 0.5039,
+      "step": 12527
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.066643523903382e-06,
+      "loss": 1.1328,
+      "step": 12528
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.065194136376558e-06,
+      "loss": 0.707,
+      "step": 12529
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.063744846644906e-06,
+      "loss": 0.6445,
+      "step": 12530
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.0622956547444455e-06,
+      "loss": 1.2109,
+      "step": 12531
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.060846560711202e-06,
+      "loss": 0.8438,
+      "step": 12532
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.059397564581182e-06,
+      "loss": 1.2656,
+      "step": 12533
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.057948666390403e-06,
+      "loss": 1.3516,
+      "step": 12534
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.056499866174873e-06,
+      "loss": 1.0234,
+      "step": 12535
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.055051163970605e-06,
+      "loss": 0.7383,
+      "step": 12536
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.053602559813602e-06,
+      "loss": 1.1406,
+      "step": 12537
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.052154053739867e-06,
+      "loss": 0.75,
+      "step": 12538
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.050705645785397e-06,
+      "loss": 0.9961,
+      "step": 12539
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.0492573359862e-06,
+      "loss": 1.4375,
+      "step": 12540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.047809124378266e-06,
+      "loss": 1.0781,
+      "step": 12541
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.046361010997592e-06,
+      "loss": 1.0391,
+      "step": 12542
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.044912995880163e-06,
+      "loss": 0.7656,
+      "step": 12543
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.043465079061974e-06,
+      "loss": 1.6172,
+      "step": 12544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.042017260579008e-06,
+      "loss": 0.6055,
+      "step": 12545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.040569540467252e-06,
+      "loss": 1.0234,
+      "step": 12546
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.039121918762682e-06,
+      "loss": 0.8828,
+      "step": 12547
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.037674395501285e-06,
+      "loss": 1.3203,
+      "step": 12548
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.03622697071903e-06,
+      "loss": 0.582,
+      "step": 12549
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.034779644451896e-06,
+      "loss": 0.2988,
+      "step": 12550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.03333241673585e-06,
+      "loss": 0.5273,
+      "step": 12551
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.031885287606866e-06,
+      "loss": 0.6328,
+      "step": 12552
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.030438257100911e-06,
+      "loss": 0.5156,
+      "step": 12553
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.0289913252539455e-06,
+      "loss": 0.2949,
+      "step": 12554
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.0275444921019314e-06,
+      "loss": 0.9766,
+      "step": 12555
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.026097757680831e-06,
+      "loss": 0.5781,
+      "step": 12556
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.024651122026599e-06,
+      "loss": 0.4238,
+      "step": 12557
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.023204585175193e-06,
+      "loss": 0.6562,
+      "step": 12558
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.02175814716256e-06,
+      "loss": 1.1953,
+      "step": 12559
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.020311808024655e-06,
+      "loss": 0.6211,
+      "step": 12560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.0188655677974225e-06,
+      "loss": 0.6602,
+      "step": 12561
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.017419426516806e-06,
+      "loss": 0.5508,
+      "step": 12562
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.015973384218749e-06,
+      "loss": 1.125,
+      "step": 12563
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.014527440939193e-06,
+      "loss": 0.9062,
+      "step": 12564
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.013081596714074e-06,
+      "loss": 0.5703,
+      "step": 12565
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.011635851579325e-06,
+      "loss": 1.0781,
+      "step": 12566
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.010190205570879e-06,
+      "loss": 0.6094,
+      "step": 12567
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.008744658724668e-06,
+      "loss": 0.8867,
+      "step": 12568
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.007299211076617e-06,
+      "loss": 0.6211,
+      "step": 12569
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.005853862662655e-06,
+      "loss": 0.6914,
+      "step": 12570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.004408613518697e-06,
+      "loss": 0.7461,
+      "step": 12571
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.002963463680671e-06,
+      "loss": 1.1172,
+      "step": 12572
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.00151841318449e-06,
+      "loss": 0.3652,
+      "step": 12573
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.000073462066072e-06,
+      "loss": 0.8477,
+      "step": 12574
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.998628610361325e-06,
+      "loss": 0.3711,
+      "step": 12575
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.997183858106166e-06,
+      "loss": 0.7695,
+      "step": 12576
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.995739205336497e-06,
+      "loss": 0.9609,
+      "step": 12577
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.994294652088224e-06,
+      "loss": 0.2559,
+      "step": 12578
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.992850198397251e-06,
+      "loss": 0.7617,
+      "step": 12579
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.991405844299478e-06,
+      "loss": 0.2793,
+      "step": 12580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.989961589830806e-06,
+      "loss": 1.1094,
+      "step": 12581
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.988517435027125e-06,
+      "loss": 0.4844,
+      "step": 12582
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.987073379924327e-06,
+      "loss": 0.7188,
+      "step": 12583
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.985629424558308e-06,
+      "loss": 0.1787,
+      "step": 12584
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.984185568964954e-06,
+      "loss": 1.3672,
+      "step": 12585
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9827418131801515e-06,
+      "loss": 0.2402,
+      "step": 12586
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.981298157239778e-06,
+      "loss": 1.1797,
+      "step": 12587
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.979854601179718e-06,
+      "loss": 0.1963,
+      "step": 12588
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.978411145035851e-06,
+      "loss": 1.0625,
+      "step": 12589
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.976967788844049e-06,
+      "loss": 0.9062,
+      "step": 12590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.975524532640187e-06,
+      "loss": 1.1406,
+      "step": 12591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.974081376460132e-06,
+      "loss": 1.0234,
+      "step": 12592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9726383203397605e-06,
+      "loss": 0.7109,
+      "step": 12593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9711953643149305e-06,
+      "loss": 1.5703,
+      "step": 12594
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9697525084215066e-06,
+      "loss": 0.5898,
+      "step": 12595
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.968309752695347e-06,
+      "loss": 0.9453,
+      "step": 12596
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.966867097172316e-06,
+      "loss": 0.3984,
+      "step": 12597
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.965424541888267e-06,
+      "loss": 0.5664,
+      "step": 12598
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.963982086879051e-06,
+      "loss": 1.0547,
+      "step": 12599
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.962539732180517e-06,
+      "loss": 1.0156,
+      "step": 12600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.961097477828517e-06,
+      "loss": 1.0938,
+      "step": 12601
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.959655323858896e-06,
+      "loss": 0.3164,
+      "step": 12602
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9582132703074976e-06,
+      "loss": 0.7344,
+      "step": 12603
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9567713172101585e-06,
+      "loss": 1.8438,
+      "step": 12604
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9553294646027195e-06,
+      "loss": 0.3887,
+      "step": 12605
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.953887712521018e-06,
+      "loss": 0.5156,
+      "step": 12606
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.952446061000885e-06,
+      "loss": 1.2031,
+      "step": 12607
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.951004510078149e-06,
+      "loss": 0.3125,
+      "step": 12608
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.949563059788643e-06,
+      "loss": 0.3613,
+      "step": 12609
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.948121710168191e-06,
+      "loss": 0.5586,
+      "step": 12610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.946680461252614e-06,
+      "loss": 1.0859,
+      "step": 12611
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.945239313077732e-06,
+      "loss": 1.2031,
+      "step": 12612
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.943798265679367e-06,
+      "loss": 0.7734,
+      "step": 12613
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9423573190933345e-06,
+      "loss": 1.0234,
+      "step": 12614
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.940916473355445e-06,
+      "loss": 0.957,
+      "step": 12615
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.939475728501507e-06,
+      "loss": 1.3359,
+      "step": 12616
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.938035084567333e-06,
+      "loss": 0.9102,
+      "step": 12617
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.936594541588727e-06,
+      "loss": 1.4297,
+      "step": 12618
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.935154099601494e-06,
+      "loss": 0.7695,
+      "step": 12619
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.933713758641431e-06,
+      "loss": 0.7695,
+      "step": 12620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.932273518744337e-06,
+      "loss": 0.7656,
+      "step": 12621
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.93083337994601e-06,
+      "loss": 1.2734,
+      "step": 12622
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.92939334228224e-06,
+      "loss": 0.9727,
+      "step": 12623
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.927953405788817e-06,
+      "loss": 1.1016,
+      "step": 12624
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.926513570501534e-06,
+      "loss": 1.1094,
+      "step": 12625
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9250738364561745e-06,
+      "loss": 0.8359,
+      "step": 12626
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.923634203688518e-06,
+      "loss": 0.4043,
+      "step": 12627
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.922194672234347e-06,
+      "loss": 1.2656,
+      "step": 12628
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.920755242129439e-06,
+      "loss": 2.0,
+      "step": 12629
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.9193159134095724e-06,
+      "loss": 1.0781,
+      "step": 12630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.917876686110518e-06,
+      "loss": 0.8633,
+      "step": 12631
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.916437560268043e-06,
+      "loss": 1.1328,
+      "step": 12632
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.914998535917921e-06,
+      "loss": 0.6758,
+      "step": 12633
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.913559613095914e-06,
+      "loss": 0.4277,
+      "step": 12634
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.9121207918377844e-06,
+      "loss": 0.6562,
+      "step": 12635
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.9106820721792925e-06,
+      "loss": 1.1328,
+      "step": 12636
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.909243454156201e-06,
+      "loss": 0.7383,
+      "step": 12637
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.907804937804258e-06,
+      "loss": 0.5234,
+      "step": 12638
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.906366523159219e-06,
+      "loss": 0.9727,
+      "step": 12639
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.904928210256835e-06,
+      "loss": 0.9688,
+      "step": 12640
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.903489999132854e-06,
+      "loss": 0.9531,
+      "step": 12641
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.902051889823021e-06,
+      "loss": 1.5859,
+      "step": 12642
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.900613882363077e-06,
+      "loss": 1.1406,
+      "step": 12643
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.899175976788762e-06,
+      "loss": 1.3047,
+      "step": 12644
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.897738173135814e-06,
+      "loss": 0.6719,
+      "step": 12645
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.89630047143997e-06,
+      "loss": 0.7383,
+      "step": 12646
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.894862871736964e-06,
+      "loss": 0.7031,
+      "step": 12647
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.893425374062517e-06,
+      "loss": 0.9609,
+      "step": 12648
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.891987978452365e-06,
+      "loss": 1.1094,
+      "step": 12649
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.890550684942229e-06,
+      "loss": 0.9102,
+      "step": 12650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.889113493567833e-06,
+      "loss": 0.9375,
+      "step": 12651
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.887676404364895e-06,
+      "loss": 0.582,
+      "step": 12652
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.886239417369138e-06,
+      "loss": 1.5469,
+      "step": 12653
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8848025326162675e-06,
+      "loss": 0.4199,
+      "step": 12654
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.883365750142002e-06,
+      "loss": 1.2812,
+      "step": 12655
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8819290699820465e-06,
+      "loss": 1.5078,
+      "step": 12656
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.880492492172114e-06,
+      "loss": 0.6367,
+      "step": 12657
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8790560167479045e-06,
+      "loss": 0.6133,
+      "step": 12658
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.877619643745125e-06,
+      "loss": 1.2891,
+      "step": 12659
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.876183373199466e-06,
+      "loss": 0.9883,
+      "step": 12660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.874747205146634e-06,
+      "loss": 0.8633,
+      "step": 12661
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.873311139622316e-06,
+      "loss": 1.2422,
+      "step": 12662
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.87187517666221e-06,
+      "loss": 1.3828,
+      "step": 12663
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8704393163019965e-06,
+      "loss": 1.3281,
+      "step": 12664
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.869003558577375e-06,
+      "loss": 0.8867,
+      "step": 12665
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.867567903524018e-06,
+      "loss": 0.8203,
+      "step": 12666
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.866132351177613e-06,
+      "loss": 0.8398,
+      "step": 12667
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.864696901573839e-06,
+      "loss": 0.8008,
+      "step": 12668
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8632615547483675e-06,
+      "loss": 0.9922,
+      "step": 12669
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.86182631073688e-06,
+      "loss": 0.6719,
+      "step": 12670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.860391169575042e-06,
+      "loss": 0.6211,
+      "step": 12671
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.858956131298524e-06,
+      "loss": 0.4688,
+      "step": 12672
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.857521195942991e-06,
+      "loss": 1.1953,
+      "step": 12673
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.85608636354411e-06,
+      "loss": 0.7422,
+      "step": 12674
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.854651634137543e-06,
+      "loss": 0.9375,
+      "step": 12675
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.853217007758942e-06,
+      "loss": 1.3125,
+      "step": 12676
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.851782484443966e-06,
+      "loss": 0.375,
+      "step": 12677
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8503480642282716e-06,
+      "loss": 0.5234,
+      "step": 12678
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8489137471475066e-06,
+      "loss": 0.3906,
+      "step": 12679
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.847479533237324e-06,
+      "loss": 0.1504,
+      "step": 12680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.846045422533361e-06,
+      "loss": 0.5273,
+      "step": 12681
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.844611415071266e-06,
+      "loss": 0.5117,
+      "step": 12682
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.843177510886682e-06,
+      "loss": 1.5938,
+      "step": 12683
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.841743710015242e-06,
+      "loss": 0.5938,
+      "step": 12684
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.840310012492583e-06,
+      "loss": 0.9062,
+      "step": 12685
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.838876418354343e-06,
+      "loss": 0.2676,
+      "step": 12686
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8374429276361435e-06,
+      "loss": 0.7891,
+      "step": 12687
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.836009540373622e-06,
+      "loss": 1.125,
+      "step": 12688
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.834576256602393e-06,
+      "loss": 0.4531,
+      "step": 12689
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.83314307635809e-06,
+      "loss": 1.0938,
+      "step": 12690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.831709999676324e-06,
+      "loss": 0.6719,
+      "step": 12691
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.830277026592719e-06,
+      "loss": 0.8516,
+      "step": 12692
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.828844157142885e-06,
+      "loss": 1.0312,
+      "step": 12693
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.827411391362437e-06,
+      "loss": 0.9102,
+      "step": 12694
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8259787292869875e-06,
+      "loss": 0.8867,
+      "step": 12695
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.82454617095214e-06,
+      "loss": 0.6602,
+      "step": 12696
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8231137163934955e-06,
+      "loss": 1.5625,
+      "step": 12697
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.821681365646663e-06,
+      "loss": 1.2812,
+      "step": 12698
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.820249118747241e-06,
+      "loss": 0.7852,
+      "step": 12699
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8188169757308255e-06,
+      "loss": 0.7227,
+      "step": 12700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.817384936633006e-06,
+      "loss": 1.0312,
+      "step": 12701
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.815953001489378e-06,
+      "loss": 0.8242,
+      "step": 12702
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.814521170335536e-06,
+      "loss": 0.543,
+      "step": 12703
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.813089443207056e-06,
+      "loss": 0.2676,
+      "step": 12704
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.811657820139529e-06,
+      "loss": 0.8789,
+      "step": 12705
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.810226301168538e-06,
+      "loss": 0.543,
+      "step": 12706
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.808794886329658e-06,
+      "loss": 1.0234,
+      "step": 12707
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8073635756584624e-06,
+      "loss": 0.6797,
+      "step": 12708
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.805932369190528e-06,
+      "loss": 0.3359,
+      "step": 12709
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.80450126696143e-06,
+      "loss": 0.9609,
+      "step": 12710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.803070269006732e-06,
+      "loss": 0.375,
+      "step": 12711
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.801639375361998e-06,
+      "loss": 0.7656,
+      "step": 12712
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.800208586062794e-06,
+      "loss": 0.4336,
+      "step": 12713
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.798777901144685e-06,
+      "loss": 0.8672,
+      "step": 12714
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.797347320643218e-06,
+      "loss": 0.4453,
+      "step": 12715
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.795916844593962e-06,
+      "loss": 1.2422,
+      "step": 12716
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.794486473032458e-06,
+      "loss": 0.9688,
+      "step": 12717
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.793056205994265e-06,
+      "loss": 1.0781,
+      "step": 12718
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.791626043514924e-06,
+      "loss": 0.8633,
+      "step": 12719
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.790195985629986e-06,
+      "loss": 0.8281,
+      "step": 12720
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7887660323749865e-06,
+      "loss": 0.8359,
+      "step": 12721
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.787336183785475e-06,
+      "loss": 1.2109,
+      "step": 12722
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7859064398969775e-06,
+      "loss": 0.3066,
+      "step": 12723
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.784476800745038e-06,
+      "loss": 1.7031,
+      "step": 12724
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.783047266365183e-06,
+      "loss": 0.5117,
+      "step": 12725
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.781617836792942e-06,
+      "loss": 0.4043,
+      "step": 12726
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.780188512063849e-06,
+      "loss": 0.4082,
+      "step": 12727
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7787592922134216e-06,
+      "loss": 0.7969,
+      "step": 12728
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.777330177277181e-06,
+      "loss": 0.9375,
+      "step": 12729
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.775901167290645e-06,
+      "loss": 0.7227,
+      "step": 12730
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7744722622893404e-06,
+      "loss": 0.9648,
+      "step": 12731
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.773043462308768e-06,
+      "loss": 0.6406,
+      "step": 12732
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.771614767384446e-06,
+      "loss": 1.7734,
+      "step": 12733
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.770186177551884e-06,
+      "loss": 0.7617,
+      "step": 12734
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.768757692846587e-06,
+      "loss": 1.1797,
+      "step": 12735
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.767329313304053e-06,
+      "loss": 0.6523,
+      "step": 12736
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.765901038959787e-06,
+      "loss": 0.9609,
+      "step": 12737
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.764472869849291e-06,
+      "loss": 0.9336,
+      "step": 12738
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.763044806008058e-06,
+      "loss": 1.2578,
+      "step": 12739
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7616168474715736e-06,
+      "loss": 1.4219,
+      "step": 12740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.760188994275338e-06,
+      "loss": 0.6289,
+      "step": 12741
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.758761246454832e-06,
+      "loss": 1.3828,
+      "step": 12742
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.757333604045544e-06,
+      "loss": 0.5898,
+      "step": 12743
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.75590606708296e-06,
+      "loss": 0.8008,
+      "step": 12744
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.754478635602554e-06,
+      "loss": 1.5547,
+      "step": 12745
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.753051309639803e-06,
+      "loss": 0.9844,
+      "step": 12746
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7516240892301835e-06,
+      "loss": 0.7852,
+      "step": 12747
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.75019697440917e-06,
+      "loss": 0.5703,
+      "step": 12748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.748769965212231e-06,
+      "loss": 0.7578,
+      "step": 12749
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.747343061674828e-06,
+      "loss": 1.0312,
+      "step": 12750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.745916263832427e-06,
+      "loss": 0.7617,
+      "step": 12751
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7444895717204974e-06,
+      "loss": 0.6172,
+      "step": 12752
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.743062985374486e-06,
+      "loss": 1.4062,
+      "step": 12753
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.741636504829855e-06,
+      "loss": 2.1406,
+      "step": 12754
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.740210130122064e-06,
+      "loss": 0.8555,
+      "step": 12755
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7387838612865544e-06,
+      "loss": 1.7266,
+      "step": 12756
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.737357698358776e-06,
+      "loss": 1.2422,
+      "step": 12757
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7359316413741754e-06,
+      "loss": 0.582,
+      "step": 12758
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.734505690368202e-06,
+      "loss": 1.1641,
+      "step": 12759
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.733079845376289e-06,
+      "loss": 0.9258,
+      "step": 12760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.731654106433873e-06,
+      "loss": 0.9883,
+      "step": 12761
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7302284735763916e-06,
+      "loss": 0.5977,
+      "step": 12762
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.728802946839282e-06,
+      "loss": 1.0781,
+      "step": 12763
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.727377526257966e-06,
+      "loss": 0.5938,
+      "step": 12764
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.725952211867878e-06,
+      "loss": 0.9453,
+      "step": 12765
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7245270037044364e-06,
+      "loss": 0.6953,
+      "step": 12766
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.723101901803069e-06,
+      "loss": 1.2109,
+      "step": 12767
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.721676906199189e-06,
+      "loss": 0.6289,
+      "step": 12768
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.720252016928218e-06,
+      "loss": 0.3965,
+      "step": 12769
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.718827234025566e-06,
+      "loss": 0.8438,
+      "step": 12770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7174025575266465e-06,
+      "loss": 0.9961,
+      "step": 12771
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.715977987466872e-06,
+      "loss": 1.6719,
+      "step": 12772
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.714553523881645e-06,
+      "loss": 0.9297,
+      "step": 12773
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.713129166806366e-06,
+      "loss": 0.6016,
+      "step": 12774
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.711704916276437e-06,
+      "loss": 0.6484,
+      "step": 12775
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.710280772327262e-06,
+      "loss": 0.582,
+      "step": 12776
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.7088567349942325e-06,
+      "loss": 0.7266,
+      "step": 12777
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.707432804312737e-06,
+      "loss": 0.3125,
+      "step": 12778
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.706008980318169e-06,
+      "loss": 0.4414,
+      "step": 12779
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.704585263045923e-06,
+      "loss": 0.2031,
+      "step": 12780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.703161652531372e-06,
+      "loss": 0.2461,
+      "step": 12781
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.701738148809903e-06,
+      "loss": 1.3359,
+      "step": 12782
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.700314751916902e-06,
+      "loss": 0.7734,
+      "step": 12783
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.698891461887741e-06,
+      "loss": 1.0312,
+      "step": 12784
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.697468278757789e-06,
+      "loss": 1.0391,
+      "step": 12785
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.6960452025624236e-06,
+      "loss": 1.1328,
+      "step": 12786
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.694622233337015e-06,
+      "loss": 0.2432,
+      "step": 12787
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.693199371116927e-06,
+      "loss": 0.918,
+      "step": 12788
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.69177661593752e-06,
+      "loss": 1.3984,
+      "step": 12789
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.690353967834158e-06,
+      "loss": 0.9688,
+      "step": 12790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.688931426842203e-06,
+      "loss": 1.2734,
+      "step": 12791
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.687508992997004e-06,
+      "loss": 0.8438,
+      "step": 12792
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.686086666333921e-06,
+      "loss": 1.3203,
+      "step": 12793
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.684664446888297e-06,
+      "loss": 0.7461,
+      "step": 12794
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.683242334695487e-06,
+      "loss": 1.1797,
+      "step": 12795
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.681820329790828e-06,
+      "loss": 1.2969,
+      "step": 12796
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.68039843220967e-06,
+      "loss": 1.0,
+      "step": 12797
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.678976641987346e-06,
+      "loss": 0.7578,
+      "step": 12798
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.677554959159195e-06,
+      "loss": 0.4629,
+      "step": 12799
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.676133383760558e-06,
+      "loss": 0.7539,
+      "step": 12800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.67471191582676e-06,
+      "loss": 0.8633,
+      "step": 12801
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.673290555393128e-06,
+      "loss": 0.6328,
+      "step": 12802
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.671869302494991e-06,
+      "loss": 0.6133,
+      "step": 12803
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.670448157167677e-06,
+      "loss": 0.5508,
+      "step": 12804
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.669027119446504e-06,
+      "loss": 0.8555,
+      "step": 12805
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.667606189366785e-06,
+      "loss": 1.0469,
+      "step": 12806
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.6661853669638385e-06,
+      "loss": 1.5312,
+      "step": 12807
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.664764652272985e-06,
+      "loss": 0.6367,
+      "step": 12808
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.663344045329524e-06,
+      "loss": 1.1094,
+      "step": 12809
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.6619235461687675e-06,
+      "loss": 0.4355,
+      "step": 12810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.6605031548260245e-06,
+      "loss": 1.2969,
+      "step": 12811
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.659082871336594e-06,
+      "loss": 1.2812,
+      "step": 12812
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.657662695735771e-06,
+      "loss": 0.9453,
+      "step": 12813
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.65624262805886e-06,
+      "loss": 1.0469,
+      "step": 12814
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.654822668341148e-06,
+      "loss": 1.125,
+      "step": 12815
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.653402816617934e-06,
+      "loss": 1.6016,
+      "step": 12816
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.651983072924498e-06,
+      "loss": 0.5938,
+      "step": 12817
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.650563437296136e-06,
+      "loss": 0.9844,
+      "step": 12818
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.649143909768122e-06,
+      "loss": 1.2422,
+      "step": 12819
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.647724490375741e-06,
+      "loss": 0.5781,
+      "step": 12820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.646305179154276e-06,
+      "loss": 0.8477,
+      "step": 12821
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.644885976138996e-06,
+      "loss": 1.5469,
+      "step": 12822
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.6434668813651736e-06,
+      "loss": 0.3848,
+      "step": 12823
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.642047894868079e-06,
+      "loss": 0.1953,
+      "step": 12824
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.640629016682986e-06,
+      "loss": 0.8008,
+      "step": 12825
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.639210246845153e-06,
+      "loss": 1.1875,
+      "step": 12826
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.63779158538984e-06,
+      "loss": 0.6875,
+      "step": 12827
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.636373032352309e-06,
+      "loss": 0.4395,
+      "step": 12828
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.634954587767823e-06,
+      "loss": 0.2773,
+      "step": 12829
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.633536251671623e-06,
+      "loss": 0.4785,
+      "step": 12830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.632118024098968e-06,
+      "loss": 0.2676,
+      "step": 12831
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.63069990508511e-06,
+      "loss": 0.8281,
+      "step": 12832
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.6292818946652895e-06,
+      "loss": 0.5625,
+      "step": 12833
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.627863992874747e-06,
+      "loss": 1.3516,
+      "step": 12834
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.626446199748725e-06,
+      "loss": 1.1641,
+      "step": 12835
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.625028515322466e-06,
+      "loss": 0.9297,
+      "step": 12836
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.623610939631199e-06,
+      "loss": 0.3887,
+      "step": 12837
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.622193472710161e-06,
+      "loss": 0.8945,
+      "step": 12838
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.620776114594575e-06,
+      "loss": 0.6836,
+      "step": 12839
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.619358865319677e-06,
+      "loss": 1.625,
+      "step": 12840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.617941724920681e-06,
+      "loss": 0.3906,
+      "step": 12841
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.616524693432818e-06,
+      "loss": 1.6562,
+      "step": 12842
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.615107770891297e-06,
+      "loss": 1.0625,
+      "step": 12843
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.6136909573313435e-06,
+      "loss": 0.7656,
+      "step": 12844
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.612274252788163e-06,
+      "loss": 0.8281,
+      "step": 12845
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.610857657296973e-06,
+      "loss": 1.0859,
+      "step": 12846
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.609441170892975e-06,
+      "loss": 1.4453,
+      "step": 12847
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.6080247936113765e-06,
+      "loss": 0.8906,
+      "step": 12848
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.606608525487385e-06,
+      "loss": 0.8984,
+      "step": 12849
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.605192366556194e-06,
+      "loss": 0.582,
+      "step": 12850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.603776316853e-06,
+      "loss": 1.0781,
+      "step": 12851
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.602360376412998e-06,
+      "loss": 1.0234,
+      "step": 12852
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.600944545271387e-06,
+      "loss": 0.7656,
+      "step": 12853
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.5995288234633494e-06,
+      "loss": 0.8047,
+      "step": 12854
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.59811321102407e-06,
+      "loss": 0.498,
+      "step": 12855
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.596697707988732e-06,
+      "loss": 0.6211,
+      "step": 12856
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.595282314392523e-06,
+      "loss": 0.8516,
+      "step": 12857
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.5938670302706134e-06,
+      "loss": 1.5469,
+      "step": 12858
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.592451855658181e-06,
+      "loss": 1.2031,
+      "step": 12859
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.5910367905904025e-06,
+      "loss": 0.4707,
+      "step": 12860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.589621835102446e-06,
+      "loss": 1.3516,
+      "step": 12861
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.588206989229471e-06,
+      "loss": 1.1875,
+      "step": 12862
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.5867922530066486e-06,
+      "loss": 1.0469,
+      "step": 12863
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.585377626469144e-06,
+      "loss": 0.9102,
+      "step": 12864
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.583963109652107e-06,
+      "loss": 1.0078,
+      "step": 12865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.582548702590703e-06,
+      "loss": 0.4707,
+      "step": 12866
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.581134405320078e-06,
+      "loss": 0.7969,
+      "step": 12867
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.579720217875388e-06,
+      "loss": 0.9102,
+      "step": 12868
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.578306140291777e-06,
+      "loss": 0.7344,
+      "step": 12869
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.576892172604396e-06,
+      "loss": 1.1016,
+      "step": 12870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.575478314848378e-06,
+      "loss": 0.6523,
+      "step": 12871
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.574064567058874e-06,
+      "loss": 1.1172,
+      "step": 12872
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.572650929271012e-06,
+      "loss": 1.0703,
+      "step": 12873
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.571237401519933e-06,
+      "loss": 1.7734,
+      "step": 12874
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.569823983840763e-06,
+      "loss": 1.1406,
+      "step": 12875
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.568410676268634e-06,
+      "loss": 0.104,
+      "step": 12876
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.566997478838675e-06,
+      "loss": 0.0898,
+      "step": 12877
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.5655843915860066e-06,
+      "loss": 0.5312,
+      "step": 12878
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.564171414545745e-06,
+      "loss": 0.5234,
+      "step": 12879
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.562758547753013e-06,
+      "loss": 0.6914,
+      "step": 12880
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.561345791242928e-06,
+      "loss": 0.8242,
+      "step": 12881
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.5599331450506e-06,
+      "loss": 0.8398,
+      "step": 12882
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.558520609211136e-06,
+      "loss": 0.7266,
+      "step": 12883
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.557108183759643e-06,
+      "loss": 0.7812,
+      "step": 12884
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.555695868731232e-06,
+      "loss": 0.3516,
+      "step": 12885
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.554283664160998e-06,
+      "loss": 1.0469,
+      "step": 12886
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.55287157008404e-06,
+      "loss": 0.5391,
+      "step": 12887
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.551459586535461e-06,
+      "loss": 0.6484,
+      "step": 12888
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.550047713550349e-06,
+      "loss": 0.8281,
+      "step": 12889
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.54863595116379e-06,
+      "loss": 0.2314,
+      "step": 12890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.547224299410881e-06,
+      "loss": 0.9883,
+      "step": 12891
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.545812758326698e-06,
+      "loss": 1.0156,
+      "step": 12892
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.5444013279463315e-06,
+      "loss": 1.0078,
+      "step": 12893
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.542990008304853e-06,
+      "loss": 0.8398,
+      "step": 12894
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.541578799437348e-06,
+      "loss": 0.3027,
+      "step": 12895
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.5401677013788825e-06,
+      "loss": 1.4219,
+      "step": 12896
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.538756714164532e-06,
+      "loss": 0.8633,
+      "step": 12897
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.537345837829367e-06,
+      "loss": 1.2031,
+      "step": 12898
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.53593507240845e-06,
+      "loss": 0.793,
+      "step": 12899
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.534524417936842e-06,
+      "loss": 1.7891,
+      "step": 12900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.533113874449605e-06,
+      "loss": 0.5938,
+      "step": 12901
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.531703441981799e-06,
+      "loss": 0.8203,
+      "step": 12902
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.530293120568475e-06,
+      "loss": 0.6289,
+      "step": 12903
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.528882910244688e-06,
+      "loss": 0.8555,
+      "step": 12904
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.527472811045489e-06,
+      "loss": 0.9062,
+      "step": 12905
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.52606282300592e-06,
+      "loss": 0.9727,
+      "step": 12906
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.524652946161024e-06,
+      "loss": 0.6953,
+      "step": 12907
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.523243180545843e-06,
+      "loss": 0.668,
+      "step": 12908
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.521833526195422e-06,
+      "loss": 0.3145,
+      "step": 12909
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.520423983144787e-06,
+      "loss": 1.2812,
+      "step": 12910
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.519014551428973e-06,
+      "loss": 0.4844,
+      "step": 12911
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.51760523108301e-06,
+      "loss": 0.8516,
+      "step": 12912
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.5161960221419305e-06,
+      "loss": 0.6016,
+      "step": 12913
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.51478692464075e-06,
+      "loss": 1.1875,
+      "step": 12914
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.513377938614499e-06,
+      "loss": 0.9336,
+      "step": 12915
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.511969064098186e-06,
+      "loss": 0.8945,
+      "step": 12916
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.510560301126839e-06,
+      "loss": 1.1562,
+      "step": 12917
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.509151649735459e-06,
+      "loss": 0.7031,
+      "step": 12918
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.507743109959067e-06,
+      "loss": 0.8125,
+      "step": 12919
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.506334681832662e-06,
+      "loss": 0.8359,
+      "step": 12920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.504926365391257e-06,
+      "loss": 0.7969,
+      "step": 12921
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.503518160669846e-06,
+      "loss": 0.9414,
+      "step": 12922
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.502110067703437e-06,
+      "loss": 0.4941,
+      "step": 12923
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.500702086527017e-06,
+      "loss": 1.3984,
+      "step": 12924
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.499294217175584e-06,
+      "loss": 0.4102,
+      "step": 12925
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.4978864596841346e-06,
+      "loss": 0.3086,
+      "step": 12926
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.496478814087652e-06,
+      "loss": 0.1768,
+      "step": 12927
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.495071280421118e-06,
+      "loss": 0.9102,
+      "step": 12928
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.493663858719519e-06,
+      "loss": 0.5742,
+      "step": 12929
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.492256549017839e-06,
+      "loss": 0.5273,
+      "step": 12930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.490849351351051e-06,
+      "loss": 0.6719,
+      "step": 12931
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.489442265754122e-06,
+      "loss": 0.4824,
+      "step": 12932
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.488035292262038e-06,
+      "loss": 0.6875,
+      "step": 12933
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.486628430909759e-06,
+      "loss": 0.418,
+      "step": 12934
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.485221681732251e-06,
+      "loss": 0.5195,
+      "step": 12935
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.483815044764479e-06,
+      "loss": 0.1611,
+      "step": 12936
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.482408520041405e-06,
+      "loss": 1.6641,
+      "step": 12937
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.481002107597987e-06,
+      "loss": 1.2578,
+      "step": 12938
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.479595807469172e-06,
+      "loss": 0.9531,
+      "step": 12939
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.478189619689918e-06,
+      "loss": 1.4375,
+      "step": 12940
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.476783544295178e-06,
+      "loss": 1.0,
+      "step": 12941
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.47537758131989e-06,
+      "loss": 0.6133,
+      "step": 12942
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.4739717307990055e-06,
+      "loss": 0.9453,
+      "step": 12943
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.472565992767458e-06,
+      "loss": 0.543,
+      "step": 12944
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.471160367260193e-06,
+      "loss": 1.0312,
+      "step": 12945
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.469754854312137e-06,
+      "loss": 0.9727,
+      "step": 12946
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.468349453958233e-06,
+      "loss": 1.4141,
+      "step": 12947
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.4669441662334e-06,
+      "loss": 0.2988,
+      "step": 12948
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.465538991172575e-06,
+      "loss": 1.2422,
+      "step": 12949
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.464133928810671e-06,
+      "loss": 0.8086,
+      "step": 12950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.46272897918262e-06,
+      "loss": 0.707,
+      "step": 12951
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.461324142323332e-06,
+      "loss": 0.9922,
+      "step": 12952
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.459919418267725e-06,
+      "loss": 0.9492,
+      "step": 12953
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.458514807050719e-06,
+      "loss": 1.0547,
+      "step": 12954
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.457110308707216e-06,
+      "loss": 1.0625,
+      "step": 12955
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.455705923272122e-06,
+      "loss": 0.2197,
+      "step": 12956
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.454301650780345e-06,
+      "loss": 0.6992,
+      "step": 12957
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.452897491266791e-06,
+      "loss": 0.5664,
+      "step": 12958
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.451493444766353e-06,
+      "loss": 1.4141,
+      "step": 12959
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.450089511313924e-06,
+      "loss": 0.4316,
+      "step": 12960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4486856909444015e-06,
+      "loss": 0.5898,
+      "step": 12961
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4472819836926806e-06,
+      "loss": 0.7031,
+      "step": 12962
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.44587838959364e-06,
+      "loss": 1.1328,
+      "step": 12963
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.444474908682172e-06,
+      "loss": 1.2422,
+      "step": 12964
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.44307154099315e-06,
+      "loss": 0.582,
+      "step": 12965
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.441668286561462e-06,
+      "loss": 0.9414,
+      "step": 12966
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.440265145421977e-06,
+      "loss": 1.1406,
+      "step": 12967
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.438862117609576e-06,
+      "loss": 1.1094,
+      "step": 12968
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.437459203159122e-06,
+      "loss": 0.707,
+      "step": 12969
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4360564021054865e-06,
+      "loss": 0.7656,
+      "step": 12970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4346537144835375e-06,
+      "loss": 0.8867,
+      "step": 12971
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.433251140328135e-06,
+      "loss": 0.2891,
+      "step": 12972
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4318486796741335e-06,
+      "loss": 1.0312,
+      "step": 12973
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.430446332556393e-06,
+      "loss": 1.4766,
+      "step": 12974
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.429044099009772e-06,
+      "loss": 0.4238,
+      "step": 12975
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.427641979069118e-06,
+      "loss": 0.9219,
+      "step": 12976
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.426239972769274e-06,
+      "loss": 0.6797,
+      "step": 12977
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.424838080145089e-06,
+      "loss": 1.0781,
+      "step": 12978
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.423436301231413e-06,
+      "loss": 0.4531,
+      "step": 12979
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4220346360630716e-06,
+      "loss": 0.4277,
+      "step": 12980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.420633084674911e-06,
+      "loss": 0.3555,
+      "step": 12981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.419231647101766e-06,
+      "loss": 1.0703,
+      "step": 12982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.417830323378465e-06,
+      "loss": 0.418,
+      "step": 12983
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.416429113539833e-06,
+      "loss": 1.3516,
+      "step": 12984
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.415028017620698e-06,
+      "loss": 0.7148,
+      "step": 12985
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.413627035655887e-06,
+      "loss": 0.3848,
+      "step": 12986
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.412226167680217e-06,
+      "loss": 0.7539,
+      "step": 12987
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4108254137285e-06,
+      "loss": 0.1621,
+      "step": 12988
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.409424773835553e-06,
+      "loss": 0.9844,
+      "step": 12989
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.408024248036193e-06,
+      "loss": 0.5859,
+      "step": 12990
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.406623836365222e-06,
+      "loss": 0.5195,
+      "step": 12991
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.405223538857449e-06,
+      "loss": 0.8008,
+      "step": 12992
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.403823355547673e-06,
+      "loss": 1.1797,
+      "step": 12993
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4024232864707006e-06,
+      "loss": 1.1719,
+      "step": 12994
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4010233316613215e-06,
+      "loss": 0.6406,
+      "step": 12995
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.399623491154336e-06,
+      "loss": 0.5781,
+      "step": 12996
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.398223764984529e-06,
+      "loss": 0.8086,
+      "step": 12997
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.396824153186698e-06,
+      "loss": 1.0312,
+      "step": 12998
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.39542465579562e-06,
+      "loss": 0.6719,
+      "step": 12999
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3940252728460865e-06,
+      "loss": 0.5195,
+      "step": 13000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.392626004372867e-06,
+      "loss": 1.0312,
+      "step": 13001
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.391226850410745e-06,
+      "loss": 0.5898,
+      "step": 13002
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.389827810994499e-06,
+      "loss": 0.5859,
+      "step": 13003
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3884288861588965e-06,
+      "loss": 1.4219,
+      "step": 13004
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.387030075938702e-06,
+      "loss": 0.2168,
+      "step": 13005
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3856313803686834e-06,
+      "loss": 0.4961,
+      "step": 13006
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.384232799483611e-06,
+      "loss": 0.8047,
+      "step": 13007
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.382834333318234e-06,
+      "loss": 1.1094,
+      "step": 13008
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.381435981907317e-06,
+      "loss": 0.9062,
+      "step": 13009
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3800377452856155e-06,
+      "loss": 0.668,
+      "step": 13010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.378639623487878e-06,
+      "loss": 0.5586,
+      "step": 13011
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3772416165488495e-06,
+      "loss": 0.543,
+      "step": 13012
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.37584372450328e-06,
+      "loss": 0.8047,
+      "step": 13013
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3744459473859155e-06,
+      "loss": 1.2578,
+      "step": 13014
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.373048285231495e-06,
+      "loss": 0.7812,
+      "step": 13015
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.371650738074748e-06,
+      "loss": 1.0156,
+      "step": 13016
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3702533059504146e-06,
+      "loss": 0.7344,
+      "step": 13017
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.368855988893231e-06,
+      "loss": 0.8008,
+      "step": 13018
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.367458786937918e-06,
+      "loss": 0.7148,
+      "step": 13019
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3660617001192095e-06,
+      "loss": 0.4453,
+      "step": 13020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.36466472847182e-06,
+      "loss": 1.0156,
+      "step": 13021
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.3632678720304766e-06,
+      "loss": 1.6875,
+      "step": 13022
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.36187113082989e-06,
+      "loss": 1.2109,
+      "step": 13023
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.360474504904782e-06,
+      "loss": 1.7266,
+      "step": 13024
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.359077994289858e-06,
+      "loss": 1.2031,
+      "step": 13025
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.35768159901983e-06,
+      "loss": 0.6641,
+      "step": 13026
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.356285319129402e-06,
+      "loss": 0.6836,
+      "step": 13027
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.354889154653282e-06,
+      "loss": 0.4707,
+      "step": 13028
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.353493105626161e-06,
+      "loss": 1.1797,
+      "step": 13029
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.352097172082742e-06,
+      "loss": 0.8828,
+      "step": 13030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.350701354057721e-06,
+      "loss": 0.5586,
+      "step": 13031
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.349305651585788e-06,
+      "loss": 0.5742,
+      "step": 13032
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.347910064701627e-06,
+      "loss": 0.5547,
+      "step": 13033
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.346514593439929e-06,
+      "loss": 0.6445,
+      "step": 13034
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.345119237835377e-06,
+      "loss": 0.4434,
+      "step": 13035
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.343723997922647e-06,
+      "loss": 0.9688,
+      "step": 13036
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.342328873736421e-06,
+      "loss": 0.707,
+      "step": 13037
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.340933865311369e-06,
+      "loss": 0.6836,
+      "step": 13038
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.339538972682165e-06,
+      "loss": 0.6367,
+      "step": 13039
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.338144195883475e-06,
+      "loss": 0.3477,
+      "step": 13040
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.336749534949969e-06,
+      "loss": 0.9648,
+      "step": 13041
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.335354989916304e-06,
+      "loss": 1.4219,
+      "step": 13042
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3339605608171456e-06,
+      "loss": 1.8125,
+      "step": 13043
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.332566247687143e-06,
+      "loss": 0.5547,
+      "step": 13044
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.331172050560962e-06,
+      "loss": 0.75,
+      "step": 13045
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.329777969473241e-06,
+      "loss": 1.4922,
+      "step": 13046
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.328384004458636e-06,
+      "loss": 0.9727,
+      "step": 13047
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.326990155551792e-06,
+      "loss": 0.9141,
+      "step": 13048
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.325596422787354e-06,
+      "loss": 0.6562,
+      "step": 13049
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.324202806199951e-06,
+      "loss": 0.7031,
+      "step": 13050
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.322809305824228e-06,
+      "loss": 1.2812,
+      "step": 13051
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.321415921694821e-06,
+      "loss": 1.2031,
+      "step": 13052
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.320022653846358e-06,
+      "loss": 0.8164,
+      "step": 13053
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.318629502313463e-06,
+      "loss": 1.6406,
+      "step": 13054
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.317236467130764e-06,
+      "loss": 0.75,
+      "step": 13055
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.315843548332888e-06,
+      "loss": 1.6797,
+      "step": 13056
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3144507459544495e-06,
+      "loss": 0.4531,
+      "step": 13057
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3130580600300645e-06,
+      "loss": 0.5234,
+      "step": 13058
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.311665490594351e-06,
+      "loss": 0.3008,
+      "step": 13059
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.310273037681919e-06,
+      "loss": 0.9062,
+      "step": 13060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.30888070132737e-06,
+      "loss": 0.3945,
+      "step": 13061
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.307488481565313e-06,
+      "loss": 0.6289,
+      "step": 13062
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.306096378430355e-06,
+      "loss": 0.4238,
+      "step": 13063
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.304704391957089e-06,
+      "loss": 0.3867,
+      "step": 13064
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3033125221801106e-06,
+      "loss": 1.1094,
+      "step": 13065
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.301920769134016e-06,
+      "loss": 1.2812,
+      "step": 13066
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.300529132853398e-06,
+      "loss": 0.6016,
+      "step": 13067
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.299137613372837e-06,
+      "loss": 1.1641,
+      "step": 13068
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.297746210726926e-06,
+      "loss": 1.2344,
+      "step": 13069
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.29635492495024e-06,
+      "loss": 0.6289,
+      "step": 13070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2949637560773624e-06,
+      "loss": 0.6953,
+      "step": 13071
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.293572704142865e-06,
+      "loss": 0.4531,
+      "step": 13072
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.292181769181326e-06,
+      "loss": 0.4336,
+      "step": 13073
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.290790951227309e-06,
+      "loss": 0.332,
+      "step": 13074
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.289400250315386e-06,
+      "loss": 0.9688,
+      "step": 13075
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.288009666480123e-06,
+      "loss": 0.3145,
+      "step": 13076
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.286619199756078e-06,
+      "loss": 0.9102,
+      "step": 13077
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.285228850177806e-06,
+      "loss": 0.4766,
+      "step": 13078
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.283838617779868e-06,
+      "loss": 0.7383,
+      "step": 13079
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.282448502596817e-06,
+      "loss": 0.6484,
+      "step": 13080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2810585046632035e-06,
+      "loss": 0.8164,
+      "step": 13081
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.279668624013566e-06,
+      "loss": 0.7031,
+      "step": 13082
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.278278860682455e-06,
+      "loss": 0.9805,
+      "step": 13083
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.276889214704415e-06,
+      "loss": 1.0078,
+      "step": 13084
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.275499686113975e-06,
+      "loss": 0.625,
+      "step": 13085
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2741102749456765e-06,
+      "loss": 0.4355,
+      "step": 13086
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.272720981234052e-06,
+      "loss": 0.875,
+      "step": 13087
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.271331805013631e-06,
+      "loss": 0.5117,
+      "step": 13088
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2699427463189325e-06,
+      "loss": 0.6445,
+      "step": 13089
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.268553805184486e-06,
+      "loss": 0.6484,
+      "step": 13090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.267164981644817e-06,
+      "loss": 1.0625,
+      "step": 13091
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.265776275734435e-06,
+      "loss": 1.0156,
+      "step": 13092
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.264387687487855e-06,
+      "loss": 0.4707,
+      "step": 13093
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.262999216939592e-06,
+      "loss": 0.3984,
+      "step": 13094
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.261610864124157e-06,
+      "loss": 1.4297,
+      "step": 13095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.260222629076049e-06,
+      "loss": 1.1172,
+      "step": 13096
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2588345118297805e-06,
+      "loss": 1.2812,
+      "step": 13097
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2574465124198395e-06,
+      "loss": 0.918,
+      "step": 13098
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.256058630880736e-06,
+      "loss": 0.6172,
+      "step": 13099
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2546708672469535e-06,
+      "loss": 1.7031,
+      "step": 13100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2532832215529916e-06,
+      "loss": 0.6367,
+      "step": 13101
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.251895693833331e-06,
+      "loss": 0.5859,
+      "step": 13102
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.250508284122462e-06,
+      "loss": 0.3457,
+      "step": 13103
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.249120992454868e-06,
+      "loss": 0.5742,
+      "step": 13104
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.247733818865028e-06,
+      "loss": 1.3828,
+      "step": 13105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.246346763387413e-06,
+      "loss": 0.3145,
+      "step": 13106
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2449598260565025e-06,
+      "loss": 0.8477,
+      "step": 13107
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.243573006906769e-06,
+      "loss": 0.8164,
+      "step": 13108
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.242186305972676e-06,
+      "loss": 1.2188,
+      "step": 13109
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.240799723288687e-06,
+      "loss": 0.5977,
+      "step": 13110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.2394132588892675e-06,
+      "loss": 0.9609,
+      "step": 13111
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.238026912808878e-06,
+      "loss": 1.4375,
+      "step": 13112
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.236640685081968e-06,
+      "loss": 0.9805,
+      "step": 13113
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.235254575742999e-06,
+      "loss": 1.2422,
+      "step": 13114
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.233868584826414e-06,
+      "loss": 0.7773,
+      "step": 13115
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2324827123666664e-06,
+      "loss": 0.8477,
+      "step": 13116
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.231096958398192e-06,
+      "loss": 0.7305,
+      "step": 13117
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.229711322955443e-06,
+      "loss": 0.8164,
+      "step": 13118
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2283258060728484e-06,
+      "loss": 0.7539,
+      "step": 13119
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.226940407784851e-06,
+      "loss": 0.7031,
+      "step": 13120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.225555128125875e-06,
+      "loss": 0.8398,
+      "step": 13121
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.22416996713036e-06,
+      "loss": 0.6445,
+      "step": 13122
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.222784924832723e-06,
+      "loss": 1.5469,
+      "step": 13123
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.221400001267393e-06,
+      "loss": 0.9375,
+      "step": 13124
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2200151964687926e-06,
+      "loss": 1.0859,
+      "step": 13125
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.218630510471337e-06,
+      "loss": 0.1416,
+      "step": 13126
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2172459433094365e-06,
+      "loss": 1.1328,
+      "step": 13127
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.215861495017508e-06,
+      "loss": 0.4102,
+      "step": 13128
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.214477165629963e-06,
+      "loss": 0.5391,
+      "step": 13129
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.213092955181205e-06,
+      "loss": 0.4551,
+      "step": 13130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.211708863705631e-06,
+      "loss": 0.6875,
+      "step": 13131
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.210324891237646e-06,
+      "loss": 1.5703,
+      "step": 13132
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.208941037811652e-06,
+      "loss": 0.2041,
+      "step": 13133
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.207557303462034e-06,
+      "loss": 0.4668,
+      "step": 13134
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.206173688223189e-06,
+      "loss": 1.0156,
+      "step": 13135
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.204790192129505e-06,
+      "loss": 0.7852,
+      "step": 13136
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.203406815215367e-06,
+      "loss": 1.4375,
+      "step": 13137
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.202023557515152e-06,
+      "loss": 0.249,
+      "step": 13138
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.200640419063245e-06,
+      "loss": 0.9805,
+      "step": 13139
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1992573998940234e-06,
+      "loss": 1.6484,
+      "step": 13140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.197874500041855e-06,
+      "loss": 1.1719,
+      "step": 13141
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.196491719541115e-06,
+      "loss": 0.5547,
+      "step": 13142
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.195109058426168e-06,
+      "loss": 1.3594,
+      "step": 13143
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1937265167313824e-06,
+      "loss": 0.9023,
+      "step": 13144
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1923440944911126e-06,
+      "loss": 0.5508,
+      "step": 13145
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.190961791739726e-06,
+      "loss": 0.334,
+      "step": 13146
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.18957960851157e-06,
+      "loss": 0.7148,
+      "step": 13147
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.188197544841004e-06,
+      "loss": 0.3555,
+      "step": 13148
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.18681560076237e-06,
+      "loss": 0.416,
+      "step": 13149
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.185433776310025e-06,
+      "loss": 0.334,
+      "step": 13150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1840520715183015e-06,
+      "loss": 0.9531,
+      "step": 13151
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.182670486421546e-06,
+      "loss": 0.7188,
+      "step": 13152
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.181289021054099e-06,
+      "loss": 0.8789,
+      "step": 13153
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.179907675450293e-06,
+      "loss": 0.5703,
+      "step": 13154
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.178526449644455e-06,
+      "loss": 0.8672,
+      "step": 13155
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.177145343670918e-06,
+      "loss": 0.625,
+      "step": 13156
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.17576435756401e-06,
+      "loss": 0.5234,
+      "step": 13157
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.174383491358053e-06,
+      "loss": 0.4629,
+      "step": 13158
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.173002745087361e-06,
+      "loss": 0.9453,
+      "step": 13159
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.171622118786255e-06,
+      "loss": 0.5117,
+      "step": 13160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.170241612489052e-06,
+      "loss": 0.6172,
+      "step": 13161
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.168861226230058e-06,
+      "loss": 1.4609,
+      "step": 13162
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.16748096004358e-06,
+      "loss": 1.2656,
+      "step": 13163
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.166100813963933e-06,
+      "loss": 0.9688,
+      "step": 13164
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1647207880254105e-06,
+      "loss": 1.2422,
+      "step": 13165
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.163340882262309e-06,
+      "loss": 1.2734,
+      "step": 13166
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.161961096708927e-06,
+      "loss": 0.9102,
+      "step": 13167
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.160581431399565e-06,
+      "loss": 1.0703,
+      "step": 13168
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.159201886368506e-06,
+      "loss": 1.0781,
+      "step": 13169
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1578224616500335e-06,
+      "loss": 0.832,
+      "step": 13170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.156443157278437e-06,
+      "loss": 0.8164,
+      "step": 13171
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.155063973288e-06,
+      "loss": 1.0625,
+      "step": 13172
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.153684909712994e-06,
+      "loss": 0.6992,
+      "step": 13173
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.152305966587701e-06,
+      "loss": 1.8672,
+      "step": 13174
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.150927143946385e-06,
+      "loss": 1.0859,
+      "step": 13175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.149548441823322e-06,
+      "loss": 1.8438,
+      "step": 13176
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.148169860252774e-06,
+      "loss": 0.9336,
+      "step": 13177
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.146791399269008e-06,
+      "loss": 0.4141,
+      "step": 13178
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1454130589062774e-06,
+      "loss": 1.0703,
+      "step": 13179
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.144034839198844e-06,
+      "loss": 0.543,
+      "step": 13180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.142656740180967e-06,
+      "loss": 0.3438,
+      "step": 13181
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.14127876188689e-06,
+      "loss": 0.8945,
+      "step": 13182
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.139900904350858e-06,
+      "loss": 0.668,
+      "step": 13183
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1385231676071234e-06,
+      "loss": 1.1328,
+      "step": 13184
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1371455516899304e-06,
+      "loss": 1.0703,
+      "step": 13185
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.135768056633512e-06,
+      "loss": 0.4668,
+      "step": 13186
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.134390682472103e-06,
+      "loss": 0.0981,
+      "step": 13187
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.13301342923994e-06,
+      "loss": 0.9844,
+      "step": 13188
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.131636296971256e-06,
+      "loss": 1.4297,
+      "step": 13189
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.130259285700271e-06,
+      "loss": 0.498,
+      "step": 13190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.128882395461217e-06,
+      "loss": 1.5156,
+      "step": 13191
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.127505626288305e-06,
+      "loss": 1.375,
+      "step": 13192
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.126128978215765e-06,
+      "loss": 0.7031,
+      "step": 13193
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1247524512778014e-06,
+      "loss": 0.6133,
+      "step": 13194
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.123376045508633e-06,
+      "loss": 0.9297,
+      "step": 13195
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.121999760942464e-06,
+      "loss": 1.4062,
+      "step": 13196
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.120623597613507e-06,
+      "loss": 0.5859,
+      "step": 13197
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1192475555559576e-06,
+      "loss": 1.2734,
+      "step": 13198
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1178716348040214e-06,
+      "loss": 0.6562,
+      "step": 13199
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.11649583539189e-06,
+      "loss": 1.0625,
+      "step": 13200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.115120157353761e-06,
+      "loss": 0.875,
+      "step": 13201
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.113744600723828e-06,
+      "loss": 0.6719,
+      "step": 13202
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1123691655362754e-06,
+      "loss": 1.0,
+      "step": 13203
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.110993851825286e-06,
+      "loss": 0.418,
+      "step": 13204
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.109618659625042e-06,
+      "loss": 0.5039,
+      "step": 13205
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.108243588969729e-06,
+      "loss": 0.8555,
+      "step": 13206
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.106868639893514e-06,
+      "loss": 0.9531,
+      "step": 13207
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1054938124305745e-06,
+      "loss": 0.2676,
+      "step": 13208
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.104119106615083e-06,
+      "loss": 0.8359,
+      "step": 13209
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1027445224812035e-06,
+      "loss": 1.7031,
+      "step": 13210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.101370060063096e-06,
+      "loss": 0.9102,
+      "step": 13211
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.099995719394924e-06,
+      "loss": 0.5234,
+      "step": 13212
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0986215005108475e-06,
+      "loss": 1.4766,
+      "step": 13213
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.097247403445021e-06,
+      "loss": 0.7266,
+      "step": 13214
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.09587342823159e-06,
+      "loss": 1.1484,
+      "step": 13215
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.094499574904709e-06,
+      "loss": 1.0391,
+      "step": 13216
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.093125843498526e-06,
+      "loss": 1.2031,
+      "step": 13217
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.091752234047174e-06,
+      "loss": 1.4688,
+      "step": 13218
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.090378746584803e-06,
+      "loss": 0.9648,
+      "step": 13219
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0890053811455396e-06,
+      "loss": 0.957,
+      "step": 13220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.087632137763528e-06,
+      "loss": 1.0156,
+      "step": 13221
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.086259016472889e-06,
+      "loss": 1.8984,
+      "step": 13222
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.084886017307757e-06,
+      "loss": 1.3203,
+      "step": 13223
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.08351314030225e-06,
+      "loss": 1.0469,
+      "step": 13224
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.082140385490495e-06,
+      "loss": 0.3945,
+      "step": 13225
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.080767752906606e-06,
+      "loss": 0.5352,
+      "step": 13226
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.079395242584704e-06,
+      "loss": 0.6523,
+      "step": 13227
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.078022854558891e-06,
+      "loss": 0.9922,
+      "step": 13228
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.076650588863284e-06,
+      "loss": 0.3262,
+      "step": 13229
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0752784455319926e-06,
+      "loss": 0.2754,
+      "step": 13230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0739064245991134e-06,
+      "loss": 1.1094,
+      "step": 13231
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.072534526098745e-06,
+      "loss": 0.6055,
+      "step": 13232
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0711627500649865e-06,
+      "loss": 0.3379,
+      "step": 13233
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.069791096531937e-06,
+      "loss": 1.0312,
+      "step": 13234
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.068419565533682e-06,
+      "loss": 0.7539,
+      "step": 13235
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.067048157104309e-06,
+      "loss": 1.2422,
+      "step": 13236
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0656768712779015e-06,
+      "loss": 0.166,
+      "step": 13237
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.064305708088548e-06,
+      "loss": 0.8359,
+      "step": 13238
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.062934667570321e-06,
+      "loss": 0.25,
+      "step": 13239
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.061563749757298e-06,
+      "loss": 1.2422,
+      "step": 13240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.060192954683556e-06,
+      "loss": 1.0859,
+      "step": 13241
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.05882228238316e-06,
+      "loss": 0.6992,
+      "step": 13242
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.057451732890173e-06,
+      "loss": 0.9922,
+      "step": 13243
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.056081306238663e-06,
+      "loss": 0.8906,
+      "step": 13244
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.054711002462693e-06,
+      "loss": 1.0781,
+      "step": 13245
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0533408215963156e-06,
+      "loss": 1.1094,
+      "step": 13246
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0519707636735886e-06,
+      "loss": 0.5742,
+      "step": 13247
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.050600828728559e-06,
+      "loss": 0.9805,
+      "step": 13248
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.049231016795281e-06,
+      "loss": 0.7656,
+      "step": 13249
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.047861327907792e-06,
+      "loss": 0.8867,
+      "step": 13250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.046491762100142e-06,
+      "loss": 0.2324,
+      "step": 13251
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.045122319406363e-06,
+      "loss": 0.4336,
+      "step": 13252
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0437529998604976e-06,
+      "loss": 0.6758,
+      "step": 13253
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.042383803496571e-06,
+      "loss": 0.7383,
+      "step": 13254
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.041014730348623e-06,
+      "loss": 0.6953,
+      "step": 13255
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0396457804506694e-06,
+      "loss": 0.4961,
+      "step": 13256
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.03827695383674e-06,
+      "loss": 0.8203,
+      "step": 13257
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0369082505408586e-06,
+      "loss": 1.3438,
+      "step": 13258
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.035539670597038e-06,
+      "loss": 0.7812,
+      "step": 13259
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0341712140392905e-06,
+      "loss": 1.4922,
+      "step": 13260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.032802880901631e-06,
+      "loss": 0.6758,
+      "step": 13261
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.03143467121807e-06,
+      "loss": 1.2266,
+      "step": 13262
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.030066585022613e-06,
+      "loss": 0.8828,
+      "step": 13263
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.028698622349253e-06,
+      "loss": 1.5781,
+      "step": 13264
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.027330783231997e-06,
+      "loss": 0.9102,
+      "step": 13265
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.025963067704845e-06,
+      "loss": 1.2578,
+      "step": 13266
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.024595475801779e-06,
+      "loss": 0.707,
+      "step": 13267
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0232280075568e-06,
+      "loss": 0.7617,
+      "step": 13268
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.021860663003884e-06,
+      "loss": 1.0,
+      "step": 13269
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.020493442177026e-06,
+      "loss": 0.5117,
+      "step": 13270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0191263451101966e-06,
+      "loss": 0.7188,
+      "step": 13271
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.017759371837383e-06,
+      "loss": 0.832,
+      "step": 13272
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.01639252239255e-06,
+      "loss": 1.0625,
+      "step": 13273
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.015025796809676e-06,
+      "loss": 0.9141,
+      "step": 13274
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.0136591951227286e-06,
+      "loss": 0.8516,
+      "step": 13275
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.012292717365674e-06,
+      "loss": 0.8008,
+      "step": 13276
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.010926363572468e-06,
+      "loss": 0.3047,
+      "step": 13277
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.0095601337770735e-06,
+      "loss": 0.5117,
+      "step": 13278
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.008194028013453e-06,
+      "loss": 0.6328,
+      "step": 13279
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.006828046315552e-06,
+      "loss": 0.4941,
+      "step": 13280
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.00546218871732e-06,
+      "loss": 0.9531,
+      "step": 13281
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.004096455252706e-06,
+      "loss": 0.2178,
+      "step": 13282
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.002730845955658e-06,
+      "loss": 0.7148,
+      "step": 13283
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.001365360860108e-06,
+      "loss": 0.2617,
+      "step": 13284
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.999999999999998e-06,
+      "loss": 0.2354,
+      "step": 13285
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.998634763409268e-06,
+      "loss": 0.1494,
+      "step": 13286
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.997269651121843e-06,
+      "loss": 0.4941,
+      "step": 13287
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.995904663171649e-06,
+      "loss": 0.4805,
+      "step": 13288
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.994539799592615e-06,
+      "loss": 0.8203,
+      "step": 13289
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.993175060418666e-06,
+      "loss": 0.8672,
+      "step": 13290
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.991810445683718e-06,
+      "loss": 0.3672,
+      "step": 13291
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.990445955421683e-06,
+      "loss": 1.0859,
+      "step": 13292
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.989081589666479e-06,
+      "loss": 0.9258,
+      "step": 13293
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.987717348452017e-06,
+      "loss": 1.0,
+      "step": 13294
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.986353231812198e-06,
+      "loss": 0.7344,
+      "step": 13295
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.984989239780933e-06,
+      "loss": 0.4297,
+      "step": 13296
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.983625372392114e-06,
+      "loss": 0.6875,
+      "step": 13297
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.982261629679647e-06,
+      "loss": 0.7656,
+      "step": 13298
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.980898011677418e-06,
+      "loss": 1.0312,
+      "step": 13299
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.979534518419326e-06,
+      "loss": 0.3594,
+      "step": 13300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.978171149939251e-06,
+      "loss": 0.6055,
+      "step": 13301
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.976807906271086e-06,
+      "loss": 0.9062,
+      "step": 13302
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.975444787448705e-06,
+      "loss": 0.6523,
+      "step": 13303
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.974081793505996e-06,
+      "loss": 1.3594,
+      "step": 13304
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.972718924476825e-06,
+      "loss": 0.2656,
+      "step": 13305
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.971356180395069e-06,
+      "loss": 0.7227,
+      "step": 13306
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.969993561294601e-06,
+      "loss": 1.2266,
+      "step": 13307
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.968631067209286e-06,
+      "loss": 0.582,
+      "step": 13308
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.96726869817298e-06,
+      "loss": 1.0703,
+      "step": 13309
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.965906454219549e-06,
+      "loss": 0.7617,
+      "step": 13310
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.964544335382853e-06,
+      "loss": 0.9727,
+      "step": 13311
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.963182341696738e-06,
+      "loss": 1.5938,
+      "step": 13312
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.96182047319506e-06,
+      "loss": 0.5977,
+      "step": 13313
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.96045872991167e-06,
+      "loss": 0.7539,
+      "step": 13314
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.959097111880408e-06,
+      "loss": 0.5312,
+      "step": 13315
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9577356191351125e-06,
+      "loss": 0.9258,
+      "step": 13316
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.956374251709625e-06,
+      "loss": 0.8828,
+      "step": 13317
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.955013009637785e-06,
+      "loss": 0.7109,
+      "step": 13318
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.953651892953421e-06,
+      "loss": 0.7773,
+      "step": 13319
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.952290901690359e-06,
+      "loss": 0.6289,
+      "step": 13320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9509300358824255e-06,
+      "loss": 1.0859,
+      "step": 13321
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9495692955634515e-06,
+      "loss": 0.7031,
+      "step": 13322
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.948208680767245e-06,
+      "loss": 0.6602,
+      "step": 13323
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9468481915276336e-06,
+      "loss": 1.5391,
+      "step": 13324
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.94548782787842e-06,
+      "loss": 0.2227,
+      "step": 13325
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9441275898534245e-06,
+      "loss": 0.9883,
+      "step": 13326
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.942767477486445e-06,
+      "loss": 1.5078,
+      "step": 13327
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9414074908112955e-06,
+      "loss": 1.0156,
+      "step": 13328
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.940047629861767e-06,
+      "loss": 0.2471,
+      "step": 13329
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.938687894671667e-06,
+      "loss": 0.9805,
+      "step": 13330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.937328285274779e-06,
+      "loss": 0.2236,
+      "step": 13331
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9359688017049055e-06,
+      "loss": 0.3789,
+      "step": 13332
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.934609443995827e-06,
+      "loss": 0.2754,
+      "step": 13333
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.933250212181332e-06,
+      "loss": 0.5898,
+      "step": 13334
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.931891106295206e-06,
+      "loss": 2.125,
+      "step": 13335
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.930532126371224e-06,
+      "loss": 0.9062,
+      "step": 13336
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.92917327244316e-06,
+      "loss": 0.9688,
+      "step": 13337
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.927814544544789e-06,
+      "loss": 1.1484,
+      "step": 13338
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9264559427098846e-06,
+      "loss": 0.9805,
+      "step": 13339
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.92509746697221e-06,
+      "loss": 1.7188,
+      "step": 13340
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.923739117365524e-06,
+      "loss": 0.4258,
+      "step": 13341
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.922380893923592e-06,
+      "loss": 0.6719,
+      "step": 13342
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.921022796680173e-06,
+      "loss": 1.0703,
+      "step": 13343
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.919664825669015e-06,
+      "loss": 1.375,
+      "step": 13344
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.918306980923876e-06,
+      "loss": 0.9961,
+      "step": 13345
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.916949262478496e-06,
+      "loss": 0.6328,
+      "step": 13346
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.915591670366628e-06,
+      "loss": 1.6875,
+      "step": 13347
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.914234204622004e-06,
+      "loss": 0.7969,
+      "step": 13348
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.912876865278371e-06,
+      "loss": 0.2432,
+      "step": 13349
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.911519652369456e-06,
+      "loss": 0.9883,
+      "step": 13350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.910162565928997e-06,
+      "loss": 1.375,
+      "step": 13351
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.908805605990723e-06,
+      "loss": 1.4766,
+      "step": 13352
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.907448772588358e-06,
+      "loss": 1.3203,
+      "step": 13353
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.906092065755622e-06,
+      "loss": 0.9883,
+      "step": 13354
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.904735485526235e-06,
+      "loss": 0.9844,
+      "step": 13355
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.90337903193392e-06,
+      "loss": 0.4473,
+      "step": 13356
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.902022705012386e-06,
+      "loss": 0.6836,
+      "step": 13357
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.900666504795337e-06,
+      "loss": 0.5156,
+      "step": 13358
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8993104313164855e-06,
+      "loss": 1.0781,
+      "step": 13359
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8979544846095385e-06,
+      "loss": 0.5352,
+      "step": 13360
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.896598664708189e-06,
+      "loss": 0.8242,
+      "step": 13361
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.895242971646138e-06,
+      "loss": 0.3887,
+      "step": 13362
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8938874054570825e-06,
+      "loss": 1.4453,
+      "step": 13363
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.892531966174713e-06,
+      "loss": 0.8789,
+      "step": 13364
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.89117665383271e-06,
+      "loss": 0.9492,
+      "step": 13365
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.889821468464764e-06,
+      "loss": 0.918,
+      "step": 13366
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.88846641010456e-06,
+      "loss": 1.0859,
+      "step": 13367
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8871114787857735e-06,
+      "loss": 0.4922,
+      "step": 13368
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.885756674542074e-06,
+      "loss": 1.2422,
+      "step": 13369
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.884401997407139e-06,
+      "loss": 0.7266,
+      "step": 13370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.883047447414641e-06,
+      "loss": 0.9297,
+      "step": 13371
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.881693024598238e-06,
+      "loss": 1.4453,
+      "step": 13372
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.880338728991599e-06,
+      "loss": 1.0391,
+      "step": 13373
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.878984560628379e-06,
+      "loss": 0.8203,
+      "step": 13374
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.877630519542239e-06,
+      "loss": 0.9961,
+      "step": 13375
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8762766057668265e-06,
+      "loss": 0.6562,
+      "step": 13376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.874922819335799e-06,
+      "loss": 0.3633,
+      "step": 13377
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.873569160282794e-06,
+      "loss": 1.1641,
+      "step": 13378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.87221562864146e-06,
+      "loss": 0.5,
+      "step": 13379
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.870862224445441e-06,
+      "loss": 1.1797,
+      "step": 13380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.869508947728372e-06,
+      "loss": 0.75,
+      "step": 13381
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.868155798523882e-06,
+      "loss": 1.2891,
+      "step": 13382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.866802776865605e-06,
+      "loss": 0.5469,
+      "step": 13383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.865449882787174e-06,
+      "loss": 0.9766,
+      "step": 13384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8640971163222095e-06,
+      "loss": 1.1484,
+      "step": 13385
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.86274447750433e-06,
+      "loss": 0.6758,
+      "step": 13386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.861391966367156e-06,
+      "loss": 1.4766,
+      "step": 13387
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.860039582944308e-06,
+      "loss": 0.9297,
+      "step": 13388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.858687327269389e-06,
+      "loss": 0.918,
+      "step": 13389
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8573351993760135e-06,
+      "loss": 0.6875,
+      "step": 13390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.855983199297788e-06,
+      "loss": 1.1094,
+      "step": 13391
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.854631327068314e-06,
+      "loss": 0.7422,
+      "step": 13392
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8532795827211855e-06,
+      "loss": 0.7773,
+      "step": 13393
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.851927966290002e-06,
+      "loss": 1.8516,
+      "step": 13394
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8505764778083615e-06,
+      "loss": 1.0156,
+      "step": 13395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.849225117309849e-06,
+      "loss": 0.5977,
+      "step": 13396
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.847873884828048e-06,
+      "loss": 0.2236,
+      "step": 13397
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8465227803965446e-06,
+      "loss": 0.5703,
+      "step": 13398
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.845171804048923e-06,
+      "loss": 1.3281,
+      "step": 13399
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.843820955818755e-06,
+      "loss": 0.6289,
+      "step": 13400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.842470235739618e-06,
+      "loss": 0.3926,
+      "step": 13401
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.841119643845077e-06,
+      "loss": 0.7969,
+      "step": 13402
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.839769180168709e-06,
+      "loss": 0.6016,
+      "step": 13403
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.838418844744066e-06,
+      "loss": 0.4375,
+      "step": 13404
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.837068637604722e-06,
+      "loss": 0.5508,
+      "step": 13405
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.835718558784223e-06,
+      "loss": 0.7539,
+      "step": 13406
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8343686083161325e-06,
+      "loss": 0.8164,
+      "step": 13407
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.833018786233996e-06,
+      "loss": 0.3477,
+      "step": 13408
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8316690925713686e-06,
+      "loss": 1.5391,
+      "step": 13409
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.830319527361787e-06,
+      "loss": 0.9922,
+      "step": 13410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.828970090638796e-06,
+      "loss": 1.2734,
+      "step": 13411
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.827620782435941e-06,
+      "loss": 0.707,
+      "step": 13412
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.826271602786752e-06,
+      "loss": 1.2656,
+      "step": 13413
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.824922551724758e-06,
+      "loss": 1.1641,
+      "step": 13414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8235736292834916e-06,
+      "loss": 1.1094,
+      "step": 13415
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.822224835496482e-06,
+      "loss": 0.7188,
+      "step": 13416
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.820876170397246e-06,
+      "loss": 0.8633,
+      "step": 13417
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.81952763401931e-06,
+      "loss": 0.5977,
+      "step": 13418
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8181792263961816e-06,
+      "loss": 0.75,
+      "step": 13419
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.816830947561383e-06,
+      "loss": 0.9531,
+      "step": 13420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.815482797548415e-06,
+      "loss": 0.5508,
+      "step": 13421
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.814134776390795e-06,
+      "loss": 1.2422,
+      "step": 13422
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.812786884122016e-06,
+      "loss": 0.7891,
+      "step": 13423
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.811439120775587e-06,
+      "loss": 0.6523,
+      "step": 13424
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8100914863849965e-06,
+      "loss": 0.75,
+      "step": 13425
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.808743980983749e-06,
+      "loss": 1.2188,
+      "step": 13426
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.807396604605324e-06,
+      "loss": 0.6836,
+      "step": 13427
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.806049357283215e-06,
+      "loss": 1.2812,
+      "step": 13428
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.80470223905091e-06,
+      "loss": 0.5977,
+      "step": 13429
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.803355249941886e-06,
+      "loss": 1.3594,
+      "step": 13430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8020083899896155e-06,
+      "loss": 0.4297,
+      "step": 13431
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.800661659227579e-06,
+      "loss": 1.2031,
+      "step": 13432
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7993150576892526e-06,
+      "loss": 0.5039,
+      "step": 13433
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.797968585408098e-06,
+      "loss": 0.9961,
+      "step": 13434
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.796622242417578e-06,
+      "loss": 0.2178,
+      "step": 13435
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.795276028751159e-06,
+      "loss": 0.9219,
+      "step": 13436
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.793929944442301e-06,
+      "loss": 0.957,
+      "step": 13437
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.792583989524456e-06,
+      "loss": 1.5859,
+      "step": 13438
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.791238164031074e-06,
+      "loss": 0.6992,
+      "step": 13439
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.789892467995613e-06,
+      "loss": 1.5391,
+      "step": 13440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.788546901451513e-06,
+      "loss": 1.0078,
+      "step": 13441
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.787201464432213e-06,
+      "loss": 0.8008,
+      "step": 13442
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.785856156971155e-06,
+      "loss": 1.1172,
+      "step": 13443
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.784510979101781e-06,
+      "loss": 0.8906,
+      "step": 13444
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.783165930857514e-06,
+      "loss": 0.5273,
+      "step": 13445
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.781821012271793e-06,
+      "loss": 0.6016,
+      "step": 13446
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.780476223378036e-06,
+      "loss": 0.3477,
+      "step": 13447
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.779131564209674e-06,
+      "loss": 0.6016,
+      "step": 13448
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.77778703480012e-06,
+      "loss": 1.0781,
+      "step": 13449
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.776442635182798e-06,
+      "loss": 0.3223,
+      "step": 13450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.775098365391112e-06,
+      "loss": 1.1719,
+      "step": 13451
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.773754225458483e-06,
+      "loss": 0.6367,
+      "step": 13452
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.772410215418309e-06,
+      "loss": 1.3203,
+      "step": 13453
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.771066335304001e-06,
+      "loss": 1.0156,
+      "step": 13454
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.769722585148954e-06,
+      "loss": 1.0469,
+      "step": 13455
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.768378964986567e-06,
+      "loss": 1.0078,
+      "step": 13456
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.767035474850238e-06,
+      "loss": 1.7266,
+      "step": 13457
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.765692114773356e-06,
+      "loss": 0.3438,
+      "step": 13458
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7643488847893036e-06,
+      "loss": 0.4023,
+      "step": 13459
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.763005784931469e-06,
+      "loss": 1.0938,
+      "step": 13460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.761662815233238e-06,
+      "loss": 1.5547,
+      "step": 13461
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.760319975727984e-06,
+      "loss": 0.3047,
+      "step": 13462
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.758977266449077e-06,
+      "loss": 0.9297,
+      "step": 13463
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7576346874298965e-06,
+      "loss": 0.707,
+      "step": 13464
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.75629223870381e-06,
+      "loss": 0.6836,
+      "step": 13465
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.754949920304176e-06,
+      "loss": 0.4238,
+      "step": 13466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7536077322643626e-06,
+      "loss": 1.0781,
+      "step": 13467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7522656746177284e-06,
+      "loss": 1.3516,
+      "step": 13468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.750923747397629e-06,
+      "loss": 0.582,
+      "step": 13469
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7495819506374086e-06,
+      "loss": 1.0156,
+      "step": 13470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.748240284370423e-06,
+      "loss": 0.875,
+      "step": 13471
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.74689874863002e-06,
+      "loss": 0.3125,
+      "step": 13472
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.74555734344954e-06,
+      "loss": 1.8047,
+      "step": 13473
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.744216068862315e-06,
+      "loss": 1.0547,
+      "step": 13474
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.742874924901687e-06,
+      "loss": 0.5391,
+      "step": 13475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.741533911600992e-06,
+      "loss": 0.9805,
+      "step": 13476
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.740193028993552e-06,
+      "loss": 0.8438,
+      "step": 13477
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.738852277112701e-06,
+      "loss": 0.8438,
+      "step": 13478
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.737511655991753e-06,
+      "loss": 0.5078,
+      "step": 13479
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7361711656640365e-06,
+      "loss": 0.2266,
+      "step": 13480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.73483080616286e-06,
+      "loss": 0.7109,
+      "step": 13481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.733490577521543e-06,
+      "loss": 1.0078,
+      "step": 13482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.73215047977339e-06,
+      "loss": 0.1348,
+      "step": 13483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.73081051295171e-06,
+      "loss": 1.1328,
+      "step": 13484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7294706770898105e-06,
+      "loss": 0.7383,
+      "step": 13485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7281309722209865e-06,
+      "loss": 1.7891,
+      "step": 13486
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.726791398378533e-06,
+      "loss": 0.5508,
+      "step": 13487
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7254519555957464e-06,
+      "loss": 0.8906,
+      "step": 13488
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7241126439059225e-06,
+      "loss": 0.4473,
+      "step": 13489
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.722773463342342e-06,
+      "loss": 0.5156,
+      "step": 13490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721434413938286e-06,
+      "loss": 0.7188,
+      "step": 13491
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.72009549572704e-06,
+      "loss": 1.0391,
+      "step": 13492
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.718756708741883e-06,
+      "loss": 0.5859,
+      "step": 13493
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717418053016083e-06,
+      "loss": 1.4609,
+      "step": 13494
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.716079528582918e-06,
+      "loss": 1.2422,
+      "step": 13495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.714741135475648e-06,
+      "loss": 0.7031,
+      "step": 13496
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.713402873727544e-06,
+      "loss": 0.7148,
+      "step": 13497
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.71206474337186e-06,
+      "loss": 0.9844,
+      "step": 13498
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7107267444418625e-06,
+      "loss": 0.3301,
+      "step": 13499
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.709388876970796e-06,
+      "loss": 1.3516,
+      "step": 13500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.708051140991921e-06,
+      "loss": 0.3613,
+      "step": 13501
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7067135365384785e-06,
+      "loss": 1.1328,
+      "step": 13502
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.705376063643719e-06,
+      "loss": 1.3203,
+      "step": 13503
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.704038722340876e-06,
+      "loss": 0.2393,
+      "step": 13504
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.702701512663194e-06,
+      "loss": 1.4375,
+      "step": 13505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.701364434643908e-06,
+      "loss": 0.7539,
+      "step": 13506
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.700027488316249e-06,
+      "loss": 0.9688,
+      "step": 13507
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.69869067371344e-06,
+      "loss": 0.5508,
+      "step": 13508
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6973539908687105e-06,
+      "loss": 0.8516,
+      "step": 13509
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.696017439815284e-06,
+      "loss": 0.9453,
+      "step": 13510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.694681020586379e-06,
+      "loss": 0.9883,
+      "step": 13511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.693344733215202e-06,
+      "loss": 0.7305,
+      "step": 13512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.692008577734974e-06,
+      "loss": 0.7031,
+      "step": 13513
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.690672554178904e-06,
+      "loss": 0.7461,
+      "step": 13514
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.689336662580191e-06,
+      "loss": 0.8789,
+      "step": 13515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.68800090297204e-06,
+      "loss": 1.2656,
+      "step": 13516
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.686665275387654e-06,
+      "loss": 0.8516,
+      "step": 13517
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.685329779860225e-06,
+      "loss": 0.7656,
+      "step": 13518
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.68399441642294e-06,
+      "loss": 0.8594,
+      "step": 13519
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.682659185108994e-06,
+      "loss": 0.7578,
+      "step": 13520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.681324085951576e-06,
+      "loss": 0.9688,
+      "step": 13521
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.67998911898386e-06,
+      "loss": 0.7734,
+      "step": 13522
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.678654284239032e-06,
+      "loss": 0.9922,
+      "step": 13523
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.677319581750263e-06,
+      "loss": 0.6406,
+      "step": 13524
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6759850115507295e-06,
+      "loss": 0.5781,
+      "step": 13525
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.674650573673596e-06,
+      "loss": 0.1592,
+      "step": 13526
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.673316268152035e-06,
+      "loss": 1.6094,
+      "step": 13527
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.671982095019202e-06,
+      "loss": 1.4922,
+      "step": 13528
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.670648054308262e-06,
+      "loss": 0.3867,
+      "step": 13529
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.669314146052366e-06,
+      "loss": 0.9492,
+      "step": 13530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.667980370284674e-06,
+      "loss": 0.1807,
+      "step": 13531
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.666646727038327e-06,
+      "loss": 0.832,
+      "step": 13532
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.665313216346474e-06,
+      "loss": 1.3438,
+      "step": 13533
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.663979838242264e-06,
+      "loss": 0.5664,
+      "step": 13534
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.662646592758832e-06,
+      "loss": 0.9766,
+      "step": 13535
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6613134799293095e-06,
+      "loss": 0.7305,
+      "step": 13536
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.659980499786835e-06,
+      "loss": 0.2773,
+      "step": 13537
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.65864765236454e-06,
+      "loss": 1.6562,
+      "step": 13538
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.657314937695551e-06,
+      "loss": 0.5273,
+      "step": 13539
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.655982355812983e-06,
+      "loss": 0.6914,
+      "step": 13540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.65464990674996e-06,
+      "loss": 0.6836,
+      "step": 13541
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6533175905396045e-06,
+      "loss": 1.2656,
+      "step": 13542
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.651985407215022e-06,
+      "loss": 1.5156,
+      "step": 13543
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.650653356809325e-06,
+      "loss": 0.7148,
+      "step": 13544
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.649321439355624e-06,
+      "loss": 1.0469,
+      "step": 13545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.647989654887018e-06,
+      "loss": 0.6562,
+      "step": 13546
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.646658003436606e-06,
+      "loss": 1.2422,
+      "step": 13547
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.645326485037486e-06,
+      "loss": 1.1172,
+      "step": 13548
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.643995099722755e-06,
+      "loss": 1.0938,
+      "step": 13549
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.642663847525496e-06,
+      "loss": 1.1562,
+      "step": 13550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.641332728478805e-06,
+      "loss": 0.8164,
+      "step": 13551
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.640001742615755e-06,
+      "loss": 1.3047,
+      "step": 13552
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.638670889969437e-06,
+      "loss": 1.5859,
+      "step": 13553
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.637340170572916e-06,
+      "loss": 0.582,
+      "step": 13554
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.636009584459277e-06,
+      "loss": 0.3926,
+      "step": 13555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6346791316615815e-06,
+      "loss": 0.3535,
+      "step": 13556
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.633348812212903e-06,
+      "loss": 0.543,
+      "step": 13557
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.632018626146298e-06,
+      "loss": 0.7734,
+      "step": 13558
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.630688573494836e-06,
+      "loss": 0.8477,
+      "step": 13559
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.629358654291563e-06,
+      "loss": 0.6211,
+      "step": 13560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.62802886856954e-06,
+      "loss": 0.3145,
+      "step": 13561
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.626699216361819e-06,
+      "loss": 0.7266,
+      "step": 13562
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6253696977014435e-06,
+      "loss": 0.7148,
+      "step": 13563
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.624040312621455e-06,
+      "loss": 0.8477,
+      "step": 13564
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.622711061154895e-06,
+      "loss": 1.2031,
+      "step": 13565
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6213819433348065e-06,
+      "loss": 1.2031,
+      "step": 13566
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.620052959194218e-06,
+      "loss": 0.8359,
+      "step": 13567
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.618724108766156e-06,
+      "loss": 0.5234,
+      "step": 13568
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.617395392083653e-06,
+      "loss": 1.25,
+      "step": 13569
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.616066809179735e-06,
+      "loss": 0.832,
+      "step": 13570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.614738360087417e-06,
+      "loss": 0.8164,
+      "step": 13571
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.61341004483972e-06,
+      "loss": 0.7812,
+      "step": 13572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.612081863469654e-06,
+      "loss": 0.9062,
+      "step": 13573
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.610753816010234e-06,
+      "loss": 0.6367,
+      "step": 13574
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.609425902494462e-06,
+      "loss": 1.6094,
+      "step": 13575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.608098122955348e-06,
+      "loss": 0.3672,
+      "step": 13576
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.606770477425885e-06,
+      "loss": 1.2578,
+      "step": 13577
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.605442965939079e-06,
+      "loss": 0.8008,
+      "step": 13578
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.604115588527913e-06,
+      "loss": 0.2539,
+      "step": 13579
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6027883452253884e-06,
+      "loss": 1.375,
+      "step": 13580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.601461236064484e-06,
+      "loss": 1.8359,
+      "step": 13581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.600134261078186e-06,
+      "loss": 0.8594,
+      "step": 13582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5988074202994805e-06,
+      "loss": 0.5469,
+      "step": 13583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.59748071376134e-06,
+      "loss": 0.6523,
+      "step": 13584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.596154141496734e-06,
+      "loss": 0.5781,
+      "step": 13585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.594827703538638e-06,
+      "loss": 0.6367,
+      "step": 13586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.593501399920021e-06,
+      "loss": 0.9453,
+      "step": 13587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.592175230673841e-06,
+      "loss": 0.7578,
+      "step": 13588
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.590849195833061e-06,
+      "loss": 0.8008,
+      "step": 13589
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.589523295430643e-06,
+      "loss": 1.2734,
+      "step": 13590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.588197529499536e-06,
+      "loss": 1.0781,
+      "step": 13591
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.586871898072687e-06,
+      "loss": 1.2344,
+      "step": 13592
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.585546401183048e-06,
+      "loss": 0.2246,
+      "step": 13593
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.584221038863564e-06,
+      "loss": 1.4766,
+      "step": 13594
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.582895811147174e-06,
+      "loss": 0.3027,
+      "step": 13595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.58157071806681e-06,
+      "loss": 0.5195,
+      "step": 13596
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.58024575965541e-06,
+      "loss": 1.2891,
+      "step": 13597
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.578920935945908e-06,
+      "loss": 0.9453,
+      "step": 13598
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.577596246971222e-06,
+      "loss": 0.4004,
+      "step": 13599
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.576271692764286e-06,
+      "loss": 0.668,
+      "step": 13600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.574947273358011e-06,
+      "loss": 0.8828,
+      "step": 13601
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.573622988785322e-06,
+      "loss": 0.7539,
+      "step": 13602
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.572298839079125e-06,
+      "loss": 0.5195,
+      "step": 13603
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.570974824272337e-06,
+      "loss": 1.1016,
+      "step": 13604
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5696509443978586e-06,
+      "loss": 0.9805,
+      "step": 13605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5683271994886e-06,
+      "loss": 0.625,
+      "step": 13606
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.567003589577454e-06,
+      "loss": 0.793,
+      "step": 13607
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5656801146973265e-06,
+      "loss": 1.2188,
+      "step": 13608
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5643567748811014e-06,
+      "loss": 0.7227,
+      "step": 13609
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.563033570161672e-06,
+      "loss": 0.7852,
+      "step": 13610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.561710500571932e-06,
+      "loss": 0.7695,
+      "step": 13611
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.560387566144758e-06,
+      "loss": 1.2266,
+      "step": 13612
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.559064766913028e-06,
+      "loss": 0.8242,
+      "step": 13613
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.557742102909621e-06,
+      "loss": 1.4219,
+      "step": 13614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.556419574167416e-06,
+      "loss": 1.1562,
+      "step": 13615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.555097180719274e-06,
+      "loss": 0.457,
+      "step": 13616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.553774922598067e-06,
+      "loss": 0.9688,
+      "step": 13617
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.552452799836659e-06,
+      "loss": 0.9609,
+      "step": 13618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5511308124679085e-06,
+      "loss": 0.793,
+      "step": 13619
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.549808960524668e-06,
+      "loss": 0.7461,
+      "step": 13620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.548487244039793e-06,
+      "loss": 0.9336,
+      "step": 13621
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5471656630461395e-06,
+      "loss": 1.2109,
+      "step": 13622
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.545844217576547e-06,
+      "loss": 1.625,
+      "step": 13623
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.544522907663858e-06,
+      "loss": 0.2969,
+      "step": 13624
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.543201733340914e-06,
+      "loss": 1.6094,
+      "step": 13625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.541880694640556e-06,
+      "loss": 1.0859,
+      "step": 13626
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.540559791595608e-06,
+      "loss": 0.2422,
+      "step": 13627
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.539239024238909e-06,
+      "loss": 1.4609,
+      "step": 13628
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5379183926032755e-06,
+      "loss": 0.5391,
+      "step": 13629
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.53659789672154e-06,
+      "loss": 0.8203,
+      "step": 13630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.535277536626512e-06,
+      "loss": 1.1797,
+      "step": 13631
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.533957312351018e-06,
+      "loss": 0.5742,
+      "step": 13632
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.532637223927865e-06,
+      "loss": 0.8164,
+      "step": 13633
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.531317271389859e-06,
+      "loss": 0.8203,
+      "step": 13634
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.529997454769809e-06,
+      "loss": 0.9844,
+      "step": 13635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.528677774100523e-06,
+      "loss": 0.8359,
+      "step": 13636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.527358229414791e-06,
+      "loss": 0.4668,
+      "step": 13637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.526038820745413e-06,
+      "loss": 0.375,
+      "step": 13638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.524719548125186e-06,
+      "loss": 0.4609,
+      "step": 13639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.523400411586894e-06,
+      "loss": 0.9336,
+      "step": 13640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5220814111633205e-06,
+      "loss": 0.832,
+      "step": 13641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.52076254688725e-06,
+      "loss": 0.5742,
+      "step": 13642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.519443818791466e-06,
+      "loss": 1.25,
+      "step": 13643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.518125226908739e-06,
+      "loss": 1.2344,
+      "step": 13644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.51680677127184e-06,
+      "loss": 0.8125,
+      "step": 13645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5154884519135385e-06,
+      "loss": 1.1406,
+      "step": 13646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.514170268866605e-06,
+      "loss": 0.4258,
+      "step": 13647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.512852222163795e-06,
+      "loss": 0.8359,
+      "step": 13648
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.511534311837873e-06,
+      "loss": 1.0234,
+      "step": 13649
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.510216537921587e-06,
+      "loss": 0.3848,
+      "step": 13650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.508898900447697e-06,
+      "loss": 1.625,
+      "step": 13651
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.507581399448942e-06,
+      "loss": 1.6484,
+      "step": 13652
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.5062640349580766e-06,
+      "loss": 0.2539,
+      "step": 13653
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.504946807007834e-06,
+      "loss": 0.75,
+      "step": 13654
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.503629715630956e-06,
+      "loss": 0.9961,
+      "step": 13655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.502312760860181e-06,
+      "loss": 1.4453,
+      "step": 13656
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.500995942728239e-06,
+      "loss": 0.6602,
+      "step": 13657
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.499679261267851e-06,
+      "loss": 0.7109,
+      "step": 13658
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4983627165117475e-06,
+      "loss": 0.6172,
+      "step": 13659
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.497046308492652e-06,
+      "loss": 0.6719,
+      "step": 13660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.49573003724328e-06,
+      "loss": 0.457,
+      "step": 13661
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.494413902796342e-06,
+      "loss": 1.0156,
+      "step": 13662
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.493097905184551e-06,
+      "loss": 0.6914,
+      "step": 13663
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.491782044440621e-06,
+      "loss": 1.0859,
+      "step": 13664
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.490466320597247e-06,
+      "loss": 0.6172,
+      "step": 13665
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.489150733687134e-06,
+      "loss": 1.8281,
+      "step": 13666
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.487835283742984e-06,
+      "loss": 1.0703,
+      "step": 13667
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.486519970797486e-06,
+      "loss": 0.9844,
+      "step": 13668
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4852047948833275e-06,
+      "loss": 0.5508,
+      "step": 13669
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4838897560331995e-06,
+      "loss": 0.6016,
+      "step": 13670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.482574854279789e-06,
+      "loss": 0.7188,
+      "step": 13671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.481260089655773e-06,
+      "loss": 1.7031,
+      "step": 13672
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.479945462193826e-06,
+      "loss": 1.0859,
+      "step": 13673
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.478630971926623e-06,
+      "loss": 1.6172,
+      "step": 13674
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.47731661888684e-06,
+      "loss": 1.0,
+      "step": 13675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.476002403107135e-06,
+      "loss": 0.5938,
+      "step": 13676
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.47468832462018e-06,
+      "loss": 0.5859,
+      "step": 13677
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.473374383458625e-06,
+      "loss": 0.4609,
+      "step": 13678
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.472060579655138e-06,
+      "loss": 0.8438,
+      "step": 13679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4707469132423595e-06,
+      "loss": 0.1855,
+      "step": 13680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.469433384252951e-06,
+      "loss": 0.7656,
+      "step": 13681
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.46811999271955e-06,
+      "loss": 0.3125,
+      "step": 13682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.466806738674807e-06,
+      "loss": 0.3535,
+      "step": 13683
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4654936221513536e-06,
+      "loss": 0.3789,
+      "step": 13684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.464180643181832e-06,
+      "loss": 0.3398,
+      "step": 13685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.46286780179887e-06,
+      "loss": 0.5391,
+      "step": 13686
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4615550980350995e-06,
+      "loss": 0.875,
+      "step": 13687
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.460242531923149e-06,
+      "loss": 0.7539,
+      "step": 13688
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.458930103495638e-06,
+      "loss": 0.3574,
+      "step": 13689
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.457617812785181e-06,
+      "loss": 1.4766,
+      "step": 13690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.456305659824399e-06,
+      "loss": 0.2832,
+      "step": 13691
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.454993644645907e-06,
+      "loss": 1.1641,
+      "step": 13692
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.453681767282304e-06,
+      "loss": 1.0078,
+      "step": 13693
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.452370027766202e-06,
+      "loss": 1.0391,
+      "step": 13694
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4510584261302044e-06,
+      "loss": 0.4824,
+      "step": 13695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.449746962406907e-06,
+      "loss": 0.6133,
+      "step": 13696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.448435636628901e-06,
+      "loss": 1.375,
+      "step": 13697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.447124448828783e-06,
+      "loss": 0.8789,
+      "step": 13698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.44581339903914e-06,
+      "loss": 0.7773,
+      "step": 13699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.444502487292559e-06,
+      "loss": 0.5273,
+      "step": 13700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.443191713621614e-06,
+      "loss": 0.8906,
+      "step": 13701
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.441881078058887e-06,
+      "loss": 0.4453,
+      "step": 13702
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.440570580636956e-06,
+      "loss": 1.2812,
+      "step": 13703
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.439260221388385e-06,
+      "loss": 0.3535,
+      "step": 13704
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.43795000034575e-06,
+      "loss": 0.3574,
+      "step": 13705
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4366399175416054e-06,
+      "loss": 0.6094,
+      "step": 13706
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.435329973008521e-06,
+      "loss": 0.5547,
+      "step": 13707
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.434020166779045e-06,
+      "loss": 1.0156,
+      "step": 13708
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4327104988857395e-06,
+      "loss": 0.9219,
+      "step": 13709
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.431400969361151e-06,
+      "loss": 1.3516,
+      "step": 13710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.430091578237823e-06,
+      "loss": 0.9688,
+      "step": 13711
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.428782325548303e-06,
+      "loss": 0.9258,
+      "step": 13712
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.427473211325134e-06,
+      "loss": 1.1484,
+      "step": 13713
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.426164235600844e-06,
+      "loss": 1.2734,
+      "step": 13714
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4248553984079725e-06,
+      "loss": 0.7031,
+      "step": 13715
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.423546699779051e-06,
+      "loss": 0.9648,
+      "step": 13716
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.422238139746602e-06,
+      "loss": 1.1484,
+      "step": 13717
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.420929718343146e-06,
+      "loss": 0.4902,
+      "step": 13718
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4196214356012044e-06,
+      "loss": 1.3125,
+      "step": 13719
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.418313291553298e-06,
+      "loss": 0.7656,
+      "step": 13720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4170052862319305e-06,
+      "loss": 0.8828,
+      "step": 13721
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.41569741966962e-06,
+      "loss": 0.2891,
+      "step": 13722
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4143896918988625e-06,
+      "loss": 0.9727,
+      "step": 13723
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.413082102952171e-06,
+      "loss": 1.2266,
+      "step": 13724
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.411774652862033e-06,
+      "loss": 0.377,
+      "step": 13725
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.410467341660954e-06,
+      "loss": 0.9688,
+      "step": 13726
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.409160169381415e-06,
+      "loss": 0.4766,
+      "step": 13727
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.407853136055915e-06,
+      "loss": 0.4277,
+      "step": 13728
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.40654624171693e-06,
+      "loss": 0.3184,
+      "step": 13729
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4052394863969485e-06,
+      "loss": 0.3496,
+      "step": 13730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.403932870128443e-06,
+      "loss": 0.8867,
+      "step": 13731
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.402626392943889e-06,
+      "loss": 0.2012,
+      "step": 13732
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4013200548757616e-06,
+      "loss": 0.4004,
+      "step": 13733
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4000138559565265e-06,
+      "loss": 0.6367,
+      "step": 13734
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.398707796218642e-06,
+      "loss": 1.125,
+      "step": 13735
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.397401875694574e-06,
+      "loss": 0.5391,
+      "step": 13736
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.396096094416784e-06,
+      "loss": 0.6289,
+      "step": 13737
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.39479045241772e-06,
+      "loss": 1.4922,
+      "step": 13738
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.39348494972983e-06,
+      "loss": 1.3672,
+      "step": 13739
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.392179586385563e-06,
+      "loss": 0.5391,
+      "step": 13740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.390874362417367e-06,
+      "loss": 0.4629,
+      "step": 13741
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.389569277857675e-06,
+      "loss": 0.5352,
+      "step": 13742
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.388264332738926e-06,
+      "loss": 0.5742,
+      "step": 13743
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.386959527093558e-06,
+      "loss": 0.3672,
+      "step": 13744
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.385654860953996e-06,
+      "loss": 0.7578,
+      "step": 13745
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.384350334352662e-06,
+      "loss": 0.7773,
+      "step": 13746
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.383045947321984e-06,
+      "loss": 1.1562,
+      "step": 13747
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.381741699894381e-06,
+      "loss": 1.3125,
+      "step": 13748
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.380437592102268e-06,
+      "loss": 1.0703,
+      "step": 13749
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.379133623978053e-06,
+      "loss": 0.7188,
+      "step": 13750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.377829795554147e-06,
+      "loss": 1.3125,
+      "step": 13751
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.376526106862961e-06,
+      "loss": 0.875,
+      "step": 13752
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3752225579368875e-06,
+      "loss": 0.625,
+      "step": 13753
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.373919148808335e-06,
+      "loss": 1.1172,
+      "step": 13754
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.372615879509686e-06,
+      "loss": 0.8984,
+      "step": 13755
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.371312750073344e-06,
+      "loss": 1.0781,
+      "step": 13756
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.370009760531685e-06,
+      "loss": 1.4297,
+      "step": 13757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.368706910917105e-06,
+      "loss": 1.3047,
+      "step": 13758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.367404201261975e-06,
+      "loss": 0.9531,
+      "step": 13759
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.366101631598676e-06,
+      "loss": 0.7188,
+      "step": 13760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.364799201959587e-06,
+      "loss": 0.3887,
+      "step": 13761
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.363496912377073e-06,
+      "loss": 1.1797,
+      "step": 13762
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.362194762883499e-06,
+      "loss": 1.0469,
+      "step": 13763
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.360892753511232e-06,
+      "loss": 1.0234,
+      "step": 13764
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.359590884292634e-06,
+      "loss": 1.6562,
+      "step": 13765
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3582891552600594e-06,
+      "loss": 0.5078,
+      "step": 13766
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.356987566445856e-06,
+      "loss": 0.6289,
+      "step": 13767
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.35568611788238e-06,
+      "loss": 0.9023,
+      "step": 13768
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.354384809601977e-06,
+      "loss": 0.5938,
+      "step": 13769
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.353083641636985e-06,
+      "loss": 0.8516,
+      "step": 13770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.351782614019746e-06,
+      "loss": 0.9922,
+      "step": 13771
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.350481726782599e-06,
+      "loss": 0.5742,
+      "step": 13772
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3491809799578735e-06,
+      "loss": 0.2969,
+      "step": 13773
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.347880373577893e-06,
+      "loss": 0.7812,
+      "step": 13774
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.346579907674987e-06,
+      "loss": 0.3535,
+      "step": 13775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3452795822814805e-06,
+      "loss": 0.4883,
+      "step": 13776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.343979397429689e-06,
+      "loss": 0.2451,
+      "step": 13777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.342679353151922e-06,
+      "loss": 0.9375,
+      "step": 13778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.341379449480494e-06,
+      "loss": 0.6641,
+      "step": 13779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.340079686447718e-06,
+      "loss": 0.6797,
+      "step": 13780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.33878006408589e-06,
+      "loss": 1.3125,
+      "step": 13781
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.3374805824273184e-06,
+      "loss": 0.9805,
+      "step": 13782
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.336181241504296e-06,
+      "loss": 0.9609,
+      "step": 13783
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.334882041349113e-06,
+      "loss": 0.543,
+      "step": 13784
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.333582981994064e-06,
+      "loss": 1.2656,
+      "step": 13785
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.332284063471437e-06,
+      "loss": 1.9688,
+      "step": 13786
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.330985285813515e-06,
+      "loss": 1.3828,
+      "step": 13787
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.329686649052567e-06,
+      "loss": 1.5156,
+      "step": 13788
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.328388153220886e-06,
+      "loss": 0.291,
+      "step": 13789
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.327089798350736e-06,
+      "loss": 1.0391,
+      "step": 13790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.325791584474385e-06,
+      "loss": 0.8164,
+      "step": 13791
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.324493511624099e-06,
+      "loss": 0.6641,
+      "step": 13792
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.323195579832144e-06,
+      "loss": 0.7305,
+      "step": 13793
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.321897789130778e-06,
+      "loss": 0.5703,
+      "step": 13794
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.32060013955225e-06,
+      "loss": 1.1641,
+      "step": 13795
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.319302631128815e-06,
+      "loss": 0.75,
+      "step": 13796
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.318005263892727e-06,
+      "loss": 1.2578,
+      "step": 13797
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.316708037876222e-06,
+      "loss": 0.8242,
+      "step": 13798
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.315410953111549e-06,
+      "loss": 0.7539,
+      "step": 13799
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.314114009630936e-06,
+      "loss": 1.3203,
+      "step": 13800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.312817207466626e-06,
+      "loss": 0.8906,
+      "step": 13801
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.311520546650844e-06,
+      "loss": 0.9766,
+      "step": 13802
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.310224027215821e-06,
+      "loss": 1.3281,
+      "step": 13803
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.308927649193776e-06,
+      "loss": 1.2891,
+      "step": 13804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.307631412616934e-06,
+      "loss": 1.6094,
+      "step": 13805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.306335317517506e-06,
+      "loss": 0.4648,
+      "step": 13806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.305039363927712e-06,
+      "loss": 0.707,
+      "step": 13807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.303743551879753e-06,
+      "loss": 0.957,
+      "step": 13808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.30244788140584e-06,
+      "loss": 1.5469,
+      "step": 13809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.30115235253818e-06,
+      "loss": 0.9062,
+      "step": 13810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2998569653089635e-06,
+      "loss": 1.2031,
+      "step": 13811
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2985617197503886e-06,
+      "loss": 1.0391,
+      "step": 13812
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.297266615894646e-06,
+      "loss": 1.4766,
+      "step": 13813
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.29597165377393e-06,
+      "loss": 0.7812,
+      "step": 13814
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.294676833420421e-06,
+      "loss": 1.1797,
+      "step": 13815
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.293382154866297e-06,
+      "loss": 0.793,
+      "step": 13816
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2920876181437395e-06,
+      "loss": 0.6719,
+      "step": 13817
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.290793223284926e-06,
+      "loss": 0.8828,
+      "step": 13818
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.28949897032202e-06,
+      "loss": 0.7305,
+      "step": 13819
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.288204859287193e-06,
+      "loss": 0.6992,
+      "step": 13820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.286910890212611e-06,
+      "loss": 0.8008,
+      "step": 13821
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.285617063130433e-06,
+      "loss": 1.4219,
+      "step": 13822
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.284323378072809e-06,
+      "loss": 1.4688,
+      "step": 13823
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.283029835071897e-06,
+      "loss": 0.418,
+      "step": 13824
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.28173643415985e-06,
+      "loss": 1.1641,
+      "step": 13825
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.280443175368807e-06,
+      "loss": 1.2578,
+      "step": 13826
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.279150058730919e-06,
+      "loss": 0.0791,
+      "step": 13827
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2778570842783155e-06,
+      "loss": 0.2734,
+      "step": 13828
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.27656425204314e-06,
+      "loss": 0.2168,
+      "step": 13829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.275271562057517e-06,
+      "loss": 0.3047,
+      "step": 13830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.273979014353582e-06,
+      "loss": 1.3125,
+      "step": 13831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.272686608963454e-06,
+      "loss": 0.1895,
+      "step": 13832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.27139434591926e-06,
+      "loss": 1.4844,
+      "step": 13833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.27010222525311e-06,
+      "loss": 1.0703,
+      "step": 13834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.268810246997128e-06,
+      "loss": 0.4004,
+      "step": 13835
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.267518411183414e-06,
+      "loss": 0.7852,
+      "step": 13836
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.266226717844082e-06,
+      "loss": 0.4238,
+      "step": 13837
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.264935167011238e-06,
+      "loss": 0.8203,
+      "step": 13838
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.263643758716978e-06,
+      "loss": 0.7305,
+      "step": 13839
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.262352492993394e-06,
+      "loss": 0.8398,
+      "step": 13840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.261061369872584e-06,
+      "loss": 1.7578,
+      "step": 13841
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.259770389386641e-06,
+      "loss": 0.4766,
+      "step": 13842
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.258479551567647e-06,
+      "loss": 0.4316,
+      "step": 13843
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2571888564476805e-06,
+      "loss": 1.0078,
+      "step": 13844
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.255898304058824e-06,
+      "loss": 0.9805,
+      "step": 13845
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.254607894433157e-06,
+      "loss": 0.875,
+      "step": 13846
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.253317627602742e-06,
+      "loss": 0.9375,
+      "step": 13847
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.252027503599654e-06,
+      "loss": 0.6289,
+      "step": 13848
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.250737522455958e-06,
+      "loss": 0.8867,
+      "step": 13849
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.249447684203712e-06,
+      "loss": 1.875,
+      "step": 13850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.248157988874971e-06,
+      "loss": 0.8906,
+      "step": 13851
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2468684365017935e-06,
+      "loss": 1.0703,
+      "step": 13852
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.24557902711623e-06,
+      "loss": 1.4219,
+      "step": 13853
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.244289760750327e-06,
+      "loss": 1.0781,
+      "step": 13854
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.243000637436122e-06,
+      "loss": 1.5156,
+      "step": 13855
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.24171165720566e-06,
+      "loss": 0.707,
+      "step": 13856
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.24042282009098e-06,
+      "loss": 1.1953,
+      "step": 13857
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.239134126124107e-06,
+      "loss": 0.707,
+      "step": 13858
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.237845575337079e-06,
+      "loss": 1.1172,
+      "step": 13859
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.236557167761914e-06,
+      "loss": 1.1328,
+      "step": 13860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.235268903430635e-06,
+      "loss": 0.8242,
+      "step": 13861
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.233980782375261e-06,
+      "loss": 0.6797,
+      "step": 13862
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2326928046278105e-06,
+      "loss": 0.6758,
+      "step": 13863
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2314049702202886e-06,
+      "loss": 1.0234,
+      "step": 13864
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.230117279184707e-06,
+      "loss": 0.8594,
+      "step": 13865
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.228829731553072e-06,
+      "loss": 0.5156,
+      "step": 13866
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2275423273573815e-06,
+      "loss": 0.8945,
+      "step": 13867
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.226255066629628e-06,
+      "loss": 0.9336,
+      "step": 13868
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2249679494018094e-06,
+      "loss": 0.7422,
+      "step": 13869
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.223680975705919e-06,
+      "loss": 0.9453,
+      "step": 13870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.222394145573939e-06,
+      "loss": 0.7383,
+      "step": 13871
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.221107459037849e-06,
+      "loss": 1.8984,
+      "step": 13872
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2198209161296315e-06,
+      "loss": 1.2188,
+      "step": 13873
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2185345168812656e-06,
+      "loss": 1.0469,
+      "step": 13874
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2172482613247165e-06,
+      "loss": 0.2637,
+      "step": 13875
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.21596214949196e-06,
+      "loss": 1.0781,
+      "step": 13876
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.214676181414952e-06,
+      "loss": 0.707,
+      "step": 13877
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.213390357125663e-06,
+      "loss": 1.25,
+      "step": 13878
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2121046766560434e-06,
+      "loss": 0.2178,
+      "step": 13879
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.210819140038054e-06,
+      "loss": 0.9609,
+      "step": 13880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2095337473036366e-06,
+      "loss": 0.957,
+      "step": 13881
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.208248498484749e-06,
+      "loss": 0.9453,
+      "step": 13882
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.206963393613323e-06,
+      "loss": 0.1309,
+      "step": 13883
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.20567843272131e-06,
+      "loss": 0.5156,
+      "step": 13884
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.204393615840637e-06,
+      "loss": 1.3828,
+      "step": 13885
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.203108943003239e-06,
+      "loss": 0.1006,
+      "step": 13886
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.201824414241051e-06,
+      "loss": 0.6641,
+      "step": 13887
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2005400295859944e-06,
+      "loss": 0.6328,
+      "step": 13888
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1992557890699855e-06,
+      "loss": 1.4922,
+      "step": 13889
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.197971692724949e-06,
+      "loss": 0.6172,
+      "step": 13890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.196687740582802e-06,
+      "loss": 0.6484,
+      "step": 13891
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1954039326754494e-06,
+      "loss": 1.5547,
+      "step": 13892
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1941202690348e-06,
+      "loss": 1.2344,
+      "step": 13893
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.192836749692764e-06,
+      "loss": 1.0234,
+      "step": 13894
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.191553374681237e-06,
+      "loss": 1.0859,
+      "step": 13895
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1902701440321115e-06,
+      "loss": 1.3438,
+      "step": 13896
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1889870577772866e-06,
+      "loss": 2.0625,
+      "step": 13897
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.187704115948654e-06,
+      "loss": 1.0859,
+      "step": 13898
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.186421318578096e-06,
+      "loss": 0.6133,
+      "step": 13899
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.185138665697493e-06,
+      "loss": 1.875,
+      "step": 13900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.183856157338725e-06,
+      "loss": 0.4531,
+      "step": 13901
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.182573793533673e-06,
+      "loss": 1.5859,
+      "step": 13902
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.181291574314201e-06,
+      "loss": 1.25,
+      "step": 13903
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.180009499712184e-06,
+      "loss": 0.7656,
+      "step": 13904
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.17872756975948e-06,
+      "loss": 0.7109,
+      "step": 13905
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.177445784487957e-06,
+      "loss": 0.3145,
+      "step": 13906
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.176164143929463e-06,
+      "loss": 1.3203,
+      "step": 13907
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1748826481158625e-06,
+      "loss": 0.832,
+      "step": 13908
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.173601297078995e-06,
+      "loss": 1.0703,
+      "step": 13909
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.172320090850716e-06,
+      "loss": 0.6602,
+      "step": 13910
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.171039029462863e-06,
+      "loss": 0.4141,
+      "step": 13911
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.16975811294728e-06,
+      "loss": 0.4609,
+      "step": 13912
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.168477341335795e-06,
+      "loss": 1.3281,
+      "step": 13913
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1671967146602455e-06,
+      "loss": 0.8945,
+      "step": 13914
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.165916232952464e-06,
+      "loss": 1.0078,
+      "step": 13915
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.164635896244269e-06,
+      "loss": 0.5938,
+      "step": 13916
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1633557045674825e-06,
+      "loss": 0.5703,
+      "step": 13917
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.162075657953921e-06,
+      "loss": 0.7148,
+      "step": 13918
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.160795756435407e-06,
+      "loss": 0.6211,
+      "step": 13919
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.159516000043743e-06,
+      "loss": 1.4141,
+      "step": 13920
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.158236388810735e-06,
+      "loss": 0.7578,
+      "step": 13921
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.15695692276819e-06,
+      "loss": 0.6953,
+      "step": 13922
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.15567760194791e-06,
+      "loss": 0.3359,
+      "step": 13923
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.154398426381684e-06,
+      "loss": 0.7344,
+      "step": 13924
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.153119396101309e-06,
+      "loss": 0.2168,
+      "step": 13925
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.151840511138577e-06,
+      "loss": 0.4297,
+      "step": 13926
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.15056177152527e-06,
+      "loss": 0.459,
+      "step": 13927
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.149283177293165e-06,
+      "loss": 0.4102,
+      "step": 13928
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.148004728474045e-06,
+      "loss": 0.2773,
+      "step": 13929
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1467264250996865e-06,
+      "loss": 0.2988,
+      "step": 13930
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.145448267201855e-06,
+      "loss": 0.7305,
+      "step": 13931
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.144170254812322e-06,
+      "loss": 0.0967,
+      "step": 13932
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1428923879628505e-06,
+      "loss": 0.8945,
+      "step": 13933
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.141614666685195e-06,
+      "loss": 1.2188,
+      "step": 13934
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.140337091011116e-06,
+      "loss": 1.375,
+      "step": 13935
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.139059660972371e-06,
+      "loss": 0.5156,
+      "step": 13936
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.137782376600703e-06,
+      "loss": 1.0,
+      "step": 13937
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1365052379278555e-06,
+      "loss": 0.5078,
+      "step": 13938
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.135228244985573e-06,
+      "loss": 0.7656,
+      "step": 13939
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.133951397805599e-06,
+      "loss": 1.7344,
+      "step": 13940
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.132674696419658e-06,
+      "loss": 0.4512,
+      "step": 13941
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.131398140859488e-06,
+      "loss": 0.6289,
+      "step": 13942
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.130121731156819e-06,
+      "loss": 0.6367,
+      "step": 13943
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1288454673433686e-06,
+      "loss": 1.0391,
+      "step": 13944
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127569349450856e-06,
+      "loss": 0.7109,
+      "step": 13945
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.126293377511e-06,
+      "loss": 0.4863,
+      "step": 13946
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.125017551555519e-06,
+      "loss": 0.332,
+      "step": 13947
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.123741871616116e-06,
+      "loss": 0.6758,
+      "step": 13948
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.122466337724494e-06,
+      "loss": 1.3281,
+      "step": 13949
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.121190949912358e-06,
+      "loss": 1.6641,
+      "step": 13950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.119915708211412e-06,
+      "loss": 0.6992,
+      "step": 13951
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1186406126533404e-06,
+      "loss": 0.7891,
+      "step": 13952
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.117365663269845e-06,
+      "loss": 0.3906,
+      "step": 13953
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.116090860092603e-06,
+      "loss": 1.125,
+      "step": 13954
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.114816203153307e-06,
+      "loss": 0.4199,
+      "step": 13955
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.11354169248363e-06,
+      "loss": 0.3594,
+      "step": 13956
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1122673281152546e-06,
+      "loss": 0.6992,
+      "step": 13957
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.110993110079847e-06,
+      "loss": 1.1094,
+      "step": 13958
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.10971903840908e-06,
+      "loss": 0.6094,
+      "step": 13959
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.108445113134624e-06,
+      "loss": 0.4844,
+      "step": 13960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.107171334288136e-06,
+      "loss": 0.9492,
+      "step": 13961
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.105897701901271e-06,
+      "loss": 1.2344,
+      "step": 13962
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.104624216005688e-06,
+      "loss": 0.6875,
+      "step": 13963
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.103350876633041e-06,
+      "loss": 0.9609,
+      "step": 13964
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.102077683814973e-06,
+      "loss": 0.9492,
+      "step": 13965
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.100804637583126e-06,
+      "loss": 0.9688,
+      "step": 13966
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.099531737969143e-06,
+      "loss": 0.6797,
+      "step": 13967
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.098258985004663e-06,
+      "loss": 1.1484,
+      "step": 13968
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.096986378721312e-06,
+      "loss": 1.125,
+      "step": 13969
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.095713919150726e-06,
+      "loss": 0.7461,
+      "step": 13970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.094441606324529e-06,
+      "loss": 0.6602,
+      "step": 13971
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.093169440274343e-06,
+      "loss": 1.1016,
+      "step": 13972
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.091897421031782e-06,
+      "loss": 0.8867,
+      "step": 13973
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.090625548628463e-06,
+      "loss": 0.5664,
+      "step": 13974
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.089353823096e-06,
+      "loss": 0.459,
+      "step": 13975
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.088082244466e-06,
+      "loss": 0.6836,
+      "step": 13976
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.08681081277006e-06,
+      "loss": 0.3594,
+      "step": 13977
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.085539528039785e-06,
+      "loss": 0.2021,
+      "step": 13978
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.084268390306774e-06,
+      "loss": 0.4434,
+      "step": 13979
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.082997399602614e-06,
+      "loss": 0.5859,
+      "step": 13980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.081726555958899e-06,
+      "loss": 0.918,
+      "step": 13981
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.08045585940721e-06,
+      "loss": 0.3398,
+      "step": 13982
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.0791853099791335e-06,
+      "loss": 0.957,
+      "step": 13983
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.077914907706241e-06,
+      "loss": 0.103,
+      "step": 13984
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.076644652620115e-06,
+      "loss": 0.9375,
+      "step": 13985
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.0753745447523185e-06,
+      "loss": 1.0938,
+      "step": 13986
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.074104584134425e-06,
+      "loss": 0.1348,
+      "step": 13987
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.072834770797992e-06,
+      "loss": 0.8398,
+      "step": 13988
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.071565104774586e-06,
+      "loss": 0.4883,
+      "step": 13989
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.070295586095757e-06,
+      "loss": 0.5,
+      "step": 13990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.069026214793059e-06,
+      "loss": 0.4199,
+      "step": 13991
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.067756990898045e-06,
+      "loss": 1.1875,
+      "step": 13992
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.066487914442258e-06,
+      "loss": 1.1016,
+      "step": 13993
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.0652189854572345e-06,
+      "loss": 1.3125,
+      "step": 13994
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.063950203974515e-06,
+      "loss": 0.4004,
+      "step": 13995
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.06268157002564e-06,
+      "loss": 1.0391,
+      "step": 13996
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.0614130836421305e-06,
+      "loss": 0.5391,
+      "step": 13997
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.060144744855518e-06,
+      "loss": 1.1562,
+      "step": 13998
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.058876553697328e-06,
+      "loss": 0.9258,
+      "step": 13999
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.057608510199077e-06,
+      "loss": 0.5195,
+      "step": 14000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.056340614392278e-06,
+      "loss": 0.8633,
+      "step": 14001
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.055072866308446e-06,
+      "loss": 0.7305,
+      "step": 14002
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.053805265979094e-06,
+      "loss": 0.6484,
+      "step": 14003
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.052537813435721e-06,
+      "loss": 0.7266,
+      "step": 14004
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.051270508709827e-06,
+      "loss": 0.8047,
+      "step": 14005
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.050003351832915e-06,
+      "loss": 0.7383,
+      "step": 14006
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.048736342836471e-06,
+      "loss": 1.0078,
+      "step": 14007
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.047469481751991e-06,
+      "loss": 0.4863,
+      "step": 14008
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.046202768610964e-06,
+      "loss": 0.6797,
+      "step": 14009
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.044936203444868e-06,
+      "loss": 0.9805,
+      "step": 14010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.04366978628518e-06,
+      "loss": 0.6719,
+      "step": 14011
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.042403517163378e-06,
+      "loss": 0.7734,
+      "step": 14012
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.041137396110937e-06,
+      "loss": 0.6055,
+      "step": 14013
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.039871423159323e-06,
+      "loss": 0.7109,
+      "step": 14014
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.038605598339996e-06,
+      "loss": 1.1484,
+      "step": 14015
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.037339921684419e-06,
+      "loss": 0.3809,
+      "step": 14016
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.036074393224053e-06,
+      "loss": 0.8438,
+      "step": 14017
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.034809012990346e-06,
+      "loss": 1.25,
+      "step": 14018
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.033543781014749e-06,
+      "loss": 0.8203,
+      "step": 14019
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.032278697328712e-06,
+      "loss": 0.5625,
+      "step": 14020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.031013761963675e-06,
+      "loss": 1.1406,
+      "step": 14021
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.029748974951072e-06,
+      "loss": 1.0547,
+      "step": 14022
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.0284843363223405e-06,
+      "loss": 1.1719,
+      "step": 14023
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.027219846108916e-06,
+      "loss": 0.4766,
+      "step": 14024
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.025955504342224e-06,
+      "loss": 1.2422,
+      "step": 14025
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.0246913110536825e-06,
+      "loss": 1.2656,
+      "step": 14026
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.0234272662747156e-06,
+      "loss": 0.5117,
+      "step": 14027
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.0221633700367436e-06,
+      "loss": 0.9609,
+      "step": 14028
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.020899622371173e-06,
+      "loss": 0.1523,
+      "step": 14029
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.019636023309418e-06,
+      "loss": 1.4609,
+      "step": 14030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.018372572882878e-06,
+      "loss": 0.4473,
+      "step": 14031
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.017109271122962e-06,
+      "loss": 0.2031,
+      "step": 14032
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.0158461180610604e-06,
+      "loss": 0.3535,
+      "step": 14033
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.014583113728575e-06,
+      "loss": 0.4883,
+      "step": 14034
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.013320258156889e-06,
+      "loss": 1.3828,
+      "step": 14035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.012057551377392e-06,
+      "loss": 0.2773,
+      "step": 14036
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.010794993421471e-06,
+      "loss": 0.2891,
+      "step": 14037
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.009532584320503e-06,
+      "loss": 0.1602,
+      "step": 14038
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.008270324105859e-06,
+      "loss": 1.4375,
+      "step": 14039
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.007008212808916e-06,
+      "loss": 1.0781,
+      "step": 14040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.005746250461044e-06,
+      "loss": 0.4512,
+      "step": 14041
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.004484437093605e-06,
+      "loss": 0.9141,
+      "step": 14042
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.003222772737957e-06,
+      "loss": 1.6328,
+      "step": 14043
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.001961257425461e-06,
+      "loss": 0.4473,
+      "step": 14044
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.000699891187473e-06,
+      "loss": 0.875,
+      "step": 14045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.999438674055335e-06,
+      "loss": 1.2812,
+      "step": 14046
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.998177606060398e-06,
+      "loss": 1.1172,
+      "step": 14047
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.99691668723401e-06,
+      "loss": 0.7969,
+      "step": 14048
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.995655917607502e-06,
+      "loss": 0.498,
+      "step": 14049
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.994395297212208e-06,
+      "loss": 0.7617,
+      "step": 14050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9931348260794625e-06,
+      "loss": 0.9062,
+      "step": 14051
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.991874504240595e-06,
+      "loss": 1.0547,
+      "step": 14052
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.990614331726929e-06,
+      "loss": 0.5703,
+      "step": 14053
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.989354308569778e-06,
+      "loss": 0.5898,
+      "step": 14054
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.988094434800465e-06,
+      "loss": 0.7773,
+      "step": 14055
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9868347104503035e-06,
+      "loss": 0.7109,
+      "step": 14056
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.9855751355505965e-06,
+      "loss": 0.5625,
+      "step": 14057
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.984315710132657e-06,
+      "loss": 0.7109,
+      "step": 14058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.983056434227779e-06,
+      "loss": 0.8633,
+      "step": 14059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.981797307867267e-06,
+      "loss": 1.1172,
+      "step": 14060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.980538331082408e-06,
+      "loss": 0.7578,
+      "step": 14061
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.979279503904502e-06,
+      "loss": 0.4844,
+      "step": 14062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.978020826364825e-06,
+      "loss": 1.3984,
+      "step": 14063
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.976762298494667e-06,
+      "loss": 0.9805,
+      "step": 14064
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.975503920325309e-06,
+      "loss": 1.4609,
+      "step": 14065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.974245691888022e-06,
+      "loss": 1.0078,
+      "step": 14066
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.972987613214076e-06,
+      "loss": 1.0234,
+      "step": 14067
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.971729684334743e-06,
+      "loss": 0.8828,
+      "step": 14068
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.970471905281289e-06,
+      "loss": 1.4844,
+      "step": 14069
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.969214276084972e-06,
+      "loss": 1.1953,
+      "step": 14070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.967956796777047e-06,
+      "loss": 0.6914,
+      "step": 14071
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.966699467388769e-06,
+      "loss": 0.8438,
+      "step": 14072
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.965442287951391e-06,
+      "loss": 1.4922,
+      "step": 14073
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.964185258496153e-06,
+      "loss": 1.2031,
+      "step": 14074
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.962928379054299e-06,
+      "loss": 1.7031,
+      "step": 14075
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.961671649657072e-06,
+      "loss": 0.7461,
+      "step": 14076
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.960415070335702e-06,
+      "loss": 0.6016,
+      "step": 14077
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.959158641121419e-06,
+      "loss": 0.9219,
+      "step": 14078
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.957902362045451e-06,
+      "loss": 0.2393,
+      "step": 14079
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.956646233139024e-06,
+      "loss": 0.6211,
+      "step": 14080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.955390254433358e-06,
+      "loss": 1.4531,
+      "step": 14081
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.954134425959662e-06,
+      "loss": 0.6602,
+      "step": 14082
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.952878747749156e-06,
+      "loss": 0.7617,
+      "step": 14083
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.9516232198330426e-06,
+      "loss": 1.4688,
+      "step": 14084
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.95036784224253e-06,
+      "loss": 0.1895,
+      "step": 14085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.949112615008821e-06,
+      "loss": 0.4824,
+      "step": 14086
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.9478575381631104e-06,
+      "loss": 0.1982,
+      "step": 14087
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.946602611736588e-06,
+      "loss": 1.1094,
+      "step": 14088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.945347835760448e-06,
+      "loss": 0.6133,
+      "step": 14089
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.944093210265879e-06,
+      "loss": 0.2734,
+      "step": 14090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.94283873528406e-06,
+      "loss": 0.543,
+      "step": 14091
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.941584410846166e-06,
+      "loss": 0.7852,
+      "step": 14092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.940330236983377e-06,
+      "loss": 1.2109,
+      "step": 14093
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.9390762137268645e-06,
+      "loss": 1.0078,
+      "step": 14094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.937822341107791e-06,
+      "loss": 1.0312,
+      "step": 14095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.9365686191573215e-06,
+      "loss": 0.8711,
+      "step": 14096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.935315047906623e-06,
+      "loss": 1.2344,
+      "step": 14097
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.934061627386845e-06,
+      "loss": 1.4453,
+      "step": 14098
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.932808357629137e-06,
+      "loss": 0.918,
+      "step": 14099
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.931555238664651e-06,
+      "loss": 1.6016,
+      "step": 14100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.9303022705245364e-06,
+      "loss": 1.4609,
+      "step": 14101
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.929049453239926e-06,
+      "loss": 0.9258,
+      "step": 14102
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.927796786841965e-06,
+      "loss": 0.9023,
+      "step": 14103
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.926544271361781e-06,
+      "loss": 0.6602,
+      "step": 14104
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.925291906830509e-06,
+      "loss": 0.9688,
+      "step": 14105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.9240396932792675e-06,
+      "loss": 0.4922,
+      "step": 14106
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.922787630739189e-06,
+      "loss": 0.6484,
+      "step": 14107
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.921535719241382e-06,
+      "loss": 0.4668,
+      "step": 14108
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.920283958816969e-06,
+      "loss": 1.1484,
+      "step": 14109
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.919032349497056e-06,
+      "loss": 0.8086,
+      "step": 14110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.917780891312757e-06,
+      "loss": 0.4609,
+      "step": 14111
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.916529584295166e-06,
+      "loss": 1.3828,
+      "step": 14112
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.9152784284753885e-06,
+      "loss": 1.1328,
+      "step": 14113
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.914027423884524e-06,
+      "loss": 1.0312,
+      "step": 14114
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.91277657055366e-06,
+      "loss": 1.5391,
+      "step": 14115
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.911525868513882e-06,
+      "loss": 1.2656,
+      "step": 14116
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.91027531779628e-06,
+      "loss": 0.5234,
+      "step": 14117
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.909024918431937e-06,
+      "loss": 0.7812,
+      "step": 14118
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.907774670451927e-06,
+      "loss": 0.832,
+      "step": 14119
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.906524573887319e-06,
+      "loss": 1.0547,
+      "step": 14120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.905274628769188e-06,
+      "loss": 0.8711,
+      "step": 14121
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.904024835128602e-06,
+      "loss": 1.4766,
+      "step": 14122
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.902775192996618e-06,
+      "loss": 0.2949,
+      "step": 14123
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.901525702404297e-06,
+      "loss": 0.5273,
+      "step": 14124
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.900276363382697e-06,
+      "loss": 0.5898,
+      "step": 14125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.899027175962865e-06,
+      "loss": 0.8867,
+      "step": 14126
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.897778140175845e-06,
+      "loss": 0.2852,
+      "step": 14127
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.896529256052685e-06,
+      "loss": 0.3223,
+      "step": 14128
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.895280523624426e-06,
+      "loss": 0.9297,
+      "step": 14129
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.894031942922098e-06,
+      "loss": 0.6836,
+      "step": 14130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.89278351397674e-06,
+      "loss": 0.8516,
+      "step": 14131
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.891535236819373e-06,
+      "loss": 0.2695,
+      "step": 14132
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.89028711148103e-06,
+      "loss": 0.332,
+      "step": 14133
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.889039137992724e-06,
+      "loss": 0.6914,
+      "step": 14134
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.887791316385478e-06,
+      "loss": 0.957,
+      "step": 14135
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.886543646690298e-06,
+      "loss": 0.6523,
+      "step": 14136
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.885296128938202e-06,
+      "loss": 1.3125,
+      "step": 14137
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.884048763160188e-06,
+      "loss": 0.5547,
+      "step": 14138
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.882801549387264e-06,
+      "loss": 0.2061,
+      "step": 14139
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.881554487650422e-06,
+      "loss": 0.3418,
+      "step": 14140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.88030757798066e-06,
+      "loss": 1.5156,
+      "step": 14141
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.879060820408972e-06,
+      "loss": 0.6211,
+      "step": 14142
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.87781421496634e-06,
+      "loss": 0.7461,
+      "step": 14143
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.876567761683745e-06,
+      "loss": 1.7266,
+      "step": 14144
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.8753214605921695e-06,
+      "loss": 0.375,
+      "step": 14145
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.874075311722592e-06,
+      "loss": 0.5195,
+      "step": 14146
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.872829315105979e-06,
+      "loss": 0.9883,
+      "step": 14147
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.871583470773298e-06,
+      "loss": 0.7148,
+      "step": 14148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.870337778755515e-06,
+      "loss": 0.5352,
+      "step": 14149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.869092239083594e-06,
+      "loss": 0.6602,
+      "step": 14150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.867846851788484e-06,
+      "loss": 0.7109,
+      "step": 14151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.866601616901141e-06,
+      "loss": 0.9258,
+      "step": 14152
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8653565344525176e-06,
+      "loss": 0.4824,
+      "step": 14153
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.864111604473557e-06,
+      "loss": 0.6719,
+      "step": 14154
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8628668269951956e-06,
+      "loss": 0.5742,
+      "step": 14155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.861622202048378e-06,
+      "loss": 0.8555,
+      "step": 14156
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.860377729664031e-06,
+      "loss": 0.9219,
+      "step": 14157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8591334098730915e-06,
+      "loss": 1.2734,
+      "step": 14158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8578892427064785e-06,
+      "loss": 0.9805,
+      "step": 14159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8566452281951215e-06,
+      "loss": 0.5703,
+      "step": 14160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.855401366369933e-06,
+      "loss": 0.9531,
+      "step": 14161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.854157657261831e-06,
+      "loss": 0.9297,
+      "step": 14162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.852914100901729e-06,
+      "loss": 1.2891,
+      "step": 14163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.851670697320531e-06,
+      "loss": 0.3066,
+      "step": 14164
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.850427446549138e-06,
+      "loss": 0.7891,
+      "step": 14165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.849184348618451e-06,
+      "loss": 1.3047,
+      "step": 14166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.847941403559371e-06,
+      "loss": 0.8086,
+      "step": 14167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.846698611402783e-06,
+      "loss": 0.7617,
+      "step": 14168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.845455972179577e-06,
+      "loss": 0.8477,
+      "step": 14169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.844213485920643e-06,
+      "loss": 0.7969,
+      "step": 14170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.842971152656857e-06,
+      "loss": 1.125,
+      "step": 14171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.841728972419093e-06,
+      "loss": 1.1719,
+      "step": 14172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.840486945238227e-06,
+      "loss": 0.7734,
+      "step": 14173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.839245071145132e-06,
+      "loss": 0.3164,
+      "step": 14174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.838003350170668e-06,
+      "loss": 0.3848,
+      "step": 14175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.836761782345696e-06,
+      "loss": 1.2734,
+      "step": 14176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8355203677010754e-06,
+      "loss": 1.1953,
+      "step": 14177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.834279106267665e-06,
+      "loss": 1.5234,
+      "step": 14178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.833037998076306e-06,
+      "loss": 0.2871,
+      "step": 14179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.831797043157855e-06,
+      "loss": 0.1895,
+      "step": 14180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.830556241543143e-06,
+      "loss": 0.085,
+      "step": 14181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.82931559326302e-06,
+      "loss": 0.7773,
+      "step": 14182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.828075098348311e-06,
+      "loss": 0.7773,
+      "step": 14183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.826834756829856e-06,
+      "loss": 0.7812,
+      "step": 14184
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.825594568738473e-06,
+      "loss": 0.8945,
+      "step": 14185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.824354534104996e-06,
+      "loss": 0.2793,
+      "step": 14186
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.823114652960235e-06,
+      "loss": 0.9219,
+      "step": 14187
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.821874925335013e-06,
+      "loss": 0.1289,
+      "step": 14188
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.820635351260134e-06,
+      "loss": 1.1094,
+      "step": 14189
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8193959307664145e-06,
+      "loss": 1.1719,
+      "step": 14190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.818156663884657e-06,
+      "loss": 1.2422,
+      "step": 14191
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.816917550645661e-06,
+      "loss": 1.3672,
+      "step": 14192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.81567859108022e-06,
+      "loss": 1.1328,
+      "step": 14193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8144397852191296e-06,
+      "loss": 1.1562,
+      "step": 14194
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.813201133093183e-06,
+      "loss": 0.9219,
+      "step": 14195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.811962634733162e-06,
+      "loss": 0.5781,
+      "step": 14196
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8107242901698403e-06,
+      "loss": 0.6797,
+      "step": 14197
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.809486099434011e-06,
+      "loss": 0.5664,
+      "step": 14198
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8082480625564387e-06,
+      "loss": 1.8672,
+      "step": 14199
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8070101795678927e-06,
+      "loss": 0.4629,
+      "step": 14200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.80577245049914e-06,
+      "loss": 1.6641,
+      "step": 14201
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8045348753809486e-06,
+      "loss": 0.8789,
+      "step": 14202
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8032974542440725e-06,
+      "loss": 0.5195,
+      "step": 14203
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.802060187119263e-06,
+      "loss": 0.8242,
+      "step": 14204
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.8008230740372743e-06,
+      "loss": 1.0156,
+      "step": 14205
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7995861150288573e-06,
+      "loss": 0.5469,
+      "step": 14206
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.798349310124748e-06,
+      "loss": 1.2188,
+      "step": 14207
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7971126593556927e-06,
+      "loss": 0.6445,
+      "step": 14208
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.795876162752421e-06,
+      "loss": 0.7734,
+      "step": 14209
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7946398203456704e-06,
+      "loss": 0.4316,
+      "step": 14210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.793403632166163e-06,
+      "loss": 1.6328,
+      "step": 14211
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.792167598244628e-06,
+      "loss": 1.1641,
+      "step": 14212
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.79093171861178e-06,
+      "loss": 0.9805,
+      "step": 14213
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7896959932983425e-06,
+      "loss": 1.2656,
+      "step": 14214
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7884604223350198e-06,
+      "loss": 1.6484,
+      "step": 14215
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7872250057525294e-06,
+      "loss": 0.6133,
+      "step": 14216
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7859897435815686e-06,
+      "loss": 0.6641,
+      "step": 14217
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7847546358528407e-06,
+      "loss": 0.9922,
+      "step": 14218
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7835196825970476e-06,
+      "loss": 0.3809,
+      "step": 14219
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.782284883844879e-06,
+      "loss": 0.7812,
+      "step": 14220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.781050239627021e-06,
+      "loss": 0.6953,
+      "step": 14221
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.779815749974163e-06,
+      "loss": 0.9844,
+      "step": 14222
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.778581414916989e-06,
+      "loss": 0.6641,
+      "step": 14223
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.777347234486175e-06,
+      "loss": 1.0859,
+      "step": 14224
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7761132087123918e-06,
+      "loss": 0.3828,
+      "step": 14225
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7748793376263114e-06,
+      "loss": 0.4414,
+      "step": 14226
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.773645621258606e-06,
+      "loss": 0.6172,
+      "step": 14227
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7724120596399303e-06,
+      "loss": 1.1797,
+      "step": 14228
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7711786528009498e-06,
+      "loss": 1.0078,
+      "step": 14229
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7699454007723124e-06,
+      "loss": 0.9102,
+      "step": 14230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.768712303584676e-06,
+      "loss": 1.4062,
+      "step": 14231
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.767479361268682e-06,
+      "loss": 0.1514,
+      "step": 14232
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.76624657385498e-06,
+      "loss": 0.2715,
+      "step": 14233
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7650139413742027e-06,
+      "loss": 0.957,
+      "step": 14234
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.763781463856988e-06,
+      "loss": 0.1826,
+      "step": 14235
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7625491413339743e-06,
+      "loss": 1.3438,
+      "step": 14236
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7613169738357834e-06,
+      "loss": 1.1875,
+      "step": 14237
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7600849613930366e-06,
+      "loss": 1.0469,
+      "step": 14238
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7588531040363575e-06,
+      "loss": 0.9844,
+      "step": 14239
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7576214017963675e-06,
+      "loss": 0.1611,
+      "step": 14240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.756389854703674e-06,
+      "loss": 0.7461,
+      "step": 14241
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.755158462788883e-06,
+      "loss": 0.4922,
+      "step": 14242
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7539272260826023e-06,
+      "loss": 1.25,
+      "step": 14243
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7526961446154364e-06,
+      "loss": 0.4336,
+      "step": 14244
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7514652184179767e-06,
+      "loss": 0.4883,
+      "step": 14245
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7502344475208183e-06,
+      "loss": 0.8867,
+      "step": 14246
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7490038319545554e-06,
+      "loss": 0.4863,
+      "step": 14247
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.74777337174977e-06,
+      "loss": 1.0703,
+      "step": 14248
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7465430669370384e-06,
+      "loss": 0.7695,
+      "step": 14249
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.745312917546945e-06,
+      "loss": 0.875,
+      "step": 14250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.744082923610065e-06,
+      "loss": 0.5352,
+      "step": 14251
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7428530851569655e-06,
+      "loss": 0.5898,
+      "step": 14252
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7416234022182095e-06,
+      "loss": 0.8945,
+      "step": 14253
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7403938748243628e-06,
+      "loss": 0.9531,
+      "step": 14254
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7391645030059887e-06,
+      "loss": 0.7578,
+      "step": 14255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7379352867936323e-06,
+      "loss": 0.4434,
+      "step": 14256
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.736706226217853e-06,
+      "loss": 1.6328,
+      "step": 14257
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7354773213091897e-06,
+      "loss": 0.6016,
+      "step": 14258
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7342485720981946e-06,
+      "loss": 0.8906,
+      "step": 14259
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7330199786153977e-06,
+      "loss": 0.8984,
+      "step": 14260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.731791540891342e-06,
+      "loss": 0.4219,
+      "step": 14261
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.730563258956552e-06,
+      "loss": 1.2109,
+      "step": 14262
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.729335132841563e-06,
+      "loss": 0.8359,
+      "step": 14263
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7281071625768907e-06,
+      "loss": 1.1094,
+      "step": 14264
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.726879348193061e-06,
+      "loss": 1.0234,
+      "step": 14265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.725651689720585e-06,
+      "loss": 0.8398,
+      "step": 14266
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7244241871899777e-06,
+      "loss": 1.0859,
+      "step": 14267
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.72319684063175e-06,
+      "loss": 0.8398,
+      "step": 14268
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7219696500764023e-06,
+      "loss": 0.7461,
+      "step": 14269
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7207426155544334e-06,
+      "loss": 1.1953,
+      "step": 14270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719515737096342e-06,
+      "loss": 1.0234,
+      "step": 14271
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7182890147326244e-06,
+      "loss": 0.8711,
+      "step": 14272
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7170624484937623e-06,
+      "loss": 1.6016,
+      "step": 14273
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7158360384102453e-06,
+      "loss": 1.6172,
+      "step": 14274
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.714609784512556e-06,
+      "loss": 0.8711,
+      "step": 14275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.71338368683117e-06,
+      "loss": 0.1196,
+      "step": 14276
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7121577453965564e-06,
+      "loss": 1.7578,
+      "step": 14277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.710931960239187e-06,
+      "loss": 0.8359,
+      "step": 14278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7097063313895323e-06,
+      "loss": 1.1406,
+      "step": 14279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7084808588780496e-06,
+      "loss": 0.6875,
+      "step": 14280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.707255542735194e-06,
+      "loss": 0.6836,
+      "step": 14281
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.706030382991421e-06,
+      "loss": 1.0938,
+      "step": 14282
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7048053796771865e-06,
+      "loss": 0.5273,
+      "step": 14283
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7035805328229278e-06,
+      "loss": 0.6133,
+      "step": 14284
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7023558424590944e-06,
+      "loss": 0.373,
+      "step": 14285
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7011313086161183e-06,
+      "loss": 0.5195,
+      "step": 14286
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6999069313244405e-06,
+      "loss": 0.2031,
+      "step": 14287
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.698682710614484e-06,
+      "loss": 0.7695,
+      "step": 14288
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.697458646516684e-06,
+      "loss": 0.9492,
+      "step": 14289
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6962347390614537e-06,
+      "loss": 1.4062,
+      "step": 14290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6950109882792217e-06,
+      "loss": 1.7656,
+      "step": 14291
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6937873942003944e-06,
+      "loss": 1.4766,
+      "step": 14292
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6925639568553895e-06,
+      "loss": 1.0156,
+      "step": 14293
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6913406762746084e-06,
+      "loss": 1.1172,
+      "step": 14294
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.690117552488457e-06,
+      "loss": 1.5703,
+      "step": 14295
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6888945855273383e-06,
+      "loss": 1.1406,
+      "step": 14296
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.687671775421644e-06,
+      "loss": 0.7461,
+      "step": 14297
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.686449122201763e-06,
+      "loss": 1.5703,
+      "step": 14298
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6852266258980864e-06,
+      "loss": 1.1172,
+      "step": 14299
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.684004286541e-06,
+      "loss": 0.5781,
+      "step": 14300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.682782104160879e-06,
+      "loss": 0.3066,
+      "step": 14301
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6815600787881e-06,
+      "loss": 0.5078,
+      "step": 14302
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6803382104530416e-06,
+      "loss": 0.5859,
+      "step": 14303
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6791164991860663e-06,
+      "loss": 0.6406,
+      "step": 14304
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6778949450175363e-06,
+      "loss": 0.6797,
+      "step": 14305
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.676673547977818e-06,
+      "loss": 0.3379,
+      "step": 14306
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6754523080972606e-06,
+      "loss": 1.2812,
+      "step": 14307
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.674231225406225e-06,
+      "loss": 1.6562,
+      "step": 14308
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6730102999350515e-06,
+      "loss": 0.291,
+      "step": 14309
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6717895317140915e-06,
+      "loss": 0.3984,
+      "step": 14310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6705689207736807e-06,
+      "loss": 0.4844,
+      "step": 14311
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.669348467144158e-06,
+      "loss": 0.8047,
+      "step": 14312
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6681281708558603e-06,
+      "loss": 0.3828,
+      "step": 14313
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.666908031939114e-06,
+      "loss": 0.9414,
+      "step": 14314
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.66568805042424e-06,
+      "loss": 1.2266,
+      "step": 14315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.664468226341562e-06,
+      "loss": 1.0312,
+      "step": 14316
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6632485597214027e-06,
+      "loss": 1.125,
+      "step": 14317
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6620290505940703e-06,
+      "loss": 0.7812,
+      "step": 14318
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.660809698989873e-06,
+      "loss": 0.5625,
+      "step": 14319
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.659590504939119e-06,
+      "loss": 0.707,
+      "step": 14320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.658371468472112e-06,
+      "loss": 0.7344,
+      "step": 14321
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6571525896191453e-06,
+      "loss": 1.3516,
+      "step": 14322
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6559338684105137e-06,
+      "loss": 0.5938,
+      "step": 14323
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6547153048765118e-06,
+      "loss": 1.125,
+      "step": 14324
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.653496899047423e-06,
+      "loss": 0.3672,
+      "step": 14325
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6522786509535245e-06,
+      "loss": 1.5781,
+      "step": 14326
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6510605606250995e-06,
+      "loss": 0.7969,
+      "step": 14327
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6498426280924247e-06,
+      "loss": 1.0469,
+      "step": 14328
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.648624853385767e-06,
+      "loss": 1.125,
+      "step": 14329
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6474072365353896e-06,
+      "loss": 0.373,
+      "step": 14330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6461897775715584e-06,
+      "loss": 0.1709,
+      "step": 14331
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6449724765245344e-06,
+      "loss": 0.9805,
+      "step": 14332
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6437553334245666e-06,
+      "loss": 1.0781,
+      "step": 14333
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.642538348301914e-06,
+      "loss": 0.2363,
+      "step": 14334
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6413215211868125e-06,
+      "loss": 0.3574,
+      "step": 14335
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.640104852109516e-06,
+      "loss": 0.8672,
+      "step": 14336
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.638888341100254e-06,
+      "loss": 0.4863,
+      "step": 14337
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6376719881892686e-06,
+      "loss": 0.1641,
+      "step": 14338
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6364557934067858e-06,
+      "loss": 0.8359,
+      "step": 14339
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6352397567830343e-06,
+      "loss": 1.1953,
+      "step": 14340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6340238783482418e-06,
+      "loss": 1.6641,
+      "step": 14341
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6328081581326224e-06,
+      "loss": 1.625,
+      "step": 14342
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.631592596166391e-06,
+      "loss": 0.8359,
+      "step": 14343
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6303771924797606e-06,
+      "loss": 0.8008,
+      "step": 14344
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6291619471029417e-06,
+      "loss": 1.0469,
+      "step": 14345
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.627946860066135e-06,
+      "loss": 1.2891,
+      "step": 14346
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6267319313995376e-06,
+      "loss": 0.8359,
+      "step": 14347
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6255171611333473e-06,
+      "loss": 0.9648,
+      "step": 14348
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.624302549297759e-06,
+      "loss": 1.9141,
+      "step": 14349
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6230880959229553e-06,
+      "loss": 1.25,
+      "step": 14350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.621873801039123e-06,
+      "loss": 1.8125,
+      "step": 14351
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.620659664676445e-06,
+      "loss": 0.3301,
+      "step": 14352
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.619445686865093e-06,
+      "loss": 1.0078,
+      "step": 14353
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.618231867635238e-06,
+      "loss": 0.3867,
+      "step": 14354
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.617018207017049e-06,
+      "loss": 0.5898,
+      "step": 14355
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.615804705040695e-06,
+      "loss": 1.1875,
+      "step": 14356
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6145913617363325e-06,
+      "loss": 1.1797,
+      "step": 14357
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.613378177134115e-06,
+      "loss": 0.5547,
+      "step": 14358
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6121651512641966e-06,
+      "loss": 0.4727,
+      "step": 14359
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6109522841567303e-06,
+      "loss": 0.9141,
+      "step": 14360
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.609739575841854e-06,
+      "loss": 0.6055,
+      "step": 14361
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.608527026349714e-06,
+      "loss": 0.7539,
+      "step": 14362
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6073146357104417e-06,
+      "loss": 1.7422,
+      "step": 14363
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6061024039541746e-06,
+      "loss": 1.1875,
+      "step": 14364
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.604890331111036e-06,
+      "loss": 1.3984,
+      "step": 14365
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6036784172111563e-06,
+      "loss": 0.6875,
+      "step": 14366
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6024666622846495e-06,
+      "loss": 0.5586,
+      "step": 14367
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6012550663616407e-06,
+      "loss": 1.7266,
+      "step": 14368
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.600043629472234e-06,
+      "loss": 0.8281,
+      "step": 14369
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.598832351646546e-06,
+      "loss": 0.7344,
+      "step": 14370
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5976212329146766e-06,
+      "loss": 0.6562,
+      "step": 14371
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.596410273306726e-06,
+      "loss": 1.3984,
+      "step": 14372
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5951994728527984e-06,
+      "loss": 1.3359,
+      "step": 14373
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5939888315829828e-06,
+      "loss": 0.875,
+      "step": 14374
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.592778349527364e-06,
+      "loss": 0.832,
+      "step": 14375
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5915680267160303e-06,
+      "loss": 0.4863,
+      "step": 14376
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.590357863179067e-06,
+      "loss": 0.6289,
+      "step": 14377
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5891478589465435e-06,
+      "loss": 0.5039,
+      "step": 14378
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5879380140485422e-06,
+      "loss": 1.4766,
+      "step": 14379
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5867283285151233e-06,
+      "loss": 0.375,
+      "step": 14380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.585518802376361e-06,
+      "loss": 0.291,
+      "step": 14381
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5843094356623088e-06,
+      "loss": 0.4609,
+      "step": 14382
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.58310022840303e-06,
+      "loss": 0.3418,
+      "step": 14383
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5818911806285727e-06,
+      "loss": 1.0078,
+      "step": 14384
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5806822923689934e-06,
+      "loss": 0.5234,
+      "step": 14385
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5794735636543297e-06,
+      "loss": 0.8398,
+      "step": 14386
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.578264994514631e-06,
+      "loss": 1.125,
+      "step": 14387
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5770565849799266e-06,
+      "loss": 0.4531,
+      "step": 14388
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5758483350802553e-06,
+      "loss": 0.5508,
+      "step": 14389
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5746402448456497e-06,
+      "loss": 0.9883,
+      "step": 14390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.573432314306132e-06,
+      "loss": 1.2969,
+      "step": 14391
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.57222454349172e-06,
+      "loss": 1.8281,
+      "step": 14392
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5710169324324364e-06,
+      "loss": 1.8281,
+      "step": 14393
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5698094811582974e-06,
+      "loss": 0.4648,
+      "step": 14394
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5686021896993085e-06,
+      "loss": 0.4336,
+      "step": 14395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5673950580854732e-06,
+      "loss": 0.5508,
+      "step": 14396
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.566188086346797e-06,
+      "loss": 0.8516,
+      "step": 14397
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.564981274513282e-06,
+      "loss": 1.125,
+      "step": 14398
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5637746226149126e-06,
+      "loss": 0.7109,
+      "step": 14399
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.562568130681685e-06,
+      "loss": 0.6016,
+      "step": 14400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5613617987435865e-06,
+      "loss": 0.4805,
+      "step": 14401
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.560155626830597e-06,
+      "loss": 0.6289,
+      "step": 14402
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.558949614972691e-06,
+      "loss": 0.2871,
+      "step": 14403
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5577437631998467e-06,
+      "loss": 0.832,
+      "step": 14404
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.556538071542036e-06,
+      "loss": 0.3262,
+      "step": 14405
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5553325400292184e-06,
+      "loss": 0.4902,
+      "step": 14406
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.554127168691364e-06,
+      "loss": 0.7891,
+      "step": 14407
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.552921957558423e-06,
+      "loss": 1.1875,
+      "step": 14408
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5517169066603574e-06,
+      "loss": 0.7148,
+      "step": 14409
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5505120160271102e-06,
+      "loss": 0.8359,
+      "step": 14410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5493072856886337e-06,
+      "loss": 0.4414,
+      "step": 14411
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.548102715674864e-06,
+      "loss": 0.9805,
+      "step": 14412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5468983060157458e-06,
+      "loss": 0.8242,
+      "step": 14413
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.545694056741207e-06,
+      "loss": 0.7695,
+      "step": 14414
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.544489967881184e-06,
+      "loss": 0.7227,
+      "step": 14415
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5432860394655965e-06,
+      "loss": 0.9844,
+      "step": 14416
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5420822715243707e-06,
+      "loss": 1.3828,
+      "step": 14417
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.540878664087428e-06,
+      "loss": 1.5312,
+      "step": 14418
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5396752171846794e-06,
+      "loss": 0.6562,
+      "step": 14419
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5384719308460314e-06,
+      "loss": 0.8477,
+      "step": 14420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5372688051013947e-06,
+      "loss": 1.0938,
+      "step": 14421
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5360658399806733e-06,
+      "loss": 1.1016,
+      "step": 14422
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.534863035513765e-06,
+      "loss": 1.1406,
+      "step": 14423
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5336603917305567e-06,
+      "loss": 0.2656,
+      "step": 14424
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5324579086609456e-06,
+      "loss": 1.4531,
+      "step": 14425
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5312555863348207e-06,
+      "loss": 1.5312,
+      "step": 14426
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.530053424782056e-06,
+      "loss": 0.8828,
+      "step": 14427
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5288514240325356e-06,
+      "loss": 0.7852,
+      "step": 14428
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.527649584116135e-06,
+      "loss": 0.4961,
+      "step": 14429
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.526447905062723e-06,
+      "loss": 0.5039,
+      "step": 14430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5252463869021613e-06,
+      "loss": 0.9453,
+      "step": 14431
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5240450296643167e-06,
+      "loss": 1.0703,
+      "step": 14432
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5228438333790503e-06,
+      "loss": 0.7305,
+      "step": 14433
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5216427980762134e-06,
+      "loss": 0.2002,
+      "step": 14434
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.520441923785652e-06,
+      "loss": 0.5469,
+      "step": 14435
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5192412105372174e-06,
+      "loss": 0.3887,
+      "step": 14436
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.518040658360755e-06,
+      "loss": 0.707,
+      "step": 14437
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.516840267286097e-06,
+      "loss": 0.1152,
+      "step": 14438
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.515640037343082e-06,
+      "loss": 0.248,
+      "step": 14439
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.514439968561536e-06,
+      "loss": 0.5352,
+      "step": 14440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.513240060971291e-06,
+      "loss": 0.9531,
+      "step": 14441
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5120403146021633e-06,
+      "loss": 0.1963,
+      "step": 14442
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5108407294839774e-06,
+      "loss": 0.4863,
+      "step": 14443
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5096413056465416e-06,
+      "loss": 0.7891,
+      "step": 14444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5084420431196687e-06,
+      "loss": 0.2617,
+      "step": 14445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.507242941933169e-06,
+      "loss": 0.6445,
+      "step": 14446
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.506044002116842e-06,
+      "loss": 0.8633,
+      "step": 14447
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5048452237004816e-06,
+      "loss": 1.6328,
+      "step": 14448
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5036466067138854e-06,
+      "loss": 1.3672,
+      "step": 14449
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.502448151186847e-06,
+      "loss": 1.1094,
+      "step": 14450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5012498571491506e-06,
+      "loss": 0.6836,
+      "step": 14451
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.5000517246305732e-06,
+      "loss": 0.6055,
+      "step": 14452
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.498853753660899e-06,
+      "loss": 0.3594,
+      "step": 14453
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.497655944269903e-06,
+      "loss": 1.8672,
+      "step": 14454
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4964582964873495e-06,
+      "loss": 0.9102,
+      "step": 14455
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.495260810343012e-06,
+      "loss": 0.6289,
+      "step": 14456
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4940634858666457e-06,
+      "loss": 0.8711,
+      "step": 14457
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4928663230880143e-06,
+      "loss": 0.5977,
+      "step": 14458
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.491669322036867e-06,
+      "loss": 0.9883,
+      "step": 14459
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4904724827429603e-06,
+      "loss": 0.5195,
+      "step": 14460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.489275805236032e-06,
+      "loss": 0.9648,
+      "step": 14461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4880792895458325e-06,
+      "loss": 0.7852,
+      "step": 14462
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.486882935702094e-06,
+      "loss": 0.3926,
+      "step": 14463
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4856867437345544e-06,
+      "loss": 0.9609,
+      "step": 14464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.484490713672939e-06,
+      "loss": 1.0469,
+      "step": 14465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.483294845546976e-06,
+      "loss": 0.7734,
+      "step": 14466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4820991393863923e-06,
+      "loss": 0.9492,
+      "step": 14467
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4809035952209003e-06,
+      "loss": 0.9102,
+      "step": 14468
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4797082130802127e-06,
+      "loss": 1.0312,
+      "step": 14469
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.478512992994042e-06,
+      "loss": 0.6172,
+      "step": 14470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4773179349920973e-06,
+      "loss": 0.7656,
+      "step": 14471
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4761230391040723e-06,
+      "loss": 0.8359,
+      "step": 14472
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4749283053596706e-06,
+      "loss": 0.6016,
+      "step": 14473
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.473733733788588e-06,
+      "loss": 0.3184,
+      "step": 14474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.472539324420511e-06,
+      "loss": 0.3711,
+      "step": 14475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.471345077285122e-06,
+      "loss": 1.0859,
+      "step": 14476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4701509924121056e-06,
+      "loss": 0.7305,
+      "step": 14477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4689570698311436e-06,
+      "loss": 0.3711,
+      "step": 14478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4677633095719063e-06,
+      "loss": 1.2734,
+      "step": 14479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4665697116640593e-06,
+      "loss": 0.3242,
+      "step": 14480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4653762761372712e-06,
+      "loss": 0.2578,
+      "step": 14481
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4641830030212077e-06,
+      "loss": 0.543,
+      "step": 14482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4629898923455195e-06,
+      "loss": 1.1016,
+      "step": 14483
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.461796944139866e-06,
+      "loss": 1.3203,
+      "step": 14484
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.46060415843389e-06,
+      "loss": 0.2793,
+      "step": 14485
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4594115352572442e-06,
+      "loss": 1.1016,
+      "step": 14486
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.458219074639563e-06,
+      "loss": 0.4492,
+      "step": 14487
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4570267766104904e-06,
+      "loss": 0.416,
+      "step": 14488
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4558346411996514e-06,
+      "loss": 0.4336,
+      "step": 14489
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.454642668436684e-06,
+      "loss": 1.4766,
+      "step": 14490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4534508583512073e-06,
+      "loss": 0.9922,
+      "step": 14491
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4522592109728457e-06,
+      "loss": 0.7852,
+      "step": 14492
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.451067726331212e-06,
+      "loss": 0.9453,
+      "step": 14493
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.449876404455923e-06,
+      "loss": 0.6172,
+      "step": 14494
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4486852453765894e-06,
+      "loss": 1.4609,
+      "step": 14495
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4474942491228137e-06,
+      "loss": 1.0078,
+      "step": 14496
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4463034157241926e-06,
+      "loss": 1.0625,
+      "step": 14497
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4451127452103274e-06,
+      "loss": 1.4141,
+      "step": 14498
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.443922237610814e-06,
+      "loss": 1.2422,
+      "step": 14499
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4427318929552365e-06,
+      "loss": 1.0469,
+      "step": 14500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4415417112731777e-06,
+      "loss": 1.2578,
+      "step": 14501
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4403516925942214e-06,
+      "loss": 0.7578,
+      "step": 14502
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4391618369479464e-06,
+      "loss": 0.5859,
+      "step": 14503
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.43797214436392e-06,
+      "loss": 0.2402,
+      "step": 14504
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.436782614871713e-06,
+      "loss": 0.7461,
+      "step": 14505
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4355932485008936e-06,
+      "loss": 0.7539,
+      "step": 14506
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.434404045281019e-06,
+      "loss": 1.125,
+      "step": 14507
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4332150052416413e-06,
+      "loss": 0.8945,
+      "step": 14508
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.432026128412317e-06,
+      "loss": 0.9688,
+      "step": 14509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4308374148225976e-06,
+      "loss": 0.9648,
+      "step": 14510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.42964886450202e-06,
+      "loss": 0.8281,
+      "step": 14511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4284604774801313e-06,
+      "loss": 0.6016,
+      "step": 14512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.427272253786461e-06,
+      "loss": 0.4941,
+      "step": 14513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4260841934505473e-06,
+      "loss": 0.8477,
+      "step": 14514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4248962965019116e-06,
+      "loss": 0.6172,
+      "step": 14515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.423708562970084e-06,
+      "loss": 1.5,
+      "step": 14516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4225209928845783e-06,
+      "loss": 1.0938,
+      "step": 14517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4213335862749175e-06,
+      "loss": 1.2656,
+      "step": 14518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.420146343170605e-06,
+      "loss": 0.9297,
+      "step": 14519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4189592636011557e-06,
+      "loss": 0.6953,
+      "step": 14520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4177723475960667e-06,
+      "loss": 1.0156,
+      "step": 14521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4165855951848416e-06,
+      "loss": 1.6797,
+      "step": 14522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4153990063969768e-06,
+      "loss": 1.0156,
+      "step": 14523
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.414212581261962e-06,
+      "loss": 1.8047,
+      "step": 14524
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4130263198092816e-06,
+      "loss": 1.4453,
+      "step": 14525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.411840222068421e-06,
+      "loss": 0.1289,
+      "step": 14526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.410654288068862e-06,
+      "loss": 0.4219,
+      "step": 14527
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.409468517840079e-06,
+      "loss": 0.249,
+      "step": 14528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.408282911411537e-06,
+      "loss": 0.8008,
+      "step": 14529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4070974688127066e-06,
+      "loss": 0.3535,
+      "step": 14530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.405912190073055e-06,
+      "loss": 0.9688,
+      "step": 14531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.404727075222035e-06,
+      "loss": 0.3262,
+      "step": 14532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4035421242891063e-06,
+      "loss": 1.5469,
+      "step": 14533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4023573373037134e-06,
+      "loss": 1.3047,
+      "step": 14534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4011727142953088e-06,
+      "loss": 0.2168,
+      "step": 14535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3999882552933295e-06,
+      "loss": 1.0078,
+      "step": 14536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3988039603272206e-06,
+      "loss": 1.2656,
+      "step": 14537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3976198294264084e-06,
+      "loss": 1.1484,
+      "step": 14538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.396435862620331e-06,
+      "loss": 0.5859,
+      "step": 14539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3952520599384077e-06,
+      "loss": 0.9805,
+      "step": 14540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3940684214100672e-06,
+      "loss": 0.6484,
+      "step": 14541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.392884947064721e-06,
+      "loss": 0.5195,
+      "step": 14542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.391701636931786e-06,
+      "loss": 0.918,
+      "step": 14543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.390518491040675e-06,
+      "loss": 0.4746,
+      "step": 14544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3893355094207914e-06,
+      "loss": 0.3965,
+      "step": 14545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3881526921015317e-06,
+      "loss": 1.5,
+      "step": 14546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.3869700391122994e-06,
+      "loss": 1.0234,
+      "step": 14547
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.38578755048249e-06,
+      "loss": 0.8555,
+      "step": 14548
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3846052262414864e-06,
+      "loss": 1.1328,
+      "step": 14549
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3834230664186775e-06,
+      "loss": 0.4941,
+      "step": 14550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3822410710434473e-06,
+      "loss": 0.8516,
+      "step": 14551
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.38105924014517e-06,
+      "loss": 0.4922,
+      "step": 14552
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.379877573753215e-06,
+      "loss": 0.4941,
+      "step": 14553
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.378696071896955e-06,
+      "loss": 0.832,
+      "step": 14554
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.377514734605759e-06,
+      "loss": 1.1875,
+      "step": 14555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.376333561908983e-06,
+      "loss": 0.3555,
+      "step": 14556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.375152553835981e-06,
+      "loss": 1.0,
+      "step": 14557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3739717104161095e-06,
+      "loss": 0.5547,
+      "step": 14558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3727910316787193e-06,
+      "loss": 1.3203,
+      "step": 14559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3716105176531488e-06,
+      "loss": 0.4434,
+      "step": 14560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3704301683687445e-06,
+      "loss": 0.8672,
+      "step": 14561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3692499838548375e-06,
+      "loss": 1.3125,
+      "step": 14562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3680699641407645e-06,
+      "loss": 0.707,
+      "step": 14563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3668901092558483e-06,
+      "loss": 1.0469,
+      "step": 14564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3657104192294197e-06,
+      "loss": 0.5742,
+      "step": 14565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3645308940907907e-06,
+      "loss": 1.1562,
+      "step": 14566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3633515338692846e-06,
+      "loss": 1.2812,
+      "step": 14567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3621723385942063e-06,
+      "loss": 0.6992,
+      "step": 14568
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.36099330829487e-06,
+      "loss": 0.8359,
+      "step": 14569
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.359814443000573e-06,
+      "loss": 0.9844,
+      "step": 14570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.358635742740617e-06,
+      "loss": 0.6328,
+      "step": 14571
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.357457207544301e-06,
+      "loss": 0.6875,
+      "step": 14572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3562788374409126e-06,
+      "loss": 0.8984,
+      "step": 14573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.355100632459737e-06,
+      "loss": 0.6562,
+      "step": 14574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.353922592630058e-06,
+      "loss": 0.9922,
+      "step": 14575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.352744717981159e-06,
+      "loss": 0.7969,
+      "step": 14576
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.351567008542309e-06,
+      "loss": 0.5898,
+      "step": 14577
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3503894643427816e-06,
+      "loss": 0.2656,
+      "step": 14578
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.349212085411845e-06,
+      "loss": 0.9219,
+      "step": 14579
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3480348717787603e-06,
+      "loss": 0.8008,
+      "step": 14580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.346857823472782e-06,
+      "loss": 0.8555,
+      "step": 14581
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3456809405231673e-06,
+      "loss": 1.2188,
+      "step": 14582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3445042229591697e-06,
+      "loss": 0.3555,
+      "step": 14583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3433276708100316e-06,
+      "loss": 0.6055,
+      "step": 14584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.342151284104991e-06,
+      "loss": 0.3008,
+      "step": 14585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3409750628732908e-06,
+      "loss": 0.3574,
+      "step": 14586
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3397990071441668e-06,
+      "loss": 0.3594,
+      "step": 14587
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.338623116946842e-06,
+      "loss": 0.4629,
+      "step": 14588
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.337447392310549e-06,
+      "loss": 1.0547,
+      "step": 14589
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3362718332645016e-06,
+      "loss": 1.2656,
+      "step": 14590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3350964398379247e-06,
+      "loss": 1.5156,
+      "step": 14591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3339212120600227e-06,
+      "loss": 0.582,
+      "step": 14592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3327461499600143e-06,
+      "loss": 2.25,
+      "step": 14593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.331571253567095e-06,
+      "loss": 0.8828,
+      "step": 14594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.330396522910474e-06,
+      "loss": 1.0469,
+      "step": 14595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3292219580193395e-06,
+      "loss": 0.6406,
+      "step": 14596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.328047558922892e-06,
+      "loss": 1.0312,
+      "step": 14597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.326873325650313e-06,
+      "loss": 0.9141,
+      "step": 14598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.32569925823079e-06,
+      "loss": 0.2871,
+      "step": 14599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3245253566935054e-06,
+      "loss": 0.6406,
+      "step": 14600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3233516210676343e-06,
+      "loss": 0.8672,
+      "step": 14601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.322178051382342e-06,
+      "loss": 1.1094,
+      "step": 14602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.321004647666802e-06,
+      "loss": 1.2578,
+      "step": 14603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3198314099501815e-06,
+      "loss": 0.375,
+      "step": 14604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3186583382616357e-06,
+      "loss": 0.2109,
+      "step": 14605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.317485432630315e-06,
+      "loss": 1.1016,
+      "step": 14606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3163126930853772e-06,
+      "loss": 0.6211,
+      "step": 14607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3151401196559717e-06,
+      "loss": 1.6406,
+      "step": 14608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.313967712371233e-06,
+      "loss": 0.832,
+      "step": 14609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3127954712603094e-06,
+      "loss": 1.1797,
+      "step": 14610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.311623396352326e-06,
+      "loss": 0.9531,
+      "step": 14611
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3104514876764225e-06,
+      "loss": 0.5078,
+      "step": 14612
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3092797452617187e-06,
+      "loss": 1.1484,
+      "step": 14613
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3081081691373418e-06,
+      "loss": 1.0547,
+      "step": 14614
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.306936759332404e-06,
+      "loss": 0.9219,
+      "step": 14615
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3057655158760236e-06,
+      "loss": 0.6406,
+      "step": 14616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.304594438797313e-06,
+      "loss": 0.7812,
+      "step": 14617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3034235281253755e-06,
+      "loss": 0.8516,
+      "step": 14618
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.302252783889308e-06,
+      "loss": 0.4922,
+      "step": 14619
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3010822061182137e-06,
+      "loss": 0.8555,
+      "step": 14620
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2999117948411865e-06,
+      "loss": 0.7266,
+      "step": 14621
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2987415500873143e-06,
+      "loss": 0.8086,
+      "step": 14622
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.297571471885678e-06,
+      "loss": 0.6875,
+      "step": 14623
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.296401560265362e-06,
+      "loss": 1.1328,
+      "step": 14624
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2952318152554475e-06,
+      "loss": 0.8242,
+      "step": 14625
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.2940622368849984e-06,
+      "loss": 0.9219,
+      "step": 14626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2928928251830895e-06,
+      "loss": 0.7031,
+      "step": 14627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.291723580178785e-06,
+      "loss": 0.6523,
+      "step": 14628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.290554501901145e-06,
+      "loss": 0.457,
+      "step": 14629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2893855903792216e-06,
+      "loss": 0.2227,
+      "step": 14630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.288216845642068e-06,
+      "loss": 0.75,
+      "step": 14631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2870482677187387e-06,
+      "loss": 0.2715,
+      "step": 14632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.285879856638271e-06,
+      "loss": 0.252,
+      "step": 14633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.284711612429703e-06,
+      "loss": 0.4512,
+      "step": 14634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2835435351220723e-06,
+      "loss": 0.5234,
+      "step": 14635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.282375624744415e-06,
+      "loss": 0.582,
+      "step": 14636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.28120788132575e-06,
+      "loss": 0.9141,
+      "step": 14637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2800403048951068e-06,
+      "loss": 1.125,
+      "step": 14638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2788728954814985e-06,
+      "loss": 0.8555,
+      "step": 14639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2777056531139463e-06,
+      "loss": 0.3086,
+      "step": 14640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.276538577821454e-06,
+      "loss": 1.2578,
+      "step": 14641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.275371669633035e-06,
+      "loss": 0.9688,
+      "step": 14642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2742049285776833e-06,
+      "loss": 0.8086,
+      "step": 14643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.273038354684401e-06,
+      "loss": 1.125,
+      "step": 14644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.271871947982187e-06,
+      "loss": 0.9766,
+      "step": 14645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.270705708500025e-06,
+      "loss": 1.4062,
+      "step": 14646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2695396362668984e-06,
+      "loss": 1.1484,
+      "step": 14647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2683737313117937e-06,
+      "loss": 0.6328,
+      "step": 14648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.267207993663688e-06,
+      "loss": 0.9023,
+      "step": 14649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.266042423351553e-06,
+      "loss": 0.8633,
+      "step": 14650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2648770204043557e-06,
+      "loss": 0.7109,
+      "step": 14651
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.263711784851061e-06,
+      "loss": 0.7109,
+      "step": 14652
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.262546716720636e-06,
+      "loss": 0.7617,
+      "step": 14653
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2613818160420276e-06,
+      "loss": 0.3047,
+      "step": 14654
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2602170828441938e-06,
+      "loss": 0.8711,
+      "step": 14655
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2590525171560847e-06,
+      "loss": 0.7578,
+      "step": 14656
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2578881190066415e-06,
+      "loss": 0.4785,
+      "step": 14657
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.256723888424801e-06,
+      "loss": 0.3926,
+      "step": 14658
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2555598254395017e-06,
+      "loss": 0.4824,
+      "step": 14659
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.254395930079678e-06,
+      "loss": 1.7891,
+      "step": 14660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2532322023742545e-06,
+      "loss": 1.2969,
+      "step": 14661
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2520686423521507e-06,
+      "loss": 0.5273,
+      "step": 14662
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.250905250042289e-06,
+      "loss": 0.9414,
+      "step": 14663
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.249742025473588e-06,
+      "loss": 1.0156,
+      "step": 14664
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2485789686749502e-06,
+      "loss": 0.2773,
+      "step": 14665
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2474160796752907e-06,
+      "loss": 1.1719,
+      "step": 14666
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.246253358503504e-06,
+      "loss": 0.9961,
+      "step": 14667
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2450908051884954e-06,
+      "loss": 0.6484,
+      "step": 14668
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2439284197591514e-06,
+      "loss": 0.7812,
+      "step": 14669
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2427662022443695e-06,
+      "loss": 0.6875,
+      "step": 14670
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.241604152673027e-06,
+      "loss": 0.9531,
+      "step": 14671
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2404422710740146e-06,
+      "loss": 0.4941,
+      "step": 14672
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2392805574762e-06,
+      "loss": 1.7422,
+      "step": 14673
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.238119011908465e-06,
+      "loss": 0.6055,
+      "step": 14674
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.236957634399671e-06,
+      "loss": 1.0391,
+      "step": 14675
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2357964249786866e-06,
+      "loss": 1.1406,
+      "step": 14676
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.234635383674375e-06,
+      "loss": 0.3398,
+      "step": 14677
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2334745105155895e-06,
+      "loss": 0.5977,
+      "step": 14678
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2323138055311785e-06,
+      "loss": 0.7305,
+      "step": 14679
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2311532687499947e-06,
+      "loss": 0.9883,
+      "step": 14680
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2299929002008845e-06,
+      "loss": 0.4062,
+      "step": 14681
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2288326999126806e-06,
+      "loss": 1.2109,
+      "step": 14682
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2276726679142256e-06,
+      "loss": 0.5781,
+      "step": 14683
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.226512804234343e-06,
+      "loss": 0.6016,
+      "step": 14684
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2253531089018684e-06,
+      "loss": 0.6523,
+      "step": 14685
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2241935819456173e-06,
+      "loss": 0.2451,
+      "step": 14686
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2230342233944146e-06,
+      "loss": 1.6172,
+      "step": 14687
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.221875033277068e-06,
+      "loss": 0.3828,
+      "step": 14688
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2207160116223957e-06,
+      "loss": 0.9414,
+      "step": 14689
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2195571584591955e-06,
+      "loss": 1.5703,
+      "step": 14690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.218398473816278e-06,
+      "loss": 1.1719,
+      "step": 14691
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2172399577224334e-06,
+      "loss": 1.5469,
+      "step": 14692
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2160816102064584e-06,
+      "loss": 1.3594,
+      "step": 14693
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2149234312971467e-06,
+      "loss": 1.2734,
+      "step": 14694
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2137654210232793e-06,
+      "loss": 1.0703,
+      "step": 14695
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2126075794136337e-06,
+      "loss": 1.3906,
+      "step": 14696
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2114499064969905e-06,
+      "loss": 0.9336,
+      "step": 14697
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2102924023021267e-06,
+      "loss": 0.3184,
+      "step": 14698
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2091350668578058e-06,
+      "loss": 0.6055,
+      "step": 14699
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.20797790019279e-06,
+      "loss": 0.8164,
+      "step": 14700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.206820902335841e-06,
+      "loss": 0.6914,
+      "step": 14701
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2056640733157206e-06,
+      "loss": 0.9531,
+      "step": 14702
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.204507413161171e-06,
+      "loss": 0.6719,
+      "step": 14703
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.203350921900945e-06,
+      "loss": 1.2422,
+      "step": 14704
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.202194599563788e-06,
+      "loss": 0.3359,
+      "step": 14705
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.201038446178437e-06,
+      "loss": 0.6641,
+      "step": 14706
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1998824617736234e-06,
+      "loss": 0.5898,
+      "step": 14707
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1987266463780798e-06,
+      "loss": 0.3594,
+      "step": 14708
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1975710000205375e-06,
+      "loss": 0.4922,
+      "step": 14709
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1964155227297144e-06,
+      "loss": 0.543,
+      "step": 14710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.195260214534327e-06,
+      "loss": 0.8008,
+      "step": 14711
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1941050754630898e-06,
+      "loss": 1.0781,
+      "step": 14712
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1929501055447175e-06,
+      "loss": 0.9102,
+      "step": 14713
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1917953048079098e-06,
+      "loss": 0.6914,
+      "step": 14714
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.190640673281372e-06,
+      "loss": 1.2109,
+      "step": 14715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1894862109937974e-06,
+      "loss": 1.1562,
+      "step": 14716
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.188331917973884e-06,
+      "loss": 0.7344,
+      "step": 14717
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1871777942503134e-06,
+      "loss": 0.6641,
+      "step": 14718
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1860238398517774e-06,
+      "loss": 0.8906,
+      "step": 14719
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.18487005480695e-06,
+      "loss": 0.7734,
+      "step": 14720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.18371643914451e-06,
+      "loss": 0.7695,
+      "step": 14721
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.182562992893132e-06,
+      "loss": 1.5703,
+      "step": 14722
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1814097160814806e-06,
+      "loss": 0.8516,
+      "step": 14723
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1802566087382167e-06,
+      "loss": 1.1641,
+      "step": 14724
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.179103670892002e-06,
+      "loss": 0.5781,
+      "step": 14725
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.177950902571495e-06,
+      "loss": 0.9023,
+      "step": 14726
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1767983038053442e-06,
+      "loss": 0.9453,
+      "step": 14727
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.175645874622191e-06,
+      "loss": 0.3965,
+      "step": 14728
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.174493615050681e-06,
+      "loss": 1.0391,
+      "step": 14729
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.173341525119458e-06,
+      "loss": 0.1543,
+      "step": 14730
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1721896048571464e-06,
+      "loss": 0.7031,
+      "step": 14731
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1710378542923815e-06,
+      "loss": 0.3242,
+      "step": 14732
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1698862734537907e-06,
+      "loss": 0.7734,
+      "step": 14733
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.168734862369992e-06,
+      "loss": 0.4961,
+      "step": 14734
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1675836210696e-06,
+      "loss": 0.1982,
+      "step": 14735
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.16643254958123e-06,
+      "loss": 0.6289,
+      "step": 14736
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1652816479334947e-06,
+      "loss": 0.2539,
+      "step": 14737
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1641309161549937e-06,
+      "loss": 1.1016,
+      "step": 14738
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1629803542743253e-06,
+      "loss": 0.9961,
+      "step": 14739
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1618299623200887e-06,
+      "loss": 0.3281,
+      "step": 14740
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.160679740320878e-06,
+      "loss": 0.123,
+      "step": 14741
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1595296883052737e-06,
+      "loss": 0.3652,
+      "step": 14742
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1583798063018676e-06,
+      "loss": 0.9336,
+      "step": 14743
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.157230094339231e-06,
+      "loss": 1.2188,
+      "step": 14744
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.156080552445945e-06,
+      "loss": 0.7109,
+      "step": 14745
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1549311806505735e-06,
+      "loss": 0.6406,
+      "step": 14746
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1537819789816894e-06,
+      "loss": 0.6484,
+      "step": 14747
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1526329474678484e-06,
+      "loss": 0.4082,
+      "step": 14748
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1514840861376127e-06,
+      "loss": 1.2969,
+      "step": 14749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.150335395019538e-06,
+      "loss": 1.1562,
+      "step": 14750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.149186874142172e-06,
+      "loss": 0.9023,
+      "step": 14751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1480385235340537e-06,
+      "loss": 1.4219,
+      "step": 14752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1468903432237295e-06,
+      "loss": 0.6914,
+      "step": 14753
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1457423332397395e-06,
+      "loss": 1.1094,
+      "step": 14754
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1445944936106123e-06,
+      "loss": 1.7891,
+      "step": 14755
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.143446824364873e-06,
+      "loss": 1.375,
+      "step": 14756
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1422993255310485e-06,
+      "loss": 1.3438,
+      "step": 14757
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1411519971376635e-06,
+      "loss": 0.5547,
+      "step": 14758
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1400048392132243e-06,
+      "loss": 0.6172,
+      "step": 14759
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.138857851786251e-06,
+      "loss": 0.8477,
+      "step": 14760
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1377110348852435e-06,
+      "loss": 0.4219,
+      "step": 14761
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.136564388538711e-06,
+      "loss": 1.0078,
+      "step": 14762
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.135417912775145e-06,
+      "loss": 1.5391,
+      "step": 14763
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1342716076230485e-06,
+      "loss": 1.2188,
+      "step": 14764
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1331254731109017e-06,
+      "loss": 0.8047,
+      "step": 14765
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1319795092672e-06,
+      "loss": 0.8672,
+      "step": 14766
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1308337161204172e-06,
+      "loss": 0.8789,
+      "step": 14767
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1296880936990383e-06,
+      "loss": 0.6367,
+      "step": 14768
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.128542642031528e-06,
+      "loss": 0.7422,
+      "step": 14769
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1273973611463605e-06,
+      "loss": 0.7539,
+      "step": 14770
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1262522510720026e-06,
+      "loss": 0.8477,
+      "step": 14771
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1251073118369112e-06,
+      "loss": 1.2578,
+      "step": 14772
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.12396254346954e-06,
+      "loss": 1.1094,
+      "step": 14773
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.122817945998344e-06,
+      "loss": 0.2246,
+      "step": 14774
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1216735194517746e-06,
+      "loss": 1.3984,
+      "step": 14775
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1205292638582705e-06,
+      "loss": 0.334,
+      "step": 14776
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1193851792462683e-06,
+      "loss": 1.0469,
+      "step": 14777
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1182412656442072e-06,
+      "loss": 0.3008,
+      "step": 14778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.11709752308052e-06,
+      "loss": 0.2021,
+      "step": 14779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1159539515836267e-06,
+      "loss": 0.2695,
+      "step": 14780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1148105511819538e-06,
+      "loss": 0.6133,
+      "step": 14781
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1136673219039217e-06,
+      "loss": 0.3477,
+      "step": 14782
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1125242637779408e-06,
+      "loss": 0.5391,
+      "step": 14783
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1113813768324163e-06,
+      "loss": 1.1406,
+      "step": 14784
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.110238661095758e-06,
+      "loss": 0.6641,
+      "step": 14785
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.109096116596371e-06,
+      "loss": 0.8594,
+      "step": 14786
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.107953743362643e-06,
+      "loss": 0.9336,
+      "step": 14787
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.106811541422974e-06,
+      "loss": 0.6445,
+      "step": 14788
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1056695108057455e-06,
+      "loss": 0.2578,
+      "step": 14789
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1045276515393484e-06,
+      "loss": 0.3301,
+      "step": 14790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.103385963652156e-06,
+      "loss": 0.7422,
+      "step": 14791
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.10224444717255e-06,
+      "loss": 0.9141,
+      "step": 14792
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.101103102128894e-06,
+      "loss": 1.3438,
+      "step": 14793
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.099961928549562e-06,
+      "loss": 0.5195,
+      "step": 14794
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.09882092646291e-06,
+      "loss": 0.3613,
+      "step": 14795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.097680095897303e-06,
+      "loss": 0.6836,
+      "step": 14796
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0965394368810886e-06,
+      "loss": 0.5586,
+      "step": 14797
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0953989494426195e-06,
+      "loss": 0.6133,
+      "step": 14798
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0942586336102452e-06,
+      "loss": 0.8867,
+      "step": 14799
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0931184894123036e-06,
+      "loss": 0.582,
+      "step": 14800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.091978516877128e-06,
+      "loss": 0.5469,
+      "step": 14801
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0908387160330533e-06,
+      "loss": 1.4844,
+      "step": 14802
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0896990869084133e-06,
+      "loss": 0.3301,
+      "step": 14803
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.088559629531528e-06,
+      "loss": 0.8672,
+      "step": 14804
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.087420343930714e-06,
+      "loss": 0.6406,
+      "step": 14805
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.08628123013429e-06,
+      "loss": 1.1641,
+      "step": 14806
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0851422881705717e-06,
+      "loss": 1.4219,
+      "step": 14807
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0840035180678595e-06,
+      "loss": 0.2949,
+      "step": 14808
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.082864919854458e-06,
+      "loss": 0.4961,
+      "step": 14809
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.08172649355867e-06,
+      "loss": 0.2734,
+      "step": 14810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0805882392087873e-06,
+      "loss": 1.3672,
+      "step": 14811
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.079450156833096e-06,
+      "loss": 0.3379,
+      "step": 14812
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0783122464598847e-06,
+      "loss": 0.6875,
+      "step": 14813
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0771745081174375e-06,
+      "loss": 0.5898,
+      "step": 14814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0760369418340275e-06,
+      "loss": 0.6289,
+      "step": 14815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.074899547637933e-06,
+      "loss": 0.6523,
+      "step": 14816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0737623255574143e-06,
+      "loss": 1.0234,
+      "step": 14817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0726252756207455e-06,
+      "loss": 0.9375,
+      "step": 14818
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0714883978561782e-06,
+      "loss": 0.6367,
+      "step": 14819
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.070351692291974e-06,
+      "loss": 0.8945,
+      "step": 14820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.06921515895638e-06,
+      "loss": 0.4727,
+      "step": 14821
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0680787978776493e-06,
+      "loss": 0.9453,
+      "step": 14822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.066942609084017e-06,
+      "loss": 1.0078,
+      "step": 14823
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0658065926037305e-06,
+      "loss": 0.2676,
+      "step": 14824
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0646707484650163e-06,
+      "loss": 0.9023,
+      "step": 14825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.063535076696107e-06,
+      "loss": 0.7109,
+      "step": 14826
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0623995773252337e-06,
+      "loss": 0.8555,
+      "step": 14827
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0612642503806133e-06,
+      "loss": 0.6445,
+      "step": 14828
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0601290958904615e-06,
+      "loss": 0.4316,
+      "step": 14829
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0589941138829924e-06,
+      "loss": 1.1406,
+      "step": 14830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.057859304386419e-06,
+      "loss": 0.625,
+      "step": 14831
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.056724667428942e-06,
+      "loss": 0.2812,
+      "step": 14832
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0555902030387587e-06,
+      "loss": 0.4102,
+      "step": 14833
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0544559112440687e-06,
+      "loss": 1.2109,
+      "step": 14834
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.053321792073065e-06,
+      "loss": 0.3691,
+      "step": 14835
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.05218784555393e-06,
+      "loss": 0.7617,
+      "step": 14836
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.051054071714853e-06,
+      "loss": 1.5156,
+      "step": 14837
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0499204705840057e-06,
+      "loss": 0.7617,
+      "step": 14838
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0487870421895695e-06,
+      "loss": 1.0391,
+      "step": 14839
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.047653786559708e-06,
+      "loss": 1.0859,
+      "step": 14840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0465207037225917e-06,
+      "loss": 0.4609,
+      "step": 14841
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0453877937063782e-06,
+      "loss": 1.0781,
+      "step": 14842
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0442550565392303e-06,
+      "loss": 0.3984,
+      "step": 14843
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0431224922492932e-06,
+      "loss": 0.249,
+      "step": 14844
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0419901008647236e-06,
+      "loss": 1.125,
+      "step": 14845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.040857882413658e-06,
+      "loss": 1.6953,
+      "step": 14846
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.039725836924241e-06,
+      "loss": 0.7969,
+      "step": 14847
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.038593964424611e-06,
+      "loss": 0.6797,
+      "step": 14848
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0374622649428964e-06,
+      "loss": 0.6406,
+      "step": 14849
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0363307385072194e-06,
+      "loss": 0.8594,
+      "step": 14850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0351993851457083e-06,
+      "loss": 0.543,
+      "step": 14851
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0340682048864845e-06,
+      "loss": 0.8711,
+      "step": 14852
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.032937197757654e-06,
+      "loss": 0.8008,
+      "step": 14853
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0318063637873318e-06,
+      "loss": 0.582,
+      "step": 14854
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.030675703003626e-06,
+      "loss": 0.9922,
+      "step": 14855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.029545215434635e-06,
+      "loss": 1.3281,
+      "step": 14856
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.028414901108452e-06,
+      "loss": 1.0156,
+      "step": 14857
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.027284760053173e-06,
+      "loss": 0.582,
+      "step": 14858
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0261547922968904e-06,
+      "loss": 0.6602,
+      "step": 14859
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.025024997867685e-06,
+      "loss": 1.2109,
+      "step": 14860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.023895376793632e-06,
+      "loss": 1.3047,
+      "step": 14861
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0227659291028112e-06,
+      "loss": 1.3594,
+      "step": 14862
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0216366548232967e-06,
+      "loss": 0.4844,
+      "step": 14863
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0205075539831485e-06,
+      "loss": 0.8984,
+      "step": 14864
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.0193786266104363e-06,
+      "loss": 0.8281,
+      "step": 14865
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.018249872733211e-06,
+      "loss": 1.1875,
+      "step": 14866
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0171212923795347e-06,
+      "loss": 0.4746,
+      "step": 14867
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.015992885577449e-06,
+      "loss": 0.9492,
+      "step": 14868
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0148646523550053e-06,
+      "loss": 0.832,
+      "step": 14869
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0137365927402397e-06,
+      "loss": 0.6914,
+      "step": 14870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0126087067611944e-06,
+      "loss": 0.9961,
+      "step": 14871
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0114809944458944e-06,
+      "loss": 1.0703,
+      "step": 14872
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0103534558223757e-06,
+      "loss": 1.8281,
+      "step": 14873
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0092260909186546e-06,
+      "loss": 0.4062,
+      "step": 14874
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0080988997627536e-06,
+      "loss": 0.2695,
+      "step": 14875
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.006971882382692e-06,
+      "loss": 0.2236,
+      "step": 14876
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0058450388064776e-06,
+      "loss": 1.1719,
+      "step": 14877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.004718369062112e-06,
+      "loss": 0.5781,
+      "step": 14878
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0035918731776016e-06,
+      "loss": 0.8164,
+      "step": 14879
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0024655511809463e-06,
+      "loss": 0.5664,
+      "step": 14880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0013394031001374e-06,
+      "loss": 0.4258,
+      "step": 14881
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.000213428963158e-06,
+      "loss": 0.832,
+      "step": 14882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9990876287980044e-06,
+      "loss": 1.5781,
+      "step": 14883
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.997962002632652e-06,
+      "loss": 0.8828,
+      "step": 14884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9968365504950726e-06,
+      "loss": 1.1797,
+      "step": 14885
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.995711272413242e-06,
+      "loss": 0.4902,
+      "step": 14886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.994586168415131e-06,
+      "loss": 0.7461,
+      "step": 14887
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.993461238528699e-06,
+      "loss": 0.3008,
+      "step": 14888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.992336482781902e-06,
+      "loss": 0.6602,
+      "step": 14889
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9912119012026985e-06,
+      "loss": 0.9492,
+      "step": 14890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9900874938190404e-06,
+      "loss": 0.9727,
+      "step": 14891
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.988963260658868e-06,
+      "loss": 1.3828,
+      "step": 14892
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.987839201750129e-06,
+      "loss": 0.7656,
+      "step": 14893
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9867153171207552e-06,
+      "loss": 1.2734,
+      "step": 14894
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9855916067986844e-06,
+      "loss": 0.7305,
+      "step": 14895
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.98446807081184e-06,
+      "loss": 0.8359,
+      "step": 14896
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9833447091881516e-06,
+      "loss": 0.7344,
+      "step": 14897
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.982221521955533e-06,
+      "loss": 0.9062,
+      "step": 14898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9810985091419065e-06,
+      "loss": 0.2412,
+      "step": 14899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.979975670775177e-06,
+      "loss": 1.0781,
+      "step": 14900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9788530068832575e-06,
+      "loss": 0.8789,
+      "step": 14901
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.977730517494044e-06,
+      "loss": 0.6602,
+      "step": 14902
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.976608202635438e-06,
+      "loss": 0.4688,
+      "step": 14903
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9754860623353367e-06,
+      "loss": 0.3184,
+      "step": 14904
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.974364096621627e-06,
+      "loss": 1.4375,
+      "step": 14905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9732423055221893e-06,
+      "loss": 0.4238,
+      "step": 14906
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.972120689064909e-06,
+      "loss": 0.6836,
+      "step": 14907
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9709992472776662e-06,
+      "loss": 1.1875,
+      "step": 14908
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9698779801883283e-06,
+      "loss": 0.3887,
+      "step": 14909
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.968756887824761e-06,
+      "loss": 1.2344,
+      "step": 14910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9676359702148304e-06,
+      "loss": 1.3594,
+      "step": 14911
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9665152273863997e-06,
+      "loss": 0.8086,
+      "step": 14912
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9653946593673165e-06,
+      "loss": 0.8945,
+      "step": 14913
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9642742661854386e-06,
+      "loss": 1.0859,
+      "step": 14914
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.963154047868604e-06,
+      "loss": 0.7969,
+      "step": 14915
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9620340044446625e-06,
+      "loss": 1.1172,
+      "step": 14916
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.960914135941444e-06,
+      "loss": 0.9141,
+      "step": 14917
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9597944423867895e-06,
+      "loss": 0.5625,
+      "step": 14918
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.95867492380852e-06,
+      "loss": 0.457,
+      "step": 14919
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9575555802344635e-06,
+      "loss": 0.625,
+      "step": 14920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9564364116924425e-06,
+      "loss": 0.6914,
+      "step": 14921
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9553174182102706e-06,
+      "loss": 1.3359,
+      "step": 14922
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.954198599815755e-06,
+      "loss": 0.9023,
+      "step": 14923
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9530799565367075e-06,
+      "loss": 1.4219,
+      "step": 14924
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.951961488400932e-06,
+      "loss": 0.3184,
+      "step": 14925
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.950843195436224e-06,
+      "loss": 1.2969,
+      "step": 14926
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9497250776703746e-06,
+      "loss": 0.3066,
+      "step": 14927
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9486071351311764e-06,
+      "loss": 1.8672,
+      "step": 14928
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.947489367846419e-06,
+      "loss": 0.8711,
+      "step": 14929
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.946371775843875e-06,
+      "loss": 0.6406,
+      "step": 14930
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.945254359151324e-06,
+      "loss": 0.4004,
+      "step": 14931
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9441371177965428e-06,
+      "loss": 0.2383,
+      "step": 14932
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.943020051807296e-06,
+      "loss": 0.1064,
+      "step": 14933
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9419031612113424e-06,
+      "loss": 1.0547,
+      "step": 14934
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9407864460364445e-06,
+      "loss": 0.5352,
+      "step": 14935
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9396699063103616e-06,
+      "loss": 0.291,
+      "step": 14936
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.93855354206084e-06,
+      "loss": 0.918,
+      "step": 14937
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.937437353315622e-06,
+      "loss": 0.332,
+      "step": 14938
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9363213401024525e-06,
+      "loss": 0.6797,
+      "step": 14939
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9352055024490733e-06,
+      "loss": 1.1484,
+      "step": 14940
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9340898403832095e-06,
+      "loss": 1.0078,
+      "step": 14941
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.932974353932596e-06,
+      "loss": 1.1875,
+      "step": 14942
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9318590431249518e-06,
+      "loss": 1.4141,
+      "step": 14943
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.930743907988002e-06,
+      "loss": 0.8555,
+      "step": 14944
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9296289485494557e-06,
+      "loss": 1.1094,
+      "step": 14945
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9285141648370296e-06,
+      "loss": 1.1094,
+      "step": 14946
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9273995568784263e-06,
+      "loss": 0.5547,
+      "step": 14947
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9262851247013523e-06,
+      "loss": 1.5078,
+      "step": 14948
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.925170868333501e-06,
+      "loss": 1.3203,
+      "step": 14949
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9240567878025706e-06,
+      "loss": 1.2109,
+      "step": 14950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.922942883136245e-06,
+      "loss": 0.6992,
+      "step": 14951
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9218291543622123e-06,
+      "loss": 1.4297,
+      "step": 14952
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.920715601508155e-06,
+      "loss": 0.5742,
+      "step": 14953
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9196022246017474e-06,
+      "loss": 1.1484,
+      "step": 14954
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.918489023670659e-06,
+      "loss": 0.8867,
+      "step": 14955
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9173759987425567e-06,
+      "loss": 0.9805,
+      "step": 14956
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9162631498451097e-06,
+      "loss": 1.2891,
+      "step": 14957
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.915150477005969e-06,
+      "loss": 0.5078,
+      "step": 14958
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.914037980252793e-06,
+      "loss": 0.8164,
+      "step": 14959
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9129256596132337e-06,
+      "loss": 0.4902,
+      "step": 14960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.911813515114934e-06,
+      "loss": 0.9648,
+      "step": 14961
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9107015467855314e-06,
+      "loss": 1.0156,
+      "step": 14962
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9095897546526664e-06,
+      "loss": 0.5625,
+      "step": 14963
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.908478138743973e-06,
+      "loss": 0.6875,
+      "step": 14964
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9073666990870787e-06,
+      "loss": 1.125,
+      "step": 14965
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9062554357096017e-06,
+      "loss": 0.9531,
+      "step": 14966
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.905144348639165e-06,
+      "loss": 0.6758,
+      "step": 14967
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.904033437903386e-06,
+      "loss": 0.3047,
+      "step": 14968
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.90292270352987e-06,
+      "loss": 1.2656,
+      "step": 14969
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.901812145546229e-06,
+      "loss": 0.9023,
+      "step": 14970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9007017639800584e-06,
+      "loss": 0.6992,
+      "step": 14971
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8995915588589607e-06,
+      "loss": 1.1875,
+      "step": 14972
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.898481530210524e-06,
+      "loss": 0.2393,
+      "step": 14973
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8973716780623416e-06,
+      "loss": 1.4375,
+      "step": 14974
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8962620024419963e-06,
+      "loss": 0.5664,
+      "step": 14975
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8951525033770635e-06,
+      "loss": 0.3301,
+      "step": 14976
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.894043180895122e-06,
+      "loss": 0.625,
+      "step": 14977
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8929340350237455e-06,
+      "loss": 0.8398,
+      "step": 14978
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8918250657904945e-06,
+      "loss": 0.4121,
+      "step": 14979
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.890716273222934e-06,
+      "loss": 0.1226,
+      "step": 14980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8896076573486264e-06,
+      "loss": 0.3086,
+      "step": 14981
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.888499218195121e-06,
+      "loss": 1.1484,
+      "step": 14982
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.887390955789963e-06,
+      "loss": 0.6055,
+      "step": 14983
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8862828701607005e-06,
+      "loss": 0.6602,
+      "step": 14984
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.885174961334878e-06,
+      "loss": 0.2061,
+      "step": 14985
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.884067229340023e-06,
+      "loss": 0.8438,
+      "step": 14986
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.882959674203675e-06,
+      "loss": 0.3066,
+      "step": 14987
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8818522959533544e-06,
+      "loss": 1.25,
+      "step": 14988
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8807450946165893e-06,
+      "loss": 1.0078,
+      "step": 14989
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8796380702208924e-06,
+      "loss": 0.2031,
+      "step": 14990
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8785312227937847e-06,
+      "loss": 1.1562,
+      "step": 14991
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8774245523627663e-06,
+      "loss": 1.0625,
+      "step": 14992
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.876318058955352e-06,
+      "loss": 1.2734,
+      "step": 14993
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8752117425990345e-06,
+      "loss": 1.0312,
+      "step": 14994
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8741056033213167e-06,
+      "loss": 1.0312,
+      "step": 14995
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.872999641149683e-06,
+      "loss": 0.9219,
+      "step": 14996
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.871893856111626e-06,
+      "loss": 0.875,
+      "step": 14997
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8707882482346305e-06,
+      "loss": 0.3711,
+      "step": 14998
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.869682817546172e-06,
+      "loss": 0.6133,
+      "step": 14999
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8685775640737222e-06,
+      "loss": 0.7031,
+      "step": 15000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 15942,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "total_flos": 5.302734647427072e+16,
+  "trial_name": null,
+  "trial_params": null
+}