diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26751 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.964060264287518,
+  "eval_steps": 500,
+  "global_step": 4455,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7251,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.684,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.7066,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6894,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.7617,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6869,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-05,
+      "loss": 0.6697,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.7293,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6381,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6163,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999998596668246e-05,
+      "loss": 0.6307,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999994386673375e-05,
+      "loss": 0.6536,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999987370016564e-05,
+      "loss": 0.6721,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999977546699793e-05,
+      "loss": 0.6227,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999964916725805e-05,
+      "loss": 0.6545,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999949480098158e-05,
+      "loss": 0.6805,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999993123682118e-05,
+      "loss": 0.661,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999910186899984e-05,
+      "loss": 0.6475,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999988633034049e-05,
+      "loss": 0.6131,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999859667149386e-05,
+      "loss": 0.6337,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999830197334157e-05,
+      "loss": 0.7843,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999797920903076e-05,
+      "loss": 0.6236,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999762837865202e-05,
+      "loss": 0.644,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999972494823038e-05,
+      "loss": 0.6497,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999684252009243e-05,
+      "loss": 0.6422,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999640749213215e-05,
+      "loss": 0.622,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999594439854504e-05,
+      "loss": 0.6153,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999954532394611e-05,
+      "loss": 0.6442,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999949340150182e-05,
+      "loss": 0.62,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999438672536202e-05,
+      "loss": 0.625,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999381137064617e-05,
+      "loss": 0.5887,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999320795103215e-05,
+      "loss": 0.5998,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999257646668936e-05,
+      "loss": 0.5878,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999191691779494e-05,
+      "loss": 0.5898,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999912293045341e-05,
+      "loss": 0.6015,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999905136270998e-05,
+      "loss": 0.5943,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998976988569283e-05,
+      "loss": 0.6411,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998899808052203e-05,
+      "loss": 0.6758,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998819821180398e-05,
+      "loss": 0.6665,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998737027976323e-05,
+      "loss": 0.5868,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998651428463205e-05,
+      "loss": 0.6656,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998563022665078e-05,
+      "loss": 0.5947,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999847181060675e-05,
+      "loss": 0.6319,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999837779231382e-05,
+      "loss": 0.5821,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999828096781268e-05,
+      "loss": 0.6216,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998181337130503e-05,
+      "loss": 0.6095,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998078900295254e-05,
+      "loss": 0.5949,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999797365733568e-05,
+      "loss": 0.6879,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999786560828132e-05,
+      "loss": 0.599,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99977547531625e-05,
+      "loss": 0.5634,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999764109201034e-05,
+      "loss": 0.6556,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999752462485673e-05,
+      "loss": 0.576,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997405351734365e-05,
+      "loss": 0.5823,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999728327267672e-05,
+      "loss": 0.5989,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997158387718057e-05,
+      "loss": 0.5854,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997030696893427e-05,
+      "loss": 0.6511,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996900200238668e-05,
+      "loss": 0.6048,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996766897790412e-05,
+      "loss": 0.5831,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996630789586065e-05,
+      "loss": 0.565,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996491875663833e-05,
+      "loss": 0.5952,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996350156062697e-05,
+      "loss": 0.5649,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999620563082244e-05,
+      "loss": 0.6088,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999605829998363e-05,
+      "loss": 0.5991,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995908163587607e-05,
+      "loss": 0.621,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999575522167651e-05,
+      "loss": 0.6172,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995599474293272e-05,
+      "loss": 0.6655,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9995440921481605e-05,
+      "loss": 0.6356,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9995279563286004e-05,
+      "loss": 0.5898,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999511539975176e-05,
+      "loss": 0.5648,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9994948430924944e-05,
+      "loss": 0.629,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9994778656852428e-05,
+      "loss": 0.6588,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999460607758185e-05,
+      "loss": 0.5803,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9994430693161662e-05,
+      "loss": 0.5956,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9994252503641076e-05,
+      "loss": 0.594,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9994071509070104e-05,
+      "loss": 0.5705,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9993887709499553e-05,
+      "loss": 0.6173,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9993701104981003e-05,
+      "loss": 0.6115,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999351169556683e-05,
+      "loss": 0.6033,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9993319481310195e-05,
+      "loss": 0.6066,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9993124462265045e-05,
+      "loss": 0.5697,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9992926638486118e-05,
+      "loss": 0.6293,
+      "step": 81
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9992726010028935e-05,
+      "loss": 0.6349,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9992522576949803e-05,
+      "loss": 0.5912,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999231633930582e-05,
+      "loss": 0.6073,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9992107297154872e-05,
+      "loss": 0.5638,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999189545055563e-05,
+      "loss": 0.587,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999168079956755e-05,
+      "loss": 0.6084,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999146334425088e-05,
+      "loss": 0.5811,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999124308466665e-05,
+      "loss": 0.5603,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9991020020876676e-05,
+      "loss": 0.6015,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9990794152943574e-05,
+      "loss": 0.63,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9990565480930734e-05,
+      "loss": 0.6245,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999033400490233e-05,
+      "loss": 0.626,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9990099724923337e-05,
+      "loss": 0.5502,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9989862641059504e-05,
+      "loss": 0.6085,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998962275337738e-05,
+      "loss": 0.6174,
+      "step": 96
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998938006194429e-05,
+      "loss": 0.6137,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9989134566828344e-05,
+      "loss": 0.5377,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9988886268098446e-05,
+      "loss": 0.6077,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9988635165824293e-05,
+      "loss": 0.5585,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998838126007635e-05,
+      "loss": 0.6025,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998812455092589e-05,
+      "loss": 0.6041,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9987865038444955e-05,
+      "loss": 0.5728,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9987602722706387e-05,
+      "loss": 0.5588,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9987337603783806e-05,
+      "loss": 0.5903,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998706968175162e-05,
+      "loss": 0.5814,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9986798956685033e-05,
+      "loss": 0.6156,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9986525428660018e-05,
+      "loss": 0.5985,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998624909775335e-05,
+      "loss": 0.6005,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998596996404259e-05,
+      "loss": 0.576,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998568802760608e-05,
+      "loss": 0.5777,
+      "step": 111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.998540328852294e-05,
+      "loss": 0.6087,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99851157468731e-05,
+      "loss": 0.5819,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9984825402737262e-05,
+      "loss": 0.6039,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9984532256196905e-05,
+      "loss": 0.6003,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9984236307334313e-05,
+      "loss": 0.6095,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.998393755623255e-05,
+      "loss": 0.6607,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9983636002975462e-05,
+      "loss": 0.6127,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9983331647647687e-05,
+      "loss": 0.6099,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9983024490334645e-05,
+      "loss": 0.6032,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.998271453112255e-05,
+      "loss": 0.5811,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.998240177009839e-05,
+      "loss": 0.6107,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.998208620734995e-05,
+      "loss": 0.5947,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99817678429658e-05,
+      "loss": 0.5572,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.998144667703529e-05,
+      "loss": 0.5635,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9981122709648558e-05,
+      "loss": 0.5965,
+      "step": 126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9980795940896544e-05,
+      "loss": 0.6202,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9980466370870947e-05,
+      "loss": 0.556,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9980133999664272e-05,
+      "loss": 0.6325,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99797988273698e-05,
+      "loss": 0.6009,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997946085408161e-05,
+      "loss": 0.6068,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9979120079894558e-05,
+      "loss": 0.5491,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9978776504904282e-05,
+      "loss": 0.5876,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997843012920722e-05,
+      "loss": 0.5731,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997808095290058e-05,
+      "loss": 0.5959,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9977728976082367e-05,
+      "loss": 0.5916,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9977374198851374e-05,
+      "loss": 0.5775,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9977016621307167e-05,
+      "loss": 0.6175,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9976656243550115e-05,
+      "loss": 0.6082,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9976293065681355e-05,
+      "loss": 0.5714,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9975927087802822e-05,
+      "loss": 0.5886,
+      "step": 141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975558310017238e-05,
+      "loss": 0.5499,
+      "step": 142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975186732428102e-05,
+      "loss": 0.5873,
+      "step": 143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99748123551397e-05,
+      "loss": 0.588,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974435178257114e-05,
+      "loss": 0.5705,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99740552018862e-05,
+      "loss": 0.5878,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997367242613361e-05,
+      "loss": 0.5783,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997328685110677e-05,
+      "loss": 0.6105,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9972898476913906e-05,
+      "loss": 0.5822,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997250730366401e-05,
+      "loss": 0.5816,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9972113331466883e-05,
+      "loss": 0.6001,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997171656043309e-05,
+      "loss": 0.567,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9971316990673997e-05,
+      "loss": 0.5925,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997091462230175e-05,
+      "loss": 0.5198,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997050945542928e-05,
+      "loss": 0.6043,
+      "step": 155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99701014901703e-05,
+      "loss": 0.5603,
+      "step": 156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996969072663931e-05,
+      "loss": 0.544,
+      "step": 157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969277164951612e-05,
+      "loss": 0.6085,
+      "step": 158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968860805223264e-05,
+      "loss": 0.5777,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968441647571124e-05,
+      "loss": 0.5832,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996801969211285e-05,
+      "loss": 0.5975,
+      "step": 161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967594938966856e-05,
+      "loss": 0.5982,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967167388252358e-05,
+      "loss": 0.5485,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966737040089362e-05,
+      "loss": 0.5663,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966303894598645e-05,
+      "loss": 0.5549,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965867951901783e-05,
+      "loss": 0.5951,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996542921212113e-05,
+      "loss": 0.5974,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996498767537982e-05,
+      "loss": 0.597,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964543341801778e-05,
+      "loss": 0.5683,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996409621151172e-05,
+      "loss": 0.5452,
+      "step": 170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9963646284635134e-05,
+      "loss": 0.6023,
+      "step": 171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99631935612983e-05,
+      "loss": 0.598,
+      "step": 172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9962738041628286e-05,
+      "loss": 0.6609,
+      "step": 173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996227972575294e-05,
+      "loss": 0.5362,
+      "step": 174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961818613800892e-05,
+      "loss": 0.6304,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9961354705901567e-05,
+      "loss": 0.573,
+      "step": 176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996088800218516e-05,
+      "loss": 0.585,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960418502782663e-05,
+      "loss": 0.5762,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995994620782585e-05,
+      "loss": 0.6077,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995947111744728e-05,
+      "loss": 0.5582,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958993231780294e-05,
+      "loss": 0.5809,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958512550959013e-05,
+      "loss": 0.5751,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995802907511835e-05,
+      "loss": 0.6164,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957542804394008e-05,
+      "loss": 0.5551,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995705373892246e-05,
+      "loss": 0.5957,
+      "step": 185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956561878840967e-05,
+      "loss": 0.5508,
+      "step": 186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9956067224287585e-05,
+      "loss": 0.5901,
+      "step": 187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955569775401144e-05,
+      "loss": 0.5543,
+      "step": 188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9955069532321257e-05,
+      "loss": 0.6488,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9954566495188333e-05,
+      "loss": 0.5791,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9954060664143555e-05,
+      "loss": 0.5738,
+      "step": 191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995355203932889e-05,
+      "loss": 0.6277,
+      "step": 192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9953040620887093e-05,
+      "loss": 0.6511,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9952526408961703e-05,
+      "loss": 0.6071,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995200940369704e-05,
+      "loss": 0.5688,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9951489605238214e-05,
+      "loss": 0.5631,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.995096701373111e-05,
+      "loss": 0.5913,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9950441629322407e-05,
+      "loss": 0.5427,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9949913452159556e-05,
+      "loss": 0.5748,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9949382482390803e-05,
+      "loss": 0.5769,
+      "step": 200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9948848720165178e-05,
+      "loss": 0.5554,
+      "step": 201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.994831216563248e-05,
+      "loss": 0.6145,
+      "step": 202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9947772818943308e-05,
+      "loss": 0.5717,
+      "step": 203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9947230680249036e-05,
+      "loss": 0.5828,
+      "step": 204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.994668574970183e-05,
+      "loss": 0.5667,
+      "step": 205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9946138027454624e-05,
+      "loss": 0.5935,
+      "step": 206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.994558751366115e-05,
+      "loss": 0.5763,
+      "step": 207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9945034208475922e-05,
+      "loss": 0.5388,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.994447811205423e-05,
+      "loss": 0.5679,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9943919224552154e-05,
+      "loss": 0.5664,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9943357546126548e-05,
+      "loss": 0.6093,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9942793076935067e-05,
+      "loss": 0.6113,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9942225817136132e-05,
+      "loss": 0.5914,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9941655766888956e-05,
+      "loss": 0.549,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.994108292635353e-05,
+      "loss": 0.5554,
+      "step": 215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9940507295690632e-05,
+      "loss": 0.5492,
+      "step": 216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9939928875061825e-05,
+      "loss": 0.5747,
+      "step": 217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9939347664629453e-05,
+      "loss": 0.5847,
+      "step": 218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9938763664556634e-05,
+      "loss": 0.6064,
+      "step": 219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9938176875007284e-05,
+      "loss": 0.5568,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9937587296146095e-05,
+      "loss": 0.5476,
+      "step": 221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9936994928138543e-05,
+      "loss": 0.5761,
+      "step": 222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.993639977115088e-05,
+      "loss": 0.5651,
+      "step": 223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.993580182535015e-05,
+      "loss": 0.5788,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9935201090904177e-05,
+      "loss": 0.5649,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9934597567981567e-05,
+      "loss": 0.564,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9933991256751707e-05,
+      "loss": 0.5939,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.993338215738477e-05,
+      "loss": 0.5858,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9932770270051706e-05,
+      "loss": 0.5521,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.993215559492426e-05,
+      "loss": 0.5711,
+      "step": 230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.993153813217494e-05,
+      "loss": 0.5948,
+      "step": 231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9930917881977053e-05,
+      "loss": 0.5852,
+      "step": 232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9930294844504677e-05,
+      "loss": 0.6058,
+      "step": 233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9929669019932686e-05,
+      "loss": 0.603,
+      "step": 234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.992904040843672e-05,
+      "loss": 0.5764,
+      "step": 235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9928409010193213e-05,
+      "loss": 0.5627,
+      "step": 236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.992777482537938e-05,
+      "loss": 0.5668,
+      "step": 237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.992713785417321e-05,
+      "loss": 0.5348,
+      "step": 238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.992649809675348e-05,
+      "loss": 0.59,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9925855553299755e-05,
+      "loss": 0.5641,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9925210223992365e-05,
+      "loss": 0.5417,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.992456210901244e-05,
+      "loss": 0.5635,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.992391120854188e-05,
+      "loss": 0.5527,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9923257522763373e-05,
+      "loss": 0.5966,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9922601051860386e-05,
+      "loss": 0.5538,
+      "step": 245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9921941796017168e-05,
+      "loss": 0.6451,
+      "step": 246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.992127975541875e-05,
+      "loss": 0.5891,
+      "step": 247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9920614930250945e-05,
+      "loss": 0.5767,
+      "step": 248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9919947320700346e-05,
+      "loss": 0.573,
+      "step": 249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.991927692695433e-05,
+      "loss": 0.5795,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9918603749201053e-05,
+      "loss": 0.5831,
+      "step": 251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9917927787629454e-05,
+      "loss": 0.5439,
+      "step": 252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9917249042429253e-05,
+      "loss": 0.5367,
+      "step": 253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9916567513790946e-05,
+      "loss": 0.6075,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9915883201905824e-05,
+      "loss": 0.5493,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9915196106965944e-05,
+      "loss": 0.5906,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.991450622916415e-05,
+      "loss": 0.5685,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9913813568694076e-05,
+      "loss": 0.583,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9913118125750115e-05,
+      "loss": 0.5709,
+      "step": 259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9912419900527467e-05,
+      "loss": 0.5741,
+      "step": 260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9911718893222088e-05,
+      "loss": 0.5701,
+      "step": 261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.991101510403074e-05,
+      "loss": 0.5916,
+      "step": 262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9910308533150946e-05,
+      "loss": 0.5826,
+      "step": 263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9909599180781016e-05,
+      "loss": 0.5845,
+      "step": 264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9908887047120046e-05,
+      "loss": 0.5835,
+      "step": 265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99081721323679e-05,
+      "loss": 0.611,
+      "step": 266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9907454436725237e-05,
+      "loss": 0.5917,
+      "step": 267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.990673396039349e-05,
+      "loss": 0.5403,
+      "step": 268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.990601070357487e-05,
+      "loss": 0.5883,
+      "step": 269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9905284666472374e-05,
+      "loss": 0.5528,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.990455584928977e-05,
+      "loss": 0.5626,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.990382425223162e-05,
+      "loss": 0.5666,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9903089875503252e-05,
+      "loss": 0.5835,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9902352719310784e-05,
+      "loss": 0.5518,
+      "step": 274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9901612783861114e-05,
+      "loss": 0.6029,
+      "step": 275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9900870069361914e-05,
+      "loss": 0.5343,
+      "step": 276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9900124576021637e-05,
+      "loss": 0.5709,
+      "step": 277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9899376304049516e-05,
+      "loss": 0.5802,
+      "step": 278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9898625253655574e-05,
+      "loss": 0.5744,
+      "step": 279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9897871425050598e-05,
+      "loss": 0.6037,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.989711481844617e-05,
+      "loss": 0.568,
+      "step": 281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9896355434054636e-05,
+      "loss": 0.5699,
+      "step": 282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9895593272089135e-05,
+      "loss": 0.5644,
+      "step": 283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9894828332763573e-05,
+      "loss": 0.5546,
+      "step": 284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.989406061629265e-05,
+      "loss": 0.5781,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.989329012289184e-05,
+      "loss": 0.5749,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.989251685277739e-05,
+      "loss": 0.5518,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.989174080616633e-05,
+      "loss": 0.6277,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9890961983276472e-05,
+      "loss": 0.569,
+      "step": 289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9890180384326404e-05,
+      "loss": 0.5386,
+      "step": 290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9889396009535496e-05,
+      "loss": 0.5793,
+      "step": 291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9888608859123895e-05,
+      "loss": 0.5583,
+      "step": 292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9887818933312532e-05,
+      "loss": 0.5799,
+      "step": 293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9887026232323105e-05,
+      "loss": 0.5621,
+      "step": 294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9886230756378102e-05,
+      "loss": 0.5738,
+      "step": 295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9885432505700786e-05,
+      "loss": 0.5704,
+      "step": 296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.98846314805152e-05,
+      "loss": 0.6115,
+      "step": 297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9883827681046167e-05,
+      "loss": 0.5488,
+      "step": 298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.988302110751928e-05,
+      "loss": 0.5473,
+      "step": 299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9882211760160924e-05,
+      "loss": 0.5707,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9881399639198248e-05,
+      "loss": 0.5873,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9880584744859195e-05,
+      "loss": 0.5659,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9879767077372474e-05,
+      "loss": 0.5966,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9878946636967576e-05,
+      "loss": 0.5478,
+      "step": 304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9878123423874777e-05,
+      "loss": 0.5877,
+      "step": 305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9877297438325115e-05,
+      "loss": 0.597,
+      "step": 306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9876468680550425e-05,
+      "loss": 0.5697,
+      "step": 307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9875637150783307e-05,
+      "loss": 0.5785,
+      "step": 308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9874802849257148e-05,
+      "loss": 0.5755,
+      "step": 309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9873965776206103e-05,
+      "loss": 0.5772,
+      "step": 310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9873125931865113e-05,
+      "loss": 0.5743,
+      "step": 311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9872283316469893e-05,
+      "loss": 0.5606,
+      "step": 312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.987143793025694e-05,
+      "loss": 0.598,
+      "step": 313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9870589773463517e-05,
+      "loss": 0.5784,
+      "step": 314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9869738846327685e-05,
+      "loss": 0.6226,
+      "step": 315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9868885149088262e-05,
+      "loss": 0.571,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9868028681984857e-05,
+      "loss": 0.5919,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9867169445257848e-05,
+      "loss": 0.5657,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.986630743914839e-05,
+      "loss": 0.516,
+      "step": 319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.986544266389843e-05,
+      "loss": 0.568,
+      "step": 320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9864575119750677e-05,
+      "loss": 0.603,
+      "step": 321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9863704806948612e-05,
+      "loss": 0.5999,
+      "step": 322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.986283172573652e-05,
+      "loss": 0.5558,
+      "step": 323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9861955876359428e-05,
+      "loss": 0.5507,
+      "step": 324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.986107725906317e-05,
+      "loss": 0.5849,
+      "step": 325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.986019587409434e-05,
+      "loss": 0.5876,
+      "step": 326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9859311721700306e-05,
+      "loss": 0.5929,
+      "step": 327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9858424802129232e-05,
+      "loss": 0.5592,
+      "step": 328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.985753511563004e-05,
+      "loss": 0.5555,
+      "step": 329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9856642662452437e-05,
+      "loss": 0.6035,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9855747442846903e-05,
+      "loss": 0.5433,
+      "step": 331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9854849457064697e-05,
+      "loss": 0.5967,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.985394870535785e-05,
+      "loss": 0.5547,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9853045187979184e-05,
+      "loss": 0.5452,
+      "step": 334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.985213890518227e-05,
+      "loss": 0.5866,
+      "step": 335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9851229857221483e-05,
+      "loss": 0.5557,
+      "step": 336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9850318044351958e-05,
+      "loss": 0.5449,
+      "step": 337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9849403466829607e-05,
+      "loss": 0.6002,
+      "step": 338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9848486124911125e-05,
+      "loss": 0.5653,
+      "step": 339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.984756601885398e-05,
+      "loss": 0.5496,
+      "step": 340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9846643148916414e-05,
+      "loss": 0.5881,
+      "step": 341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9845717515357443e-05,
+      "loss": 0.5904,
+      "step": 342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9844789118436863e-05,
+      "loss": 0.5717,
+      "step": 343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9843857958415243e-05,
+      "loss": 0.5834,
+      "step": 344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.984292403555393e-05,
+      "loss": 0.564,
+      "step": 345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9841987350115043e-05,
+      "loss": 0.5871,
+      "step": 346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9841047902361475e-05,
+      "loss": 0.5872,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.984010569255691e-05,
+      "loss": 0.5544,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.983916072096578e-05,
+      "loss": 0.5841,
+      "step": 349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9838212987853312e-05,
+      "loss": 0.5427,
+      "step": 350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9837262493485505e-05,
+      "loss": 0.5476,
+      "step": 351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9836309238129127e-05,
+      "loss": 0.539,
+      "step": 352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9835353222051728e-05,
+      "loss": 0.5847,
+      "step": 353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983439444552163e-05,
+      "loss": 0.5792,
+      "step": 354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9833432908807927e-05,
+      "loss": 0.5919,
+      "step": 355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983246861218049e-05,
+      "loss": 0.5298,
+      "step": 356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9831501555909964e-05,
+      "loss": 0.594,
+      "step": 357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9830531740267772e-05,
+      "loss": 0.5643,
+      "step": 358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9829559165526103e-05,
+      "loss": 0.5999,
+      "step": 359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9828583831957935e-05,
+      "loss": 0.5895,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9827605739837004e-05,
+      "loss": 0.5686,
+      "step": 361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982662488943783e-05,
+      "loss": 0.612,
+      "step": 362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9825641281035707e-05,
+      "loss": 0.5482,
+      "step": 363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9824654914906694e-05,
+      "loss": 0.5584,
+      "step": 364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9823665791327635e-05,
+      "loss": 0.5635,
+      "step": 365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9822673910576148e-05,
+      "loss": 0.6189,
+      "step": 366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9821679272930614e-05,
+      "loss": 0.5824,
+      "step": 367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9820681878670197e-05,
+      "loss": 0.5695,
+      "step": 368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.981968172807483e-05,
+      "loss": 0.5872,
+      "step": 369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9818678821425227e-05,
+      "loss": 0.556,
+      "step": 370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9817673159002864e-05,
+      "loss": 0.5743,
+      "step": 371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.981666474109e-05,
+      "loss": 0.5832,
+      "step": 372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.981565356796966e-05,
+      "loss": 0.6243,
+      "step": 373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.981463963992565e-05,
+      "loss": 0.5777,
+      "step": 374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.981362295724255e-05,
+      "loss": 0.6033,
+      "step": 375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98126035202057e-05,
+      "loss": 0.5583,
+      "step": 376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9811581329101224e-05,
+      "loss": 0.578,
+      "step": 377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.981055638421602e-05,
+      "loss": 0.6008,
+      "step": 378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.980952868583775e-05,
+      "loss": 0.5858,
+      "step": 379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.980849823425486e-05,
+      "loss": 0.5801,
+      "step": 380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.980746502975656e-05,
+      "loss": 0.5443,
+      "step": 381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9806429072632835e-05,
+      "loss": 0.5571,
+      "step": 382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9805390363174447e-05,
+      "loss": 0.5739,
+      "step": 383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9804348901672922e-05,
+      "loss": 0.5362,
+      "step": 384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9803304688420568e-05,
+      "loss": 0.529,
+      "step": 385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.980225772371046e-05,
+      "loss": 0.5601,
+      "step": 386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9801208007836442e-05,
+      "loss": 0.5871,
+      "step": 387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9800155541093135e-05,
+      "loss": 0.5552,
+      "step": 388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9799100323775933e-05,
+      "loss": 0.5537,
+      "step": 389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9798042356181e-05,
+      "loss": 0.5929,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.979698163860527e-05,
+      "loss": 0.5586,
+      "step": 391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9795918171346448e-05,
+      "loss": 0.5758,
+      "step": 392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9794851954703024e-05,
+      "loss": 0.5652,
+      "step": 393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9793782988974233e-05,
+      "loss": 0.5536,
+      "step": 394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9792711274460116e-05,
+      "loss": 0.5974,
+      "step": 395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9791636811461454e-05,
+      "loss": 0.543,
+      "step": 396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9790559600279816e-05,
+      "loss": 0.5734,
+      "step": 397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9789479641217538e-05,
+      "loss": 0.5894,
+      "step": 398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9788396934577733e-05,
+      "loss": 0.587,
+      "step": 399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.978731148066428e-05,
+      "loss": 0.5684,
+      "step": 400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9786223279781823e-05,
+      "loss": 0.5832,
+      "step": 401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9785132332235787e-05,
+      "loss": 0.5801,
+      "step": 402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9784038638332363e-05,
+      "loss": 0.5815,
+      "step": 403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9782942198378516e-05,
+      "loss": 0.5693,
+      "step": 404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.978184301268198e-05,
+      "loss": 0.57,
+      "step": 405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.978074108155126e-05,
+      "loss": 0.5642,
+      "step": 406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9779636405295624e-05,
+      "loss": 0.5553,
+      "step": 407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9778528984225126e-05,
+      "loss": 0.5476,
+      "step": 408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977741881865058e-05,
+      "loss": 0.5916,
+      "step": 409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977630590888357e-05,
+      "loss": 0.586,
+      "step": 410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977519025523645e-05,
+      "loss": 0.5014,
+      "step": 411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9774071858022354e-05,
+      "loss": 0.5413,
+      "step": 412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9772950717555174e-05,
+      "loss": 0.567,
+      "step": 413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9771826834149577e-05,
+      "loss": 0.599,
+      "step": 414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9770700208120996e-05,
+      "loss": 0.607,
+      "step": 415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9769570839785638e-05,
+      "loss": 0.5516,
+      "step": 416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9768438729460487e-05,
+      "loss": 0.5413,
+      "step": 417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9767303877463275e-05,
+      "loss": 0.5653,
+      "step": 418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.976616628411253e-05,
+      "loss": 0.5838,
+      "step": 419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9765025949727526e-05,
+      "loss": 0.5575,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9763882874628318e-05,
+      "loss": 0.581,
+      "step": 421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9762737059135735e-05,
+      "loss": 0.5922,
+      "step": 422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.976158850357136e-05,
+      "loss": 0.5656,
+      "step": 423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760437208257564e-05,
+      "loss": 0.5436,
+      "step": 424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975928317351747e-05,
+      "loss": 0.5676,
+      "step": 425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758126399674974e-05,
+      "loss": 0.6124,
+      "step": 426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975696688705475e-05,
+      "loss": 0.5773,
+      "step": 427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9755804635982233e-05,
+      "loss": 0.5855,
+      "step": 428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9754639646783627e-05,
+      "loss": 0.5294,
+      "step": 429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975347191978591e-05,
+      "loss": 0.5977,
+      "step": 430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9752301455316808e-05,
+      "loss": 0.5552,
+      "step": 431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9751128253704848e-05,
+      "loss": 0.5413,
+      "step": 432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97499523152793e-05,
+      "loss": 0.5817,
+      "step": 433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9748773640370212e-05,
+      "loss": 0.5867,
+      "step": 434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97475922293084e-05,
+      "loss": 0.5578,
+      "step": 435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9746408082425443e-05,
+      "loss": 0.5483,
+      "step": 436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9745221200053694e-05,
+      "loss": 0.5564,
+      "step": 437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.974403158252627e-05,
+      "loss": 0.5528,
+      "step": 438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742839230177056e-05,
+      "loss": 0.582,
+      "step": 439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741644143340707e-05,
+      "loss": 0.5711,
+      "step": 440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740446322352643e-05,
+      "loss": 0.6,
+      "step": 441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739245767549048e-05,
+      "loss": 0.5799,
+      "step": 442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738042479266886e-05,
+      "loss": 0.5483,
+      "step": 443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973683645784387e-05,
+      "loss": 0.5807,
+      "step": 444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735627703618494e-05,
+      "loss": 0.596,
+      "step": 445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973441621693002e-05,
+      "loss": 0.6182,
+      "step": 446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973320199811846e-05,
+      "loss": 0.5435,
+      "step": 447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9731985047524612e-05,
+      "loss": 0.569,
+      "step": 448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9730765365490033e-05,
+      "loss": 0.5865,
+      "step": 449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9729542952357045e-05,
+      "loss": 0.5882,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.972831780846874e-05,
+      "loss": 0.5532,
+      "step": 451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.972708993416897e-05,
+      "loss": 0.5525,
+      "step": 452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725859329802363e-05,
+      "loss": 0.6435,
+      "step": 453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724625995714307e-05,
+      "loss": 0.5676,
+      "step": 454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723389932250955e-05,
+      "loss": 0.5593,
+      "step": 455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722151139759232e-05,
+      "loss": 0.5657,
+      "step": 456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720909618586824e-05,
+      "loss": 0.5595,
+      "step": 457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971966536908218e-05,
+      "loss": 0.5849,
+      "step": 458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718418391594526e-05,
+      "loss": 0.5468,
+      "step": 459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717168686473845e-05,
+      "loss": 0.6109,
+      "step": 460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715916254070883e-05,
+      "loss": 0.5672,
+      "step": 461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971466109473716e-05,
+      "loss": 0.5713,
+      "step": 462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713403208824957e-05,
+      "loss": 0.5531,
+      "step": 463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712142596687314e-05,
+      "loss": 0.6088,
+      "step": 464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710879258678045e-05,
+      "loss": 0.542,
+      "step": 465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709613195151736e-05,
+      "loss": 0.579,
+      "step": 466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708344406463714e-05,
+      "loss": 0.5614,
+      "step": 467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9707072892970095e-05,
+      "loss": 0.5719,
+      "step": 468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705798655027747e-05,
+      "loss": 0.592,
+      "step": 469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704521692994305e-05,
+      "loss": 0.5665,
+      "step": 470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703242007228172e-05,
+      "loss": 0.5465,
+      "step": 471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970195959808851e-05,
+      "loss": 0.541,
+      "step": 472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970067446593525e-05,
+      "loss": 0.5469,
+      "step": 473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699386611129082e-05,
+      "loss": 0.5905,
+      "step": 474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969809603403147e-05,
+      "loss": 0.6295,
+      "step": 475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696802735004627e-05,
+      "loss": 0.5631,
+      "step": 476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695506714411543e-05,
+      "loss": 0.5421,
+      "step": 477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694207972615967e-05,
+      "loss": 0.5511,
+      "step": 478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692906509982416e-05,
+      "loss": 0.5908,
+      "step": 479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969160232687616e-05,
+      "loss": 0.5496,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690295423663244e-05,
+      "loss": 0.5526,
+      "step": 481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968898580071047e-05,
+      "loss": 0.592,
+      "step": 482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9687673458385406e-05,
+      "loss": 0.5887,
+      "step": 483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968635839705638e-05,
+      "loss": 0.5629,
+      "step": 484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968504061709249e-05,
+      "loss": 0.5538,
+      "step": 485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968372011886359e-05,
+      "loss": 0.5865,
+      "step": 486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9682396902740294e-05,
+      "loss": 0.5421,
+      "step": 487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968107096909399e-05,
+      "loss": 0.5903,
+      "step": 488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679742318296827e-05,
+      "loss": 0.6114,
+      "step": 489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96784109507217e-05,
+      "loss": 0.5873,
+      "step": 490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677076866742295e-05,
+      "loss": 0.5646,
+      "step": 491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675740066733035e-05,
+      "loss": 0.6312,
+      "step": 492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967440055106911e-05,
+      "loss": 0.5399,
+      "step": 493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967305832012649e-05,
+      "loss": 0.5604,
+      "step": 494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671713374281883e-05,
+      "loss": 0.5732,
+      "step": 495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670365713912776e-05,
+      "loss": 0.5533,
+      "step": 496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966901533939741e-05,
+      "loss": 0.5677,
+      "step": 497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9667662251114793e-05,
+      "loss": 0.5533,
+      "step": 498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9666306449444682e-05,
+      "loss": 0.5834,
+      "step": 499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9664947934767614e-05,
+      "loss": 0.587,
+      "step": 500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9663586707464876e-05,
+      "loss": 0.5544,
+      "step": 501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9662222767918518e-05,
+      "loss": 0.5128,
+      "step": 502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9660856116511354e-05,
+      "loss": 0.6291,
+      "step": 503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9659486753626954e-05,
+      "loss": 0.6063,
+      "step": 504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965811467964965e-05,
+      "loss": 0.5799,
+      "step": 505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9656739894964544e-05,
+      "loss": 0.5628,
+      "step": 506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9655362399957486e-05,
+      "loss": 0.5554,
+      "step": 507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9653982195015095e-05,
+      "loss": 0.5484,
+      "step": 508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965259928052475e-05,
+      "loss": 0.5388,
+      "step": 509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965121365687458e-05,
+      "loss": 0.575,
+      "step": 510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964982532445349e-05,
+      "loss": 0.5393,
+      "step": 511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648434283651144e-05,
+      "loss": 0.5949,
+      "step": 512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9647040534857948e-05,
+      "loss": 0.5473,
+      "step": 513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9645644078465088e-05,
+      "loss": 0.541,
+      "step": 514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9644244914864502e-05,
+      "loss": 0.5781,
+      "step": 515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9642843044448886e-05,
+      "loss": 0.642,
+      "step": 516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9641438467611696e-05,
+      "loss": 0.5579,
+      "step": 517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9640031184747152e-05,
+      "loss": 0.5438,
+      "step": 518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9638621196250232e-05,
+      "loss": 0.5216,
+      "step": 519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9637208502516673e-05,
+      "loss": 0.5954,
+      "step": 520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9635793103942964e-05,
+      "loss": 0.5758,
+      "step": 521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9634375000926367e-05,
+      "loss": 0.5904,
+      "step": 522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9632954193864894e-05,
+      "loss": 0.5624,
+      "step": 523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9631530683157316e-05,
+      "loss": 0.548,
+      "step": 524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9630104469203165e-05,
+      "loss": 0.5552,
+      "step": 525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9628675552402734e-05,
+      "loss": 0.5761,
+      "step": 526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962724393315707e-05,
+      "loss": 0.5475,
+      "step": 527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9625809611867977e-05,
+      "loss": 0.5524,
+      "step": 528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9624372588938025e-05,
+      "loss": 0.5618,
+      "step": 529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9622932864770538e-05,
+      "loss": 0.535,
+      "step": 530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9621490439769594e-05,
+      "loss": 0.5327,
+      "step": 531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9620045314340037e-05,
+      "loss": 0.5834,
+      "step": 532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9618597488887462e-05,
+      "loss": 0.5423,
+      "step": 533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9617146963818233e-05,
+      "loss": 0.5831,
+      "step": 534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9615693739539452e-05,
+      "loss": 0.5654,
+      "step": 535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9614237816459e-05,
+      "loss": 0.5795,
+      "step": 536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.96127791949855e-05,
+      "loss": 0.5691,
+      "step": 537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9611317875528338e-05,
+      "loss": 0.537,
+      "step": 538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9609853858497655e-05,
+      "loss": 0.5723,
+      "step": 539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9608387144304363e-05,
+      "loss": 0.5443,
+      "step": 540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9606917733360106e-05,
+      "loss": 0.5524,
+      "step": 541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9605445626077305e-05,
+      "loss": 0.5858,
+      "step": 542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9603970822869125e-05,
+      "loss": 0.573,
+      "step": 543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9602493324149502e-05,
+      "loss": 0.5726,
+      "step": 544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.960101313033312e-05,
+      "loss": 0.5576,
+      "step": 545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9599530241835407e-05,
+      "loss": 0.5967,
+      "step": 546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9598044659072573e-05,
+      "loss": 0.5562,
+      "step": 547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9596556382461567e-05,
+      "loss": 0.5736,
+      "step": 548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9595065412420097e-05,
+      "loss": 0.5825,
+      "step": 549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.959357174936663e-05,
+      "loss": 0.5675,
+      "step": 550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9592075393720385e-05,
+      "loss": 0.5587,
+      "step": 551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.959057634590134e-05,
+      "loss": 0.5538,
+      "step": 552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9589074606330227e-05,
+      "loss": 0.5589,
+      "step": 553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9587570175428533e-05,
+      "loss": 0.5796,
+      "step": 554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95860630536185e-05,
+      "loss": 0.5732,
+      "step": 555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9584553241323133e-05,
+      "loss": 0.5658,
+      "step": 556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9583040738966178e-05,
+      "loss": 0.5565,
+      "step": 557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.958152554697215e-05,
+      "loss": 0.5791,
+      "step": 558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9580007665766306e-05,
+      "loss": 0.6232,
+      "step": 559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9578487095774666e-05,
+      "loss": 0.5555,
+      "step": 560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9576963837424e-05,
+      "loss": 0.5638,
+      "step": 561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9575437891141843e-05,
+      "loss": 0.5563,
+      "step": 562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9573909257356474e-05,
+      "loss": 0.5773,
+      "step": 563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9572377936496926e-05,
+      "loss": 0.5515,
+      "step": 564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.957084392899299e-05,
+      "loss": 0.5485,
+      "step": 565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9569307235275214e-05,
+      "loss": 0.6222,
+      "step": 566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9567767855774892e-05,
+      "loss": 0.5472,
+      "step": 567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.956622579092408e-05,
+      "loss": 0.6121,
+      "step": 568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9564681041155576e-05,
+      "loss": 0.5712,
+      "step": 569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.956313360690295e-05,
+      "loss": 0.623,
+      "step": 570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.956158348860051e-05,
+      "loss": 0.5701,
+      "step": 571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9560030686683316e-05,
+      "loss": 0.5747,
+      "step": 572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9558475201587198e-05,
+      "loss": 0.6337,
+      "step": 573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.955691703374872e-05,
+      "loss": 0.5592,
+      "step": 574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.955535618360521e-05,
+      "loss": 0.6027,
+      "step": 575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.955379265159475e-05,
+      "loss": 0.5717,
+      "step": 576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.955222643815616e-05,
+      "loss": 0.6055,
+      "step": 577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9550657543729038e-05,
+      "loss": 0.5543,
+      "step": 578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9549085968753707e-05,
+      "loss": 0.5172,
+      "step": 579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9547511713671264e-05,
+      "loss": 0.5725,
+      "step": 580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9545934778923545e-05,
+      "loss": 0.5769,
+      "step": 581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9544355164953143e-05,
+      "loss": 0.555,
+      "step": 582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9542772872203404e-05,
+      "loss": 0.5751,
+      "step": 583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9541187901118423e-05,
+      "loss": 0.6036,
+      "step": 584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.953960025214305e-05,
+      "loss": 0.6035,
+      "step": 585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.953800992572288e-05,
+      "loss": 0.5255,
+      "step": 586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.953641692230427e-05,
+      "loss": 0.5611,
+      "step": 587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.953482124233432e-05,
+      "loss": 0.5736,
+      "step": 588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.953322288626088e-05,
+      "loss": 0.5818,
+      "step": 589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9531621854532562e-05,
+      "loss": 0.5095,
+      "step": 590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9530018147598717e-05,
+      "loss": 0.5513,
+      "step": 591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9528411765909452e-05,
+      "loss": 0.5914,
+      "step": 592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9526802709915623e-05,
+      "loss": 0.5548,
+      "step": 593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9525190980068843e-05,
+      "loss": 0.5366,
+      "step": 594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9523576576821463e-05,
+      "loss": 0.5367,
+      "step": 595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.95219595006266e-05,
+      "loss": 0.5409,
+      "step": 596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9520339751938103e-05,
+      "loss": 0.5539,
+      "step": 597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9518717331210594e-05,
+      "loss": 0.5785,
+      "step": 598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.951709223889942e-05,
+      "loss": 0.5517,
+      "step": 599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9515464475460692e-05,
+      "loss": 0.5972,
+      "step": 600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9513834041351277e-05,
+      "loss": 0.5328,
+      "step": 601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9512200937028767e-05,
+      "loss": 0.6033,
+      "step": 602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.5525,
+      "step": 603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9508926719578683e-05,
+      "loss": 0.5965,
+      "step": 604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9507285607370065e-05,
+      "loss": 0.5734,
+      "step": 605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9505641826786282e-05,
+      "loss": 0.56,
+      "step": 606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9503995378288697e-05,
+      "loss": 0.5585,
+      "step": 607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.950234626233941e-05,
+      "loss": 0.5904,
+      "step": 608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9500694479401266e-05,
+      "loss": 0.5417,
+      "step": 609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.949904002993787e-05,
+      "loss": 0.5612,
+      "step": 610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.949738291441357e-05,
+      "loss": 0.5519,
+      "step": 611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9495723133293465e-05,
+      "loss": 0.5326,
+      "step": 612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9494060687043393e-05,
+      "loss": 0.6025,
+      "step": 613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9492395576129953e-05,
+      "loss": 0.5566,
+      "step": 614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9490727801020485e-05,
+      "loss": 0.5631,
+      "step": 615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9489057362183074e-05,
+      "loss": 0.5973,
+      "step": 616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9487384260086557e-05,
+      "loss": 0.5548,
+      "step": 617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9485708495200517e-05,
+      "loss": 0.5686,
+      "step": 618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.948403006799529e-05,
+      "loss": 0.5612,
+      "step": 619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9482348978941947e-05,
+      "loss": 0.6014,
+      "step": 620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9480665228512314e-05,
+      "loss": 0.5584,
+      "step": 621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9478978817178967e-05,
+      "loss": 0.5733,
+      "step": 622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9477289745415225e-05,
+      "loss": 0.5554,
+      "step": 623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.947559801369515e-05,
+      "loss": 0.5889,
+      "step": 624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9473903622493554e-05,
+      "loss": 0.517,
+      "step": 625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9472206572286003e-05,
+      "loss": 0.5979,
+      "step": 626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.947050686354879e-05,
+      "loss": 0.5714,
+      "step": 627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.946880449675898e-05,
+      "loss": 0.5673,
+      "step": 628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9467099472394357e-05,
+      "loss": 0.5075,
+      "step": 629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.946539179093347e-05,
+      "loss": 0.5806,
+      "step": 630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9463681452855607e-05,
+      "loss": 0.5948,
+      "step": 631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9461968458640802e-05,
+      "loss": 0.5642,
+      "step": 632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9460252808769836e-05,
+      "loss": 0.5767,
+      "step": 633
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9458534503724236e-05,
+      "loss": 0.5629,
+      "step": 634
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.945681354398627e-05,
+      "loss": 0.5585,
+      "step": 635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.945508993003895e-05,
+      "loss": 0.5811,
+      "step": 636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.945336366236604e-05,
+      "loss": 0.5724,
+      "step": 637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9451634741452047e-05,
+      "loss": 0.5595,
+      "step": 638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.944990316778222e-05,
+      "loss": 0.5611,
+      "step": 639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.944816894184255e-05,
+      "loss": 0.5891,
+      "step": 640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.944643206411978e-05,
+      "loss": 0.5402,
+      "step": 641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.944469253510139e-05,
+      "loss": 0.5984,
+      "step": 642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9442950355275613e-05,
+      "loss": 0.5142,
+      "step": 643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.944120552513141e-05,
+      "loss": 0.5582,
+      "step": 644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.943945804515851e-05,
+      "loss": 0.5764,
+      "step": 645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9437707915847355e-05,
+      "loss": 0.558,
+      "step": 646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.943595513768916e-05,
+      "loss": 0.5538,
+      "step": 647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.943419971117587e-05,
+      "loss": 0.5488,
+      "step": 648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.943244163680017e-05,
+      "loss": 0.5556,
+      "step": 649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9430680915055492e-05,
+      "loss": 0.5655,
+      "step": 650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9428917546436014e-05,
+      "loss": 0.5852,
+      "step": 651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9427151531436652e-05,
+      "loss": 0.5719,
+      "step": 652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.942538287055307e-05,
+      "loss": 0.5358,
+      "step": 653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.942361156428167e-05,
+      "loss": 0.5587,
+      "step": 654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9421837613119597e-05,
+      "loss": 0.5437,
+      "step": 655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9420061017564743e-05,
+      "loss": 0.5874,
+      "step": 656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.941828177811573e-05,
+      "loss": 0.5545,
+      "step": 657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.941649989527194e-05,
+      "loss": 0.5799,
+      "step": 658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9414715369533485e-05,
+      "loss": 0.5957,
+      "step": 659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.941292820140122e-05,
+      "loss": 0.5477,
+      "step": 660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9411138391376742e-05,
+      "loss": 0.5512,
+      "step": 661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9409345939962393e-05,
+      "loss": 0.541,
+      "step": 662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9407550847661256e-05,
+      "loss": 0.5706,
+      "step": 663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9405753114977145e-05,
+      "loss": 0.5745,
+      "step": 664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.940395274241463e-05,
+      "loss": 0.6005,
+      "step": 665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.940214973047901e-05,
+      "loss": 0.5466,
+      "step": 666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9400344079676335e-05,
+      "loss": 0.562,
+      "step": 667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.939853579051339e-05,
+      "loss": 0.5724,
+      "step": 668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9396724863497695e-05,
+      "loss": 0.5713,
+      "step": 669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9394911299137522e-05,
+      "loss": 0.6093,
+      "step": 670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9393095097941873e-05,
+      "loss": 0.5525,
+      "step": 671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.93912762604205e-05,
+      "loss": 0.5629,
+      "step": 672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9389454787083884e-05,
+      "loss": 0.604,
+      "step": 673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.938763067844326e-05,
+      "loss": 0.5516,
+      "step": 674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.938580393501058e-05,
+      "loss": 0.5371,
+      "step": 675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9383974557298563e-05,
+      "loss": 0.557,
+      "step": 676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.938214254582065e-05,
+      "loss": 0.5684,
+      "step": 677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9380307901091014e-05,
+      "loss": 0.5569,
+      "step": 678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9378470623624594e-05,
+      "loss": 0.551,
+      "step": 679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9376630713937043e-05,
+      "loss": 0.6101,
+      "step": 680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9374788172544765e-05,
+      "loss": 0.569,
+      "step": 681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9372942999964895e-05,
+      "loss": 0.5669,
+      "step": 682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9371095196715316e-05,
+      "loss": 0.5921,
+      "step": 683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9369244763314645e-05,
+      "loss": 0.5578,
+      "step": 684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9367391700282228e-05,
+      "loss": 0.5318,
+      "step": 685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9365536008138172e-05,
+      "loss": 0.557,
+      "step": 686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.936367768740329e-05,
+      "loss": 0.539,
+      "step": 687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.936181673859916e-05,
+      "loss": 0.6057,
+      "step": 688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.935995316224809e-05,
+      "loss": 0.5804,
+      "step": 689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9358086958873116e-05,
+      "loss": 0.5773,
+      "step": 690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.935621812899802e-05,
+      "loss": 0.5493,
+      "step": 691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9354346673147323e-05,
+      "loss": 0.5728,
+      "step": 692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9352472591846282e-05,
+      "loss": 0.5789,
+      "step": 693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.935059588562088e-05,
+      "loss": 0.5603,
+      "step": 694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9348716554997854e-05,
+      "loss": 0.5734,
+      "step": 695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9346834600504664e-05,
+      "loss": 0.5711,
+      "step": 696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9344950022669507e-05,
+      "loss": 0.5498,
+      "step": 697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9343062822021332e-05,
+      "loss": 0.5745,
+      "step": 698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9341172999089805e-05,
+      "loss": 0.5516,
+      "step": 699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9339280554405336e-05,
+      "loss": 0.5672,
+      "step": 700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9337385488499074e-05,
+      "loss": 0.5404,
+      "step": 701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9335487801902896e-05,
+      "loss": 0.5741,
+      "step": 702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9333587495149418e-05,
+      "loss": 0.5714,
+      "step": 703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9331684568772e-05,
+      "loss": 0.5669,
+      "step": 704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9329779023304724e-05,
+      "loss": 0.5737,
+      "step": 705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.932787085928241e-05,
+      "loss": 0.5828,
+      "step": 706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.932596007724062e-05,
+      "loss": 0.5557,
+      "step": 707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9324046677715644e-05,
+      "loss": 0.5757,
+      "step": 708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9322130661244508e-05,
+      "loss": 0.6158,
+      "step": 709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9320212028364976e-05,
+      "loss": 0.531,
+      "step": 710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9318290779615545e-05,
+      "loss": 0.5411,
+      "step": 711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.931636691553544e-05,
+      "loss": 0.5593,
+      "step": 712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9314440436664626e-05,
+      "loss": 0.5658,
+      "step": 713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9312511343543802e-05,
+      "loss": 0.5591,
+      "step": 714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9310579636714402e-05,
+      "loss": 0.5617,
+      "step": 715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.930864531671859e-05,
+      "loss": 0.5429,
+      "step": 716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9306708384099258e-05,
+      "loss": 0.5369,
+      "step": 717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9304768839400046e-05,
+      "loss": 0.5455,
+      "step": 718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9302826683165315e-05,
+      "loss": 0.5672,
+      "step": 719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9300881915940163e-05,
+      "loss": 0.5997,
+      "step": 720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9298934538270423e-05,
+      "loss": 0.5621,
+      "step": 721
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9296984550702656e-05,
+      "loss": 0.5844,
+      "step": 722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929503195378416e-05,
+      "loss": 0.6057,
+      "step": 723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929307674806296e-05,
+      "loss": 0.5534,
+      "step": 724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.929111893408782e-05,
+      "loss": 0.5135,
+      "step": 725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.928915851240823e-05,
+      "loss": 0.5553,
+      "step": 726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.928719548357442e-05,
+      "loss": 0.5783,
+      "step": 727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9285229848137335e-05,
+      "loss": 0.5537,
+      "step": 728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9283261606648672e-05,
+      "loss": 0.5447,
+      "step": 729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.928129075966085e-05,
+      "loss": 0.5886,
+      "step": 730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9279317307727012e-05,
+      "loss": 0.5832,
+      "step": 731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9277341251401047e-05,
+      "loss": 0.5591,
+      "step": 732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9275362591237564e-05,
+      "loss": 0.5592,
+      "step": 733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.927338132779191e-05,
+      "loss": 0.536,
+      "step": 734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9271397461620154e-05,
+      "loss": 0.5591,
+      "step": 735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9269410993279104e-05,
+      "loss": 0.5325,
+      "step": 736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.926742192332629e-05,
+      "loss": 0.5783,
+      "step": 737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9265430252319984e-05,
+      "loss": 0.5576,
+      "step": 738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9263435980819177e-05,
+      "loss": 0.6244,
+      "step": 739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9261439109383594e-05,
+      "loss": 0.5723,
+      "step": 740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9259439638573688e-05,
+      "loss": 0.5856,
+      "step": 741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.925743756895065e-05,
+      "loss": 0.5135,
+      "step": 742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9255432901076386e-05,
+      "loss": 0.5601,
+      "step": 743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9253425635513543e-05,
+      "loss": 0.5737,
+      "step": 744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.925141577282549e-05,
+      "loss": 0.5594,
+      "step": 745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9249403313576328e-05,
+      "loss": 0.5622,
+      "step": 746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9247388258330892e-05,
+      "loss": 0.5669,
+      "step": 747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9245370607654733e-05,
+      "loss": 0.5544,
+      "step": 748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9243350362114143e-05,
+      "loss": 0.5297,
+      "step": 749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9241327522276133e-05,
+      "loss": 0.5723,
+      "step": 750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9239302088708452e-05,
+      "loss": 0.5341,
+      "step": 751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9237274061979564e-05,
+      "loss": 0.6035,
+      "step": 752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.923524344265867e-05,
+      "loss": 0.5262,
+      "step": 753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9233210231315697e-05,
+      "loss": 0.5501,
+      "step": 754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.92311744285213e-05,
+      "loss": 0.5783,
+      "step": 755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9229136034846858e-05,
+      "loss": 0.5909,
+      "step": 756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9227095050864485e-05,
+      "loss": 0.5896,
+      "step": 757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.922505147714701e-05,
+      "loss": 0.5653,
+      "step": 758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9223005314268e-05,
+      "loss": 0.5263,
+      "step": 759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.922095656280174e-05,
+      "loss": 0.5538,
+      "step": 760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.921890522332325e-05,
+      "loss": 0.5532,
+      "step": 761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.921685129640827e-05,
+      "loss": 0.5551,
+      "step": 762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9214794782633264e-05,
+      "loss": 0.5391,
+      "step": 763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.921273568257543e-05,
+      "loss": 0.5658,
+      "step": 764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9210673996812694e-05,
+      "loss": 0.5353,
+      "step": 765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.920860972592369e-05,
+      "loss": 0.549,
+      "step": 766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.92065428704878e-05,
+      "loss": 0.5588,
+      "step": 767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9204473431085114e-05,
+      "loss": 0.5849,
+      "step": 768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9202401408296457e-05,
+      "loss": 0.5869,
+      "step": 769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9200326802703374e-05,
+      "loss": 0.5734,
+      "step": 770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9198249614888138e-05,
+      "loss": 0.5652,
+      "step": 771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919616984543375e-05,
+      "loss": 0.6093,
+      "step": 772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919408749492392e-05,
+      "loss": 0.562,
+      "step": 773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.919200256394311e-05,
+      "loss": 0.5512,
+      "step": 774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9189915053076472e-05,
+      "loss": 0.5685,
+      "step": 775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9187824962909912e-05,
+      "loss": 0.513,
+      "step": 776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.918573229403005e-05,
+      "loss": 0.5912,
+      "step": 777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9183637047024218e-05,
+      "loss": 0.5767,
+      "step": 778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9181539222480484e-05,
+      "loss": 0.5945,
+      "step": 779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9179438820987645e-05,
+      "loss": 0.5604,
+      "step": 780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9177335843135202e-05,
+      "loss": 0.5295,
+      "step": 781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.91752302895134e-05,
+      "loss": 0.5536,
+      "step": 782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9173122160713187e-05,
+      "loss": 0.549,
+      "step": 783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.917101145732625e-05,
+      "loss": 0.5432,
+      "step": 784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9168898179944994e-05,
+      "loss": 0.5829,
+      "step": 785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9166782329162538e-05,
+      "loss": 0.5574,
+      "step": 786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9164663905572734e-05,
+      "loss": 0.5803,
+      "step": 787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9162542909770153e-05,
+      "loss": 0.5806,
+      "step": 788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.916041934235009e-05,
+      "loss": 0.5486,
+      "step": 789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9158293203908552e-05,
+      "loss": 0.5411,
+      "step": 790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9156164495042283e-05,
+      "loss": 0.5524,
+      "step": 791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9154033216348732e-05,
+      "loss": 0.5782,
+      "step": 792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915189936842608e-05,
+      "loss": 0.5995,
+      "step": 793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9149762951873224e-05,
+      "loss": 0.5672,
+      "step": 794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.914762396728979e-05,
+      "loss": 0.5396,
+      "step": 795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9145482415276113e-05,
+      "loss": 0.5665,
+      "step": 796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9143338296433262e-05,
+      "loss": 0.5891,
+      "step": 797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.914119161136301e-05,
+      "loss": 0.5535,
+      "step": 798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9139042360667865e-05,
+      "loss": 0.5399,
+      "step": 799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9136890544951046e-05,
+      "loss": 0.555,
+      "step": 800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.91347361648165e-05,
+      "loss": 0.5976,
+      "step": 801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.913257922086889e-05,
+      "loss": 0.5292,
+      "step": 802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9130419713713588e-05,
+      "loss": 0.5547,
+      "step": 803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9128257643956703e-05,
+      "loss": 0.5403,
+      "step": 804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.912609301220505e-05,
+      "loss": 0.5898,
+      "step": 805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9123925819066172e-05,
+      "loss": 0.5698,
+      "step": 806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9121756065148333e-05,
+      "loss": 0.6113,
+      "step": 807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9119583751060495e-05,
+      "loss": 0.5581,
+      "step": 808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9117408877412366e-05,
+      "loss": 0.5537,
+      "step": 809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9115231444814356e-05,
+      "loss": 0.5646,
+      "step": 810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9113051453877595e-05,
+      "loss": 0.5092,
+      "step": 811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.911086890521394e-05,
+      "loss": 0.5775,
+      "step": 812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.910868379943595e-05,
+      "loss": 0.5368,
+      "step": 813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.910649613715691e-05,
+      "loss": 0.5578,
+      "step": 814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9104305918990832e-05,
+      "loss": 0.5948,
+      "step": 815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9102113145552434e-05,
+      "loss": 0.5245,
+      "step": 816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9099917817457152e-05,
+      "loss": 0.5549,
+      "step": 817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9097719935321137e-05,
+      "loss": 0.5862,
+      "step": 818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.909551949976127e-05,
+      "loss": 0.5024,
+      "step": 819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9093316511395128e-05,
+      "loss": 0.5631,
+      "step": 820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9091110970841024e-05,
+      "loss": 0.5404,
+      "step": 821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9088902878717978e-05,
+      "loss": 0.5944,
+      "step": 822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.908669223564572e-05,
+      "loss": 0.5525,
+      "step": 823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9084479042244714e-05,
+      "loss": 0.6176,
+      "step": 824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.908226329913612e-05,
+      "loss": 0.5513,
+      "step": 825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9080045006941828e-05,
+      "loss": 0.5757,
+      "step": 826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9077824166284434e-05,
+      "loss": 0.579,
+      "step": 827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9075600777787256e-05,
+      "loss": 0.5777,
+      "step": 828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9073374842074323e-05,
+      "loss": 0.5759,
+      "step": 829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9071146359770384e-05,
+      "loss": 0.5795,
+      "step": 830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.906891533150089e-05,
+      "loss": 0.5528,
+      "step": 831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9066681757892026e-05,
+      "loss": 0.5879,
+      "step": 832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9064445639570675e-05,
+      "loss": 0.6238,
+      "step": 833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.906220697716444e-05,
+      "loss": 0.5294,
+      "step": 834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9059965771301644e-05,
+      "loss": 0.5764,
+      "step": 835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.905772202261131e-05,
+      "loss": 0.5574,
+      "step": 836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9055475731723187e-05,
+      "loss": 0.5675,
+      "step": 837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.905322689926773e-05,
+      "loss": 0.5536,
+      "step": 838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.905097552587612e-05,
+      "loss": 0.5418,
+      "step": 839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9048721612180232e-05,
+      "loss": 0.5389,
+      "step": 840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9046465158812666e-05,
+      "loss": 0.5604,
+      "step": 841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9044206166406734e-05,
+      "loss": 0.5755,
+      "step": 842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.904194463559646e-05,
+      "loss": 0.5955,
+      "step": 843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9039680567016576e-05,
+      "loss": 0.5437,
+      "step": 844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9037413961302534e-05,
+      "loss": 0.5514,
+      "step": 845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.903514481909049e-05,
+      "loss": 0.544,
+      "step": 846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9032873141017316e-05,
+      "loss": 0.5387,
+      "step": 847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9030598927720603e-05,
+      "loss": 0.5393,
+      "step": 848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9028322179838636e-05,
+      "loss": 0.518,
+      "step": 849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9026042898010428e-05,
+      "loss": 0.5585,
+      "step": 850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9023761082875695e-05,
+      "loss": 0.6034,
+      "step": 851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9021476735074865e-05,
+      "loss": 0.5387,
+      "step": 852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9019189855249078e-05,
+      "loss": 0.5533,
+      "step": 853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9016900444040184e-05,
+      "loss": 0.5328,
+      "step": 854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9014608502090744e-05,
+      "loss": 0.5882,
+      "step": 855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.901231403004403e-05,
+      "loss": 0.5415,
+      "step": 856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9010017028544023e-05,
+      "loss": 0.6163,
+      "step": 857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9007717498235412e-05,
+      "loss": 0.5451,
+      "step": 858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.90054154397636e-05,
+      "loss": 0.5427,
+      "step": 859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9003110853774694e-05,
+      "loss": 0.6107,
+      "step": 860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9000803740915515e-05,
+      "loss": 0.5687,
+      "step": 861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.89984941018336e-05,
+      "loss": 0.5573,
+      "step": 862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8996181937177173e-05,
+      "loss": 0.5551,
+      "step": 863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.899386724759519e-05,
+      "loss": 0.5538,
+      "step": 864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.89915500337373e-05,
+      "loss": 0.5824,
+      "step": 865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.898923029625388e-05,
+      "loss": 0.6187,
+      "step": 866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8986908035795986e-05,
+      "loss": 0.6308,
+      "step": 867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.898458325301541e-05,
+      "loss": 0.5597,
+      "step": 868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8982255948564635e-05,
+      "loss": 0.5874,
+      "step": 869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8979926123096858e-05,
+      "loss": 0.5634,
+      "step": 870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8977593777265978e-05,
+      "loss": 0.5411,
+      "step": 871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8975258911726616e-05,
+      "loss": 0.5591,
+      "step": 872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8972921527134085e-05,
+      "loss": 0.5356,
+      "step": 873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8970581624144412e-05,
+      "loss": 0.5493,
+      "step": 874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.896823920341432e-05,
+      "loss": 0.5647,
+      "step": 875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8965894265601263e-05,
+      "loss": 0.6083,
+      "step": 876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8963546811363375e-05,
+      "loss": 0.5754,
+      "step": 877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8961196841359507e-05,
+      "loss": 0.5571,
+      "step": 878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8958844356249223e-05,
+      "loss": 0.5892,
+      "step": 879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.895648935669278e-05,
+      "loss": 0.5377,
+      "step": 880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.895413184335116e-05,
+      "loss": 0.5311,
+      "step": 881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.895177181688602e-05,
+      "loss": 0.524,
+      "step": 882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.894940927795975e-05,
+      "loss": 0.5935,
+      "step": 883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.894704422723543e-05,
+      "loss": 0.5717,
+      "step": 884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8944676665376858e-05,
+      "loss": 0.5819,
+      "step": 885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8942306593048522e-05,
+      "loss": 0.5643,
+      "step": 886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8939934010915627e-05,
+      "loss": 0.5349,
+      "step": 887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.893755891964407e-05,
+      "loss": 0.5445,
+      "step": 888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8935181319900467e-05,
+      "loss": 0.526,
+      "step": 889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8932801212352124e-05,
+      "loss": 0.5165,
+      "step": 890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.893041859766706e-05,
+      "loss": 0.5332,
+      "step": 891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8928033476513996e-05,
+      "loss": 0.5564,
+      "step": 892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.892564584956235e-05,
+      "loss": 0.5446,
+      "step": 893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8923255717482256e-05,
+      "loss": 0.5496,
+      "step": 894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8920863080944534e-05,
+      "loss": 0.5787,
+      "step": 895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8918467940620728e-05,
+      "loss": 0.5992,
+      "step": 896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8916070297183066e-05,
+      "loss": 0.5864,
+      "step": 897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.891367015130449e-05,
+      "loss": 0.5299,
+      "step": 898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.891126750365863e-05,
+      "loss": 0.5238,
+      "step": 899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8908862354919843e-05,
+      "loss": 0.5545,
+      "step": 900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8906454705763163e-05,
+      "loss": 0.5887,
+      "step": 901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8904044556864343e-05,
+      "loss": 0.5461,
+      "step": 902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8901631908899824e-05,
+      "loss": 0.5636,
+      "step": 903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8899216762546758e-05,
+      "loss": 0.5595,
+      "step": 904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8896799118482995e-05,
+      "loss": 0.5476,
+      "step": 905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.889437897738709e-05,
+      "loss": 0.5804,
+      "step": 906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8891956339938287e-05,
+      "loss": 0.5566,
+      "step": 907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8889531206816546e-05,
+      "loss": 0.5321,
+      "step": 908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.888710357870252e-05,
+      "loss": 0.5781,
+      "step": 909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.888467345627756e-05,
+      "loss": 0.5637,
+      "step": 910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.888224084022372e-05,
+      "loss": 0.5695,
+      "step": 911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8879805731223752e-05,
+      "loss": 0.5982,
+      "step": 912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8877368129961115e-05,
+      "loss": 0.5444,
+      "step": 913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8874928037119953e-05,
+      "loss": 0.5021,
+      "step": 914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8872485453385124e-05,
+      "loss": 0.5272,
+      "step": 915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8870040379442175e-05,
+      "loss": 0.5637,
+      "step": 916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8867592815977362e-05,
+      "loss": 0.5697,
+      "step": 917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.886514276367763e-05,
+      "loss": 0.5661,
+      "step": 918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8862690223230625e-05,
+      "loss": 0.5432,
+      "step": 919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8860235195324695e-05,
+      "loss": 0.5486,
+      "step": 920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8857777680648884e-05,
+      "loss": 0.5475,
+      "step": 921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885531767989293e-05,
+      "loss": 0.5502,
+      "step": 922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8852855193747274e-05,
+      "loss": 0.536,
+      "step": 923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8850390222903057e-05,
+      "loss": 0.5224,
+      "step": 924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8847922768052105e-05,
+      "loss": 0.5563,
+      "step": 925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8845452829886957e-05,
+      "loss": 0.5282,
+      "step": 926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8842980409100845e-05,
+      "loss": 0.5655,
+      "step": 927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8840505506387684e-05,
+      "loss": 0.5653,
+      "step": 928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.88380281224421e-05,
+      "loss": 0.5747,
+      "step": 929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8835548257959413e-05,
+      "loss": 0.5567,
+      "step": 930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8833065913635634e-05,
+      "loss": 0.5576,
+      "step": 931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8830581090167477e-05,
+      "loss": 0.5928,
+      "step": 932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8828093788252348e-05,
+      "loss": 0.6024,
+      "step": 933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882560400858835e-05,
+      "loss": 0.5754,
+      "step": 934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8823111751874277e-05,
+      "loss": 0.6022,
+      "step": 935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882061701880962e-05,
+      "loss": 0.5968,
+      "step": 936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8818119810094576e-05,
+      "loss": 0.5366,
+      "step": 937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8815620126430016e-05,
+      "loss": 0.5694,
+      "step": 938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8813117968517524e-05,
+      "loss": 0.5477,
+      "step": 939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.881061333705937e-05,
+      "loss": 0.5917,
+      "step": 940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.880810623275852e-05,
+      "loss": 0.5322,
+      "step": 941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8805596656318635e-05,
+      "loss": 0.5812,
+      "step": 942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8803084608444063e-05,
+      "loss": 0.5655,
+      "step": 943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880057008983986e-05,
+      "loss": 0.541,
+      "step": 944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.879805310121176e-05,
+      "loss": 0.5597,
+      "step": 945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.87955336432662e-05,
+      "loss": 0.5374,
+      "step": 946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8793011716710302e-05,
+      "loss": 0.5676,
+      "step": 947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8790487322251896e-05,
+      "loss": 0.5439,
+      "step": 948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8787960460599482e-05,
+      "loss": 0.5994,
+      "step": 949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8785431132462278e-05,
+      "loss": 0.531,
+      "step": 950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8782899338550175e-05,
+      "loss": 0.5691,
+      "step": 951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.878036507957376e-05,
+      "loss": 0.5385,
+      "step": 952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8777828356244316e-05,
+      "loss": 0.5474,
+      "step": 953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.877528916927382e-05,
+      "loss": 0.5952,
+      "step": 954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8772747519374927e-05,
+      "loss": 0.5825,
+      "step": 955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8770203407261004e-05,
+      "loss": 0.5566,
+      "step": 956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.876765683364609e-05,
+      "loss": 0.5683,
+      "step": 957
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8765107799244927e-05,
+      "loss": 0.5128,
+      "step": 958
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8762556304772936e-05,
+      "loss": 0.5252,
+      "step": 959
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8760002350946244e-05,
+      "loss": 0.5762,
+      "step": 960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8757445938481654e-05,
+      "loss": 0.5178,
+      "step": 961
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8754887068096672e-05,
+      "loss": 0.5271,
+      "step": 962
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.875232574050948e-05,
+      "loss": 0.5369,
+      "step": 963
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8749761956438956e-05,
+      "loss": 0.6264,
+      "step": 964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8747195716604675e-05,
+      "loss": 0.5599,
+      "step": 965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8744627021726887e-05,
+      "loss": 0.5862,
+      "step": 966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.874205587252654e-05,
+      "loss": 0.5156,
+      "step": 967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8739482269725272e-05,
+      "loss": 0.5341,
+      "step": 968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8736906214045405e-05,
+      "loss": 0.518,
+      "step": 969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.873432770620995e-05,
+      "loss": 0.5551,
+      "step": 970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8731746746942606e-05,
+      "loss": 0.5534,
+      "step": 971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8729163336967767e-05,
+      "loss": 0.5365,
+      "step": 972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8726577477010506e-05,
+      "loss": 0.5266,
+      "step": 973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8723989167796585e-05,
+      "loss": 0.5612,
+      "step": 974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.872139841005246e-05,
+      "loss": 0.5674,
+      "step": 975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8718805204505265e-05,
+      "loss": 0.56,
+      "step": 976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8716209551882826e-05,
+      "loss": 0.5646,
+      "step": 977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8713611452913657e-05,
+      "loss": 0.5802,
+      "step": 978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8711010908326957e-05,
+      "loss": 0.5417,
+      "step": 979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8708407918852608e-05,
+      "loss": 0.5585,
+      "step": 980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.870580248522119e-05,
+      "loss": 0.5539,
+      "step": 981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.870319460816395e-05,
+      "loss": 0.5482,
+      "step": 982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8700584288412836e-05,
+      "loss": 0.5743,
+      "step": 983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8697971526700482e-05,
+      "loss": 0.5737,
+      "step": 984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8695356323760197e-05,
+      "loss": 0.5375,
+      "step": 985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8692738680325976e-05,
+      "loss": 0.5992,
+      "step": 986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8690118597132514e-05,
+      "loss": 0.5642,
+      "step": 987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.868749607491517e-05,
+      "loss": 0.5322,
+      "step": 988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8684871114410006e-05,
+      "loss": 0.5381,
+      "step": 989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8682243716353754e-05,
+      "loss": 0.5528,
+      "step": 990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.867961388148384e-05,
+      "loss": 0.5554,
+      "step": 991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8676981610538366e-05,
+      "loss": 0.5264,
+      "step": 992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.867434690425613e-05,
+      "loss": 0.5763,
+      "step": 993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8671709763376596e-05,
+      "loss": 0.6016,
+      "step": 994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8669070188639924e-05,
+      "loss": 0.5204,
+      "step": 995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8666428180786956e-05,
+      "loss": 0.5293,
+      "step": 996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8663783740559214e-05,
+      "loss": 0.5339,
+      "step": 997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8661136868698903e-05,
+      "loss": 0.5588,
+      "step": 998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.865848756594891e-05,
+      "loss": 0.5394,
+      "step": 999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8655835833052808e-05,
+      "loss": 0.5522,
+      "step": 1000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8653181670754845e-05,
+      "loss": 0.6408,
+      "step": 1001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8650525079799955e-05,
+      "loss": 0.6042,
+      "step": 1002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8647866060933756e-05,
+      "loss": 0.5654,
+      "step": 1003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8645204614902548e-05,
+      "loss": 0.5483,
+      "step": 1004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8642540742453302e-05,
+      "loss": 0.5671,
+      "step": 1005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8639874444333687e-05,
+      "loss": 0.5773,
+      "step": 1006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8637205721292033e-05,
+      "loss": 0.5795,
+      "step": 1007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8634534574077365e-05,
+      "loss": 0.5965,
+      "step": 1008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8631861003439388e-05,
+      "loss": 0.5701,
+      "step": 1009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8629185010128478e-05,
+      "loss": 0.5733,
+      "step": 1010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8626506594895696e-05,
+      "loss": 0.5599,
+      "step": 1011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8623825758492788e-05,
+      "loss": 0.5474,
+      "step": 1012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8621142501672175e-05,
+      "loss": 0.5942,
+      "step": 1013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.861845682518695e-05,
+      "loss": 0.578,
+      "step": 1014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8615768729790893e-05,
+      "loss": 0.5267,
+      "step": 1015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8613078216238468e-05,
+      "loss": 0.548,
+      "step": 1016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8610385285284806e-05,
+      "loss": 0.5534,
+      "step": 1017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8607689937685728e-05,
+      "loss": 0.554,
+      "step": 1018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8604992174197716e-05,
+      "loss": 0.5538,
+      "step": 1019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8602291995577957e-05,
+      "loss": 0.5392,
+      "step": 1020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8599589402584288e-05,
+      "loss": 0.552,
+      "step": 1021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8596884395975242e-05,
+      "loss": 0.5482,
+      "step": 1022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.859417697651002e-05,
+      "loss": 0.5613,
+      "step": 1023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.859146714494851e-05,
+      "loss": 0.5874,
+      "step": 1024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8588754902051262e-05,
+      "loss": 0.5678,
+      "step": 1025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8586040248579515e-05,
+      "loss": 0.5738,
+      "step": 1026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.858332318529518e-05,
+      "loss": 0.5775,
+      "step": 1027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.858060371296085e-05,
+      "loss": 0.569,
+      "step": 1028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.857788183233978e-05,
+      "loss": 0.5369,
+      "step": 1029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.857515754419592e-05,
+      "loss": 0.5448,
+      "step": 1030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8572430849293882e-05,
+      "loss": 0.5283,
+      "step": 1031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8569701748398954e-05,
+      "loss": 0.5402,
+      "step": 1032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.856697024227711e-05,
+      "loss": 0.5652,
+      "step": 1033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8564236331694988e-05,
+      "loss": 0.5265,
+      "step": 1034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8561500017419902e-05,
+      "loss": 0.5431,
+      "step": 1035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8558761300219846e-05,
+      "loss": 0.5301,
+      "step": 1036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8556020180863487e-05,
+      "loss": 0.5852,
+      "step": 1037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8553276660120164e-05,
+      "loss": 0.5673,
+      "step": 1038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.855053073875989e-05,
+      "loss": 0.5536,
+      "step": 1039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8547782417553355e-05,
+      "loss": 0.5695,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.854503169727192e-05,
+      "loss": 0.547,
+      "step": 1041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8542278578687614e-05,
+      "loss": 0.5649,
+      "step": 1042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.853952306257315e-05,
+      "loss": 0.5144,
+      "step": 1043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.853676514970191e-05,
+      "loss": 0.5283,
+      "step": 1044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8534004840847943e-05,
+      "loss": 0.5587,
+      "step": 1045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.853124213678598e-05,
+      "loss": 0.5484,
+      "step": 1046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8528477038291416e-05,
+      "loss": 0.5718,
+      "step": 1047
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8525709546140314e-05,
+      "loss": 0.5829,
+      "step": 1048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.852293966110943e-05,
+      "loss": 0.6006,
+      "step": 1049
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8520167383976168e-05,
+      "loss": 0.5346,
+      "step": 1050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.851739271551862e-05,
+      "loss": 0.5626,
+      "step": 1051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8514615656515534e-05,
+      "loss": 0.553,
+      "step": 1052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.851183620774634e-05,
+      "loss": 0.5863,
+      "step": 1053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8509054369991136e-05,
+      "loss": 0.5691,
+      "step": 1054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.850627014403069e-05,
+      "loss": 0.5806,
+      "step": 1055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8503483530646443e-05,
+      "loss": 0.6041,
+      "step": 1056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.85006945306205e-05,
+      "loss": 0.5592,
+      "step": 1057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8497903144735643e-05,
+      "loss": 0.5467,
+      "step": 1058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8495109373775317e-05,
+      "loss": 0.5648,
+      "step": 1059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.849231321852364e-05,
+      "loss": 0.5405,
+      "step": 1060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8489514679765402e-05,
+      "loss": 0.5315,
+      "step": 1061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8486713758286052e-05,
+      "loss": 0.5274,
+      "step": 1062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8483910454871724e-05,
+      "loss": 0.5696,
+      "step": 1063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.84811047703092e-05,
+      "loss": 0.54,
+      "step": 1064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8478296705385953e-05,
+      "loss": 0.5545,
+      "step": 1065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.84754862608901e-05,
+      "loss": 0.5261,
+      "step": 1066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8472673437610448e-05,
+      "loss": 0.5648,
+      "step": 1067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8469858236336456e-05,
+      "loss": 0.5383,
+      "step": 1068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.846704065785826e-05,
+      "loss": 0.5471,
+      "step": 1069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8464220702966656e-05,
+      "loss": 0.5655,
+      "step": 1070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8461398372453113e-05,
+      "loss": 0.56,
+      "step": 1071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8458573667109765e-05,
+      "loss": 0.5315,
+      "step": 1072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.845574658772941e-05,
+      "loss": 0.5664,
+      "step": 1073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8452917135105513e-05,
+      "loss": 0.6063,
+      "step": 1074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8450085310032206e-05,
+      "loss": 0.5747,
+      "step": 1075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.844725111330429e-05,
+      "loss": 0.5459,
+      "step": 1076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.844441454571723e-05,
+      "loss": 0.5482,
+      "step": 1077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.844157560806715e-05,
+      "loss": 0.5597,
+      "step": 1078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8438734301150845e-05,
+      "loss": 0.5262,
+      "step": 1079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8435890625765776e-05,
+      "loss": 0.5731,
+      "step": 1080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8433044582710067e-05,
+      "loss": 0.5268,
+      "step": 1081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8430196172782505e-05,
+      "loss": 0.5854,
+      "step": 1082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8427345396782547e-05,
+      "loss": 0.5544,
+      "step": 1083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8424492255510305e-05,
+      "loss": 0.5715,
+      "step": 1084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8421636749766563e-05,
+      "loss": 0.5682,
+      "step": 1085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.841877888035276e-05,
+      "loss": 0.561,
+      "step": 1086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8415918648071014e-05,
+      "loss": 0.5562,
+      "step": 1087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8413056053724086e-05,
+      "loss": 0.5626,
+      "step": 1088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.841019109811541e-05,
+      "loss": 0.5773,
+      "step": 1089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8407323782049093e-05,
+      "loss": 0.5336,
+      "step": 1090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8404454106329886e-05,
+      "loss": 0.516,
+      "step": 1091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.840158207176321e-05,
+      "loss": 0.5546,
+      "step": 1092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8398707679155154e-05,
+      "loss": 0.5857,
+      "step": 1093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8395830929312457e-05,
+      "loss": 0.5659,
+      "step": 1094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8392951823042525e-05,
+      "loss": 0.5407,
+      "step": 1095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8390070361153436e-05,
+      "loss": 0.562,
+      "step": 1096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.838718654445391e-05,
+      "loss": 0.5819,
+      "step": 1097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.838430037375334e-05,
+      "loss": 0.5558,
+      "step": 1098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.838141184986178e-05,
+      "loss": 0.576,
+      "step": 1099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8378520973589937e-05,
+      "loss": 0.5851,
+      "step": 1100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8375627745749184e-05,
+      "loss": 0.5467,
+      "step": 1101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8372732167151556e-05,
+      "loss": 0.5627,
+      "step": 1102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.836983423860974e-05,
+      "loss": 0.5335,
+      "step": 1103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.836693396093709e-05,
+      "loss": 0.566,
+      "step": 1104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8364031334947612e-05,
+      "loss": 0.5558,
+      "step": 1105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.836112636145598e-05,
+      "loss": 0.5288,
+      "step": 1106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8358219041277523e-05,
+      "loss": 0.5459,
+      "step": 1107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8355309375228225e-05,
+      "loss": 0.5884,
+      "step": 1108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.835239736412473e-05,
+      "loss": 0.5693,
+      "step": 1109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8349483008784346e-05,
+      "loss": 0.5611,
+      "step": 1110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.834656631002503e-05,
+      "loss": 0.5495,
+      "step": 1111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8343647268665407e-05,
+      "loss": 0.5381,
+      "step": 1112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8340725885524747e-05,
+      "loss": 0.5539,
+      "step": 1113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8337802161422987e-05,
+      "loss": 0.5514,
+      "step": 1114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.833487609718072e-05,
+      "loss": 0.5761,
+      "step": 1115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8331947693619195e-05,
+      "loss": 0.5259,
+      "step": 1116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8329016951560307e-05,
+      "loss": 0.5635,
+      "step": 1117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8326083871826626e-05,
+      "loss": 0.5649,
+      "step": 1118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8323148455241363e-05,
+      "loss": 0.5648,
+      "step": 1119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8320210702628397e-05,
+      "loss": 0.5527,
+      "step": 1120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.831727061481225e-05,
+      "loss": 0.5816,
+      "step": 1121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.831432819261811e-05,
+      "loss": 0.5715,
+      "step": 1122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8311383436871814e-05,
+      "loss": 0.5798,
+      "step": 1123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8308436348399855e-05,
+      "loss": 0.5713,
+      "step": 1124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8305486928029383e-05,
+      "loss": 0.5625,
+      "step": 1125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.83025351765882e-05,
+      "loss": 0.5452,
+      "step": 1126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8299581094904765e-05,
+      "loss": 0.5403,
+      "step": 1127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.829662468380819e-05,
+      "loss": 0.5742,
+      "step": 1128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8293665944128237e-05,
+      "loss": 0.5546,
+      "step": 1129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8290704876695325e-05,
+      "loss": 0.544,
+      "step": 1130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.828774148234053e-05,
+      "loss": 0.5635,
+      "step": 1131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.828477576189557e-05,
+      "loss": 0.5788,
+      "step": 1132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.828180771619283e-05,
+      "loss": 0.5658,
+      "step": 1133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8278837346065336e-05,
+      "loss": 0.5797,
+      "step": 1134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8275864652346772e-05,
+      "loss": 0.5697,
+      "step": 1135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8272889635871475e-05,
+      "loss": 0.5545,
+      "step": 1136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.826991229747443e-05,
+      "loss": 0.5676,
+      "step": 1137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8266932637991278e-05,
+      "loss": 0.5892,
+      "step": 1138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8263950658258306e-05,
+      "loss": 0.5726,
+      "step": 1139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.826096635911246e-05,
+      "loss": 0.5479,
+      "step": 1140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8257979741391326e-05,
+      "loss": 0.5627,
+      "step": 1141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.825499080593315e-05,
+      "loss": 0.5281,
+      "step": 1142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.825199955357683e-05,
+      "loss": 0.5572,
+      "step": 1143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.82490059851619e-05,
+      "loss": 0.5584,
+      "step": 1144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8246010101528566e-05,
+      "loss": 0.5633,
+      "step": 1145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8243011903517663e-05,
+      "loss": 0.5161,
+      "step": 1146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8240011391970688e-05,
+      "loss": 0.5349,
+      "step": 1147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8237008567729783e-05,
+      "loss": 0.5578,
+      "step": 1148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8234003431637734e-05,
+      "loss": 0.5572,
+      "step": 1149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.823099598453799e-05,
+      "loss": 0.5674,
+      "step": 1150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8227986227274638e-05,
+      "loss": 0.56,
+      "step": 1151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.822497416069241e-05,
+      "loss": 0.5024,
+      "step": 1152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8221959785636703e-05,
+      "loss": 0.5567,
+      "step": 1153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8218943102953545e-05,
+      "loss": 0.5724,
+      "step": 1154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8215924113489613e-05,
+      "loss": 0.5166,
+      "step": 1155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8212902818092236e-05,
+      "loss": 0.5596,
+      "step": 1156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8209879217609394e-05,
+      "loss": 0.5518,
+      "step": 1157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8206853312889713e-05,
+      "loss": 0.5058,
+      "step": 1158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8203825104782457e-05,
+      "loss": 0.5774,
+      "step": 1159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.820079459413754e-05,
+      "loss": 0.5464,
+      "step": 1160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8197761781805532e-05,
+      "loss": 0.5645,
+      "step": 1161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8194726668637636e-05,
+      "loss": 0.5986,
+      "step": 1162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8191689255485707e-05,
+      "loss": 0.5357,
+      "step": 1163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8188649543202244e-05,
+      "loss": 0.5441,
+      "step": 1164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8185607532640396e-05,
+      "loss": 0.6002,
+      "step": 1165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8182563224653947e-05,
+      "loss": 0.5582,
+      "step": 1166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.817951662009734e-05,
+      "loss": 0.5546,
+      "step": 1167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8176467719825645e-05,
+      "loss": 0.5573,
+      "step": 1168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8173416524694595e-05,
+      "loss": 0.5548,
+      "step": 1169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8170363035560544e-05,
+      "loss": 0.5505,
+      "step": 1170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.816730725328052e-05,
+      "loss": 0.5436,
+      "step": 1171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.816424917871217e-05,
+      "loss": 0.5785,
+      "step": 1172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.816118881271379e-05,
+      "loss": 0.5582,
+      "step": 1173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.815812615614433e-05,
+      "loss": 0.5828,
+      "step": 1174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8155061209863368e-05,
+      "loss": 0.5446,
+      "step": 1175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.815199397473113e-05,
+      "loss": 0.5474,
+      "step": 1176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.814892445160849e-05,
+      "loss": 0.5278,
+      "step": 1177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8145852641356963e-05,
+      "loss": 0.5898,
+      "step": 1178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8142778544838695e-05,
+      "loss": 0.5891,
+      "step": 1179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8139702162916485e-05,
+      "loss": 0.5793,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.813662349645377e-05,
+      "loss": 0.5586,
+      "step": 1181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8133542546314628e-05,
+      "loss": 0.5638,
+      "step": 1182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.813045931336378e-05,
+      "loss": 0.5568,
+      "step": 1183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.812737379846658e-05,
+      "loss": 0.5527,
+      "step": 1184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8124286002489034e-05,
+      "loss": 0.5288,
+      "step": 1185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.812119592629778e-05,
+      "loss": 0.5394,
+      "step": 1186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8118103570760097e-05,
+      "loss": 0.5451,
+      "step": 1187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8115008936743908e-05,
+      "loss": 0.5415,
+      "step": 1188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8111912025117774e-05,
+      "loss": 0.5766,
+      "step": 1189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8108812836750887e-05,
+      "loss": 0.537,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.810571137251309e-05,
+      "loss": 0.5629,
+      "step": 1191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8102607633274863e-05,
+      "loss": 0.6042,
+      "step": 1192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.809950161990731e-05,
+      "loss": 0.5561,
+      "step": 1193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8096393333282195e-05,
+      "loss": 0.5187,
+      "step": 1194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8093282774271908e-05,
+      "loss": 0.5787,
+      "step": 1195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.5906,
+      "step": 1196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8087054842588565e-05,
+      "loss": 0.5603,
+      "step": 1197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808393747166348e-05,
+      "loss": 0.5595,
+      "step": 1198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808081783184916e-05,
+      "loss": 0.5911,
+      "step": 1199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.807769592402119e-05,
+      "loss": 0.5518,
+      "step": 1200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8074571749055778e-05,
+      "loss": 0.5511,
+      "step": 1201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8071445307829775e-05,
+      "loss": 0.5431,
+      "step": 1202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8068316601220667e-05,
+      "loss": 0.5305,
+      "step": 1203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8065185630106583e-05,
+      "loss": 0.5709,
+      "step": 1204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8062052395366275e-05,
+      "loss": 0.5471,
+      "step": 1205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.805891689787914e-05,
+      "loss": 0.521,
+      "step": 1206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8055779138525205e-05,
+      "loss": 0.565,
+      "step": 1207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8052639118185132e-05,
+      "loss": 0.5485,
+      "step": 1208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8049496837740223e-05,
+      "loss": 0.5483,
+      "step": 1209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8046352298072408e-05,
+      "loss": 0.5808,
+      "step": 1210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.804320550006425e-05,
+      "loss": 0.5419,
+      "step": 1211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8040056444598958e-05,
+      "loss": 0.5123,
+      "step": 1212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8036905132560362e-05,
+      "loss": 0.5991,
+      "step": 1213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8033751564832926e-05,
+      "loss": 0.5723,
+      "step": 1214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.803059574230175e-05,
+      "loss": 0.5062,
+      "step": 1215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8027437665852574e-05,
+      "loss": 0.5779,
+      "step": 1216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8024277336371755e-05,
+      "loss": 0.5962,
+      "step": 1217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.80211147547463e-05,
+      "loss": 0.575,
+      "step": 1218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8017949921863832e-05,
+      "loss": 0.5319,
+      "step": 1219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8014782838612616e-05,
+      "loss": 0.5337,
+      "step": 1220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8011613505881545e-05,
+      "loss": 0.5396,
+      "step": 1221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.800844192456015e-05,
+      "loss": 0.5938,
+      "step": 1222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.800526809553858e-05,
+      "loss": 0.5384,
+      "step": 1223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.800209201970762e-05,
+      "loss": 0.5173,
+      "step": 1224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7998913697958693e-05,
+      "loss": 0.5705,
+      "step": 1225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7995733131183846e-05,
+      "loss": 0.5306,
+      "step": 1226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.799255032027576e-05,
+      "loss": 0.5502,
+      "step": 1227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7989365266127735e-05,
+      "loss": 0.5615,
+      "step": 1228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7986177969633717e-05,
+      "loss": 0.5426,
+      "step": 1229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7982988431688266e-05,
+      "loss": 0.5941,
+      "step": 1230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.797979665318658e-05,
+      "loss": 0.5623,
+      "step": 1231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7976602635024485e-05,
+      "loss": 0.5853,
+      "step": 1232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7973406378098434e-05,
+      "loss": 0.4976,
+      "step": 1233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7970207883305512e-05,
+      "loss": 0.5476,
+      "step": 1234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7967007151543425e-05,
+      "loss": 0.5423,
+      "step": 1235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.796380418371051e-05,
+      "loss": 0.5634,
+      "step": 1236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7960598980705734e-05,
+      "loss": 0.5151,
+      "step": 1237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.795739154342869e-05,
+      "loss": 0.5512,
+      "step": 1238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7954181872779598e-05,
+      "loss": 0.586,
+      "step": 1239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7950969969659303e-05,
+      "loss": 0.5718,
+      "step": 1240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.794775583496928e-05,
+      "loss": 0.5715,
+      "step": 1241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7944539469611625e-05,
+      "loss": 0.5841,
+      "step": 1242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7941320874489065e-05,
+      "loss": 0.5748,
+      "step": 1243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7938100050504953e-05,
+      "loss": 0.5816,
+      "step": 1244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7934876998563263e-05,
+      "loss": 0.5453,
+      "step": 1245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7931651719568603e-05,
+      "loss": 0.5642,
+      "step": 1246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7928424214426196e-05,
+      "loss": 0.5858,
+      "step": 1247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7925194484041893e-05,
+      "loss": 0.6241,
+      "step": 1248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7921962529322173e-05,
+      "loss": 0.5986,
+      "step": 1249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7918728351174136e-05,
+      "loss": 0.587,
+      "step": 1250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.791549195050551e-05,
+      "loss": 0.5381,
+      "step": 1251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.791225332822464e-05,
+      "loss": 0.5746,
+      "step": 1252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7909012485240497e-05,
+      "loss": 0.5194,
+      "step": 1253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.790576942246268e-05,
+      "loss": 0.5305,
+      "step": 1254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.790252414080141e-05,
+      "loss": 0.5204,
+      "step": 1255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7899276641167516e-05,
+      "loss": 0.5574,
+      "step": 1256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7896026924472478e-05,
+      "loss": 0.5867,
+      "step": 1257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.789277499162837e-05,
+      "loss": 0.5388,
+      "step": 1258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7889520843547908e-05,
+      "loss": 0.5574,
+      "step": 1259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.788626448114442e-05,
+      "loss": 0.5374,
+      "step": 1260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7883005905331855e-05,
+      "loss": 0.5887,
+      "step": 1261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.787974511702479e-05,
+      "loss": 0.5416,
+      "step": 1262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.787648211713841e-05,
+      "loss": 0.5464,
+      "step": 1263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7873216906588536e-05,
+      "loss": 0.5829,
+      "step": 1264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7869949486291604e-05,
+      "loss": 0.54,
+      "step": 1265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7866679857164663e-05,
+      "loss": 0.5434,
+      "step": 1266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.786340802012539e-05,
+      "loss": 0.5229,
+      "step": 1267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7860133976092083e-05,
+      "loss": 0.5692,
+      "step": 1268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7856857725983653e-05,
+      "loss": 0.5863,
+      "step": 1269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7853579270719635e-05,
+      "loss": 0.5444,
+      "step": 1270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.785029861122018e-05,
+      "loss": 0.5632,
+      "step": 1271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7847015748406055e-05,
+      "loss": 0.5744,
+      "step": 1272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7843730683198658e-05,
+      "loss": 0.5433,
+      "step": 1273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7840443416519985e-05,
+      "loss": 0.5505,
+      "step": 1274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7837153949292674e-05,
+      "loss": 0.5459,
+      "step": 1275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7833862282439956e-05,
+      "loss": 0.5358,
+      "step": 1276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7830568416885697e-05,
+      "loss": 0.5456,
+      "step": 1277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7827272353554376e-05,
+      "loss": 0.556,
+      "step": 1278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.782397409337108e-05,
+      "loss": 0.5292,
+      "step": 1279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.782067363726153e-05,
+      "loss": 0.5772,
+      "step": 1280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7817370986152042e-05,
+      "loss": 0.551,
+      "step": 1281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7814066140969565e-05,
+      "loss": 0.5204,
+      "step": 1282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7810759102641657e-05,
+      "loss": 0.5896,
+      "step": 1283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7807449872096492e-05,
+      "loss": 0.5536,
+      "step": 1284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7804138450262862e-05,
+      "loss": 0.577,
+      "step": 1285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7800824838070166e-05,
+      "loss": 0.5369,
+      "step": 1286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.779750903644843e-05,
+      "loss": 0.5514,
+      "step": 1287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7794191046328283e-05,
+      "loss": 0.5477,
+      "step": 1288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.779087086864098e-05,
+      "loss": 0.5724,
+      "step": 1289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7787548504318372e-05,
+      "loss": 0.544,
+      "step": 1290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7784223954292944e-05,
+      "loss": 0.5232,
+      "step": 1291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7780897219497783e-05,
+      "loss": 0.5792,
+      "step": 1292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7777568300866587e-05,
+      "loss": 0.5664,
+      "step": 1293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.777423719933368e-05,
+      "loss": 0.5499,
+      "step": 1294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7770903915833986e-05,
+      "loss": 0.5658,
+      "step": 1295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7767568451303042e-05,
+      "loss": 0.5446,
+      "step": 1296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7764230806677005e-05,
+      "loss": 0.5919,
+      "step": 1297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7760890982892638e-05,
+      "loss": 0.5649,
+      "step": 1298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.775754898088732e-05,
+      "loss": 0.5406,
+      "step": 1299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.775420480159903e-05,
+      "loss": 0.5525,
+      "step": 1300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.775085844596638e-05,
+      "loss": 0.5279,
+      "step": 1301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7747509914928568e-05,
+      "loss": 0.5058,
+      "step": 1302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7744159209425416e-05,
+      "loss": 0.5327,
+      "step": 1303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.774080633039736e-05,
+      "loss": 0.5858,
+      "step": 1304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7737451278785435e-05,
+      "loss": 0.5439,
+      "step": 1305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.773409405553129e-05,
+      "loss": 0.5526,
+      "step": 1306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.773073466157719e-05,
+      "loss": 0.5348,
+      "step": 1307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7727373097866002e-05,
+      "loss": 0.498,
+      "step": 1308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7724009365341204e-05,
+      "loss": 0.5656,
+      "step": 1309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.772064346494688e-05,
+      "loss": 0.5645,
+      "step": 1310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7717275397627724e-05,
+      "loss": 0.5431,
+      "step": 1311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7713905164329044e-05,
+      "loss": 0.5695,
+      "step": 1312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.771053276599675e-05,
+      "loss": 0.5572,
+      "step": 1313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7707158203577356e-05,
+      "loss": 0.5517,
+      "step": 1314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7703781478017995e-05,
+      "loss": 0.5426,
+      "step": 1315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.770040259026639e-05,
+      "loss": 0.5556,
+      "step": 1316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7697021541270897e-05,
+      "loss": 0.5778,
+      "step": 1317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.769363833198045e-05,
+      "loss": 0.5986,
+      "step": 1318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7690252963344606e-05,
+      "loss": 0.5727,
+      "step": 1319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.768686543631352e-05,
+      "loss": 0.536,
+      "step": 1320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7683475751837963e-05,
+      "loss": 0.5263,
+      "step": 1321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7680083910869305e-05,
+      "loss": 0.5217,
+      "step": 1322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7676689914359517e-05,
+      "loss": 0.5059,
+      "step": 1323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7673293763261182e-05,
+      "loss": 0.5205,
+      "step": 1324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7669895458527487e-05,
+      "loss": 0.5552,
+      "step": 1325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.766649500111222e-05,
+      "loss": 0.5728,
+      "step": 1326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7663092391969773e-05,
+      "loss": 0.5808,
+      "step": 1327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.765968763205515e-05,
+      "loss": 0.5567,
+      "step": 1328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7656280722323945e-05,
+      "loss": 0.5205,
+      "step": 1329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.765287166373237e-05,
+      "loss": 0.5659,
+      "step": 1330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7649460457237223e-05,
+      "loss": 0.5307,
+      "step": 1331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7646047103795928e-05,
+      "loss": 0.5007,
+      "step": 1332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7642631604366493e-05,
+      "loss": 0.5241,
+      "step": 1333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7639213959907525e-05,
+      "loss": 0.5569,
+      "step": 1334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7635794171378257e-05,
+      "loss": 0.519,
+      "step": 1335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7632372239738497e-05,
+      "loss": 0.5643,
+      "step": 1336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.762894816594867e-05,
+      "loss": 0.5979,
+      "step": 1337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7625521950969802e-05,
+      "loss": 0.5538,
+      "step": 1338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.762209359576351e-05,
+      "loss": 0.5651,
+      "step": 1339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.761866310129202e-05,
+      "loss": 0.5428,
+      "step": 1340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7615230468518157e-05,
+      "loss": 0.5413,
+      "step": 1341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.761179569840535e-05,
+      "loss": 0.5716,
+      "step": 1342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7608358791917615e-05,
+      "loss": 0.549,
+      "step": 1343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.760491975001958e-05,
+      "loss": 0.5682,
+      "step": 1344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.760147857367647e-05,
+      "loss": 0.5648,
+      "step": 1345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7598035263854103e-05,
+      "loss": 0.5652,
+      "step": 1346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7594589821518905e-05,
+      "loss": 0.5656,
+      "step": 1347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7591142247637895e-05,
+      "loss": 0.5546,
+      "step": 1348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7587692543178684e-05,
+      "loss": 0.5348,
+      "step": 1349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7584240709109498e-05,
+      "loss": 0.549,
+      "step": 1350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7580786746399146e-05,
+      "loss": 0.5672,
+      "step": 1351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7577330656017037e-05,
+      "loss": 0.5469,
+      "step": 1352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7573872438933183e-05,
+      "loss": 0.5214,
+      "step": 1353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.757041209611819e-05,
+      "loss": 0.5863,
+      "step": 1354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7566949628543252e-05,
+      "loss": 0.5102,
+      "step": 1355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7563485037180177e-05,
+      "loss": 0.5402,
+      "step": 1356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7560018323001354e-05,
+      "loss": 0.5629,
+      "step": 1357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7556549486979775e-05,
+      "loss": 0.5137,
+      "step": 1358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7553078530089024e-05,
+      "loss": 0.5946,
+      "step": 1359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.754960545330328e-05,
+      "loss": 0.5536,
+      "step": 1360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7546130257597322e-05,
+      "loss": 0.5257,
+      "step": 1361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7542652943946523e-05,
+      "loss": 0.5616,
+      "step": 1362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.753917351332684e-05,
+      "loss": 0.5994,
+      "step": 1363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.753569196671484e-05,
+      "loss": 0.588,
+      "step": 1364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.753220830508767e-05,
+      "loss": 0.5721,
+      "step": 1365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.752872252942308e-05,
+      "loss": 0.5262,
+      "step": 1366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.752523464069941e-05,
+      "loss": 0.5495,
+      "step": 1367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7521744639895593e-05,
+      "loss": 0.5531,
+      "step": 1368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7518252527991154e-05,
+      "loss": 0.58,
+      "step": 1369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7514758305966206e-05,
+      "loss": 0.5573,
+      "step": 1370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.751126197480147e-05,
+      "loss": 0.5405,
+      "step": 1371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.750776353547824e-05,
+      "loss": 0.5244,
+      "step": 1372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7504262988978417e-05,
+      "loss": 0.5468,
+      "step": 1373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.750076033628448e-05,
+      "loss": 0.517,
+      "step": 1374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7497255578379514e-05,
+      "loss": 0.559,
+      "step": 1375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7493748716247174e-05,
+      "loss": 0.5307,
+      "step": 1376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.749023975087173e-05,
+      "loss": 0.5896,
+      "step": 1377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7486728683238025e-05,
+      "loss": 0.5933,
+      "step": 1378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.74832155143315e-05,
+      "loss": 0.4987,
+      "step": 1379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7479700245138184e-05,
+      "loss": 0.5837,
+      "step": 1380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.747618287664469e-05,
+      "loss": 0.5429,
+      "step": 1381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7472663409838225e-05,
+      "loss": 0.5695,
+      "step": 1382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.746914184570659e-05,
+      "loss": 0.5263,
+      "step": 1383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7465618185238167e-05,
+      "loss": 0.561,
+      "step": 1384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.746209242942193e-05,
+      "loss": 0.5063,
+      "step": 1385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.745856457924744e-05,
+      "loss": 0.5745,
+      "step": 1386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7455034635704845e-05,
+      "loss": 0.5506,
+      "step": 1387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.745150259978488e-05,
+      "loss": 0.6118,
+      "step": 1388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.744796847247887e-05,
+      "loss": 0.5525,
+      "step": 1389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7444432254778725e-05,
+      "loss": 0.5814,
+      "step": 1390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7440893947676943e-05,
+      "loss": 0.5578,
+      "step": 1391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.743735355216661e-05,
+      "loss": 0.5782,
+      "step": 1392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7433811069241392e-05,
+      "loss": 0.5566,
+      "step": 1393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7430266499895547e-05,
+      "loss": 0.5485,
+      "step": 1394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7426719845123914e-05,
+      "loss": 0.5494,
+      "step": 1395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7423171105921924e-05,
+      "loss": 0.5581,
+      "step": 1396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7419620283285585e-05,
+      "loss": 0.5435,
+      "step": 1397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.741606737821149e-05,
+      "loss": 0.5617,
+      "step": 1398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.741251239169683e-05,
+      "loss": 0.5174,
+      "step": 1399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7408955324739363e-05,
+      "loss": 0.5197,
+      "step": 1400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.740539617833744e-05,
+      "loss": 0.5342,
+      "step": 1401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7401834953489993e-05,
+      "loss": 0.5741,
+      "step": 1402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7398271651196538e-05,
+      "loss": 0.5477,
+      "step": 1403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.739470627245717e-05,
+      "loss": 0.5801,
+      "step": 1404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7391138818272578e-05,
+      "loss": 0.5585,
+      "step": 1405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7387569289644025e-05,
+      "loss": 0.5288,
+      "step": 1406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7383997687573354e-05,
+      "loss": 0.5547,
+      "step": 1407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7380424013062996e-05,
+      "loss": 0.5832,
+      "step": 1408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.737684826711596e-05,
+      "loss": 0.5243,
+      "step": 1409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.737327045073584e-05,
+      "loss": 0.545,
+      "step": 1410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7369690564926808e-05,
+      "loss": 0.5605,
+      "step": 1411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.736610861069361e-05,
+      "loss": 0.567,
+      "step": 1412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7362524589041593e-05,
+      "loss": 0.5514,
+      "step": 1413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7358938500976662e-05,
+      "loss": 0.5287,
+      "step": 1414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7355350347505312e-05,
+      "loss": 0.5304,
+      "step": 1415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.735176012963462e-05,
+      "loss": 0.4766,
+      "step": 1416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.734816784837224e-05,
+      "loss": 0.5207,
+      "step": 1417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7344573504726397e-05,
+      "loss": 0.5648,
+      "step": 1418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7340977099705914e-05,
+      "loss": 0.5315,
+      "step": 1419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7337378634320173e-05,
+      "loss": 0.5466,
+      "step": 1420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7333778109579144e-05,
+      "loss": 0.5923,
+      "step": 1421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7330175526493373e-05,
+      "loss": 0.5792,
+      "step": 1422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7326570886073986e-05,
+      "loss": 0.5649,
+      "step": 1423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7322964189332682e-05,
+      "loss": 0.5476,
+      "step": 1424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7319355437281737e-05,
+      "loss": 0.5452,
+      "step": 1425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7315744630934007e-05,
+      "loss": 0.5462,
+      "step": 1426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.731213177130293e-05,
+      "loss": 0.5276,
+      "step": 1427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7308516859402507e-05,
+      "loss": 0.5519,
+      "step": 1428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7304899896247328e-05,
+      "loss": 0.5934,
+      "step": 1429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.730128088285255e-05,
+      "loss": 0.5544,
+      "step": 1430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7297659820233906e-05,
+      "loss": 0.5212,
+      "step": 1431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7294036709407707e-05,
+      "loss": 0.533,
+      "step": 1432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7290411551390845e-05,
+      "loss": 0.5498,
+      "step": 1433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7286784347200768e-05,
+      "loss": 0.5629,
+      "step": 1434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7283155097855525e-05,
+      "loss": 0.5244,
+      "step": 1435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.727952380437371e-05,
+      "loss": 0.5323,
+      "step": 1436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7275890467774515e-05,
+      "loss": 0.5459,
+      "step": 1437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.727225508907769e-05,
+      "loss": 0.5407,
+      "step": 1438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7268617669303564e-05,
+      "loss": 0.535,
+      "step": 1439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7264978209473035e-05,
+      "loss": 0.5439,
+      "step": 1440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7261336710607587e-05,
+      "loss": 0.5545,
+      "step": 1441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7257693173729256e-05,
+      "loss": 0.5662,
+      "step": 1442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.725404759986067e-05,
+      "loss": 0.5379,
+      "step": 1443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7250399990025005e-05,
+      "loss": 0.5472,
+      "step": 1444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.724675034524604e-05,
+      "loss": 0.5184,
+      "step": 1445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7243098666548094e-05,
+      "loss": 0.517,
+      "step": 1446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.723944495495607e-05,
+      "loss": 0.5782,
+      "step": 1447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7235789211495455e-05,
+      "loss": 0.5171,
+      "step": 1448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.723213143719228e-05,
+      "loss": 0.5429,
+      "step": 1449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7228471633073164e-05,
+      "loss": 0.4996,
+      "step": 1450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.722480980016529e-05,
+      "loss": 0.523,
+      "step": 1451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7221145939496416e-05,
+      "loss": 0.604,
+      "step": 1452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7217480052094855e-05,
+      "loss": 0.5223,
+      "step": 1453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7213812138989505e-05,
+      "loss": 0.5559,
+      "step": 1454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7210142201209825e-05,
+      "loss": 0.5616,
+      "step": 1455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7206470239785842e-05,
+      "loss": 0.5441,
+      "step": 1456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7202796255748154e-05,
+      "loss": 0.5509,
+      "step": 1457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7199120250127922e-05,
+      "loss": 0.5611,
+      "step": 1458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7195442223956876e-05,
+      "loss": 0.5476,
+      "step": 1459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.719176217826732e-05,
+      "loss": 0.565,
+      "step": 1460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.718808011409211e-05,
+      "loss": 0.5307,
+      "step": 1461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.718439603246469e-05,
+      "loss": 0.5496,
+      "step": 1462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7180709934419047e-05,
+      "loss": 0.559,
+      "step": 1463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7177021820989746e-05,
+      "loss": 0.5849,
+      "step": 1464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7173331693211922e-05,
+      "loss": 0.5817,
+      "step": 1465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7169639552121268e-05,
+      "loss": 0.5533,
+      "step": 1466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.716594539875404e-05,
+      "loss": 0.5608,
+      "step": 1467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7162249234147063e-05,
+      "loss": 0.5828,
+      "step": 1468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.715855105933773e-05,
+      "loss": 0.546,
+      "step": 1469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7154850875363987e-05,
+      "loss": 0.5544,
+      "step": 1470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7151148683264362e-05,
+      "loss": 0.598,
+      "step": 1471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7147444484077928e-05,
+      "loss": 0.5401,
+      "step": 1472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.714373827884433e-05,
+      "loss": 0.5325,
+      "step": 1473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7140030068603777e-05,
+      "loss": 0.5642,
+      "step": 1474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7136319854397037e-05,
+      "loss": 0.5241,
+      "step": 1475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7132607637265443e-05,
+      "loss": 0.5673,
+      "step": 1476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7128893418250892e-05,
+      "loss": 0.5445,
+      "step": 1477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7125177198395835e-05,
+      "loss": 0.5406,
+      "step": 1478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.712145897874329e-05,
+      "loss": 0.5535,
+      "step": 1479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7117738760336846e-05,
+      "loss": 0.5634,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7114016544220633e-05,
+      "loss": 0.5694,
+      "step": 1481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7110292331439353e-05,
+      "loss": 0.5582,
+      "step": 1482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.710656612303827e-05,
+      "loss": 0.5437,
+      "step": 1483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7102837920063206e-05,
+      "loss": 0.5281,
+      "step": 1484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7099107723560537e-05,
+      "loss": 0.5499,
+      "step": 1485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7095375534577208e-05,
+      "loss": 0.5285,
+      "step": 1486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.709164135416072e-05,
+      "loss": 0.5201,
+      "step": 1487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.708790518335913e-05,
+      "loss": 0.5495,
+      "step": 1488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7084167023221054e-05,
+      "loss": 0.5412,
+      "step": 1489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7080426874795666e-05,
+      "loss": 0.5768,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7076684739132705e-05,
+      "loss": 0.507,
+      "step": 1491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7072940617282463e-05,
+      "loss": 0.565,
+      "step": 1492
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7069194510295785e-05,
+      "loss": 0.5262,
+      "step": 1493
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7065446419224073e-05,
+      "loss": 0.5473,
+      "step": 1494
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7061696345119304e-05,
+      "loss": 0.5041,
+      "step": 1495
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7057944289033983e-05,
+      "loss": 0.5967,
+      "step": 1496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7054190252021195e-05,
+      "loss": 0.5512,
+      "step": 1497
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.705043423513457e-05,
+      "loss": 0.5361,
+      "step": 1498
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7046676239428287e-05,
+      "loss": 0.5605,
+      "step": 1499
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7042916265957107e-05,
+      "loss": 0.5367,
+      "step": 1500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.703915431577631e-05,
+      "loss": 0.5324,
+      "step": 1501
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7035390389941757e-05,
+      "loss": 0.5222,
+      "step": 1502
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7031624489509858e-05,
+      "loss": 0.5858,
+      "step": 1503
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.702785661553757e-05,
+      "loss": 0.5545,
+      "step": 1504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.702408676908241e-05,
+      "loss": 0.5632,
+      "step": 1505
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7020314951202443e-05,
+      "loss": 0.5266,
+      "step": 1506
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7016541162956296e-05,
+      "loss": 0.5519,
+      "step": 1507
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.701276540540315e-05,
+      "loss": 0.5277,
+      "step": 1508
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.700898767960272e-05,
+      "loss": 0.5308,
+      "step": 1509
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7005207986615293e-05,
+      "loss": 0.5625,
+      "step": 1510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7001426327501702e-05,
+      "loss": 0.5516,
+      "step": 1511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.699764270332333e-05,
+      "loss": 0.5446,
+      "step": 1512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6993857115142117e-05,
+      "loss": 0.4884,
+      "step": 1513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6990069564020548e-05,
+      "loss": 0.494,
+      "step": 1514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.698628005102166e-05,
+      "loss": 0.502,
+      "step": 1515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.698248857720904e-05,
+      "loss": 0.4559,
+      "step": 1516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6978695143646827e-05,
+      "loss": 0.5284,
+      "step": 1517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6974899751399722e-05,
+      "loss": 0.5137,
+      "step": 1518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6971102401532946e-05,
+      "loss": 0.4768,
+      "step": 1519
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6967303095112297e-05,
+      "loss": 0.5221,
+      "step": 1520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6963501833204112e-05,
+      "loss": 0.4772,
+      "step": 1521
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.695969861687528e-05,
+      "loss": 0.4521,
+      "step": 1522
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6955893447193225e-05,
+      "loss": 0.4778,
+      "step": 1523
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6952086325225945e-05,
+      "loss": 0.5071,
+      "step": 1524
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6948277252041957e-05,
+      "loss": 0.4821,
+      "step": 1525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.694446622871035e-05,
+      "loss": 0.5019,
+      "step": 1526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6940653256300745e-05,
+      "loss": 0.4926,
+      "step": 1527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6936838335883316e-05,
+      "loss": 0.5114,
+      "step": 1528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6933021468528786e-05,
+      "loss": 0.5159,
+      "step": 1529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6929202655308414e-05,
+      "loss": 0.5111,
+      "step": 1530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.692538189729402e-05,
+      "loss": 0.5076,
+      "step": 1531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6921559195557958e-05,
+      "loss": 0.479,
+      "step": 1532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.691773455117313e-05,
+      "loss": 0.5125,
+      "step": 1533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.691390796521299e-05,
+      "loss": 0.4801,
+      "step": 1534
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.691007943875153e-05,
+      "loss": 0.4866,
+      "step": 1535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6906248972863285e-05,
+      "loss": 0.4519,
+      "step": 1536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6902416568623344e-05,
+      "loss": 0.4643,
+      "step": 1537
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6898582227107326e-05,
+      "loss": 0.5046,
+      "step": 1538
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.689474594939141e-05,
+      "loss": 0.4605,
+      "step": 1539
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.689090773655231e-05,
+      "loss": 0.5053,
+      "step": 1540
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6887067589667278e-05,
+      "loss": 0.5014,
+      "step": 1541
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.688322550981411e-05,
+      "loss": 0.4573,
+      "step": 1542
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6879381498071163e-05,
+      "loss": 0.4963,
+      "step": 1543
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6875535555517302e-05,
+      "loss": 0.4797,
+      "step": 1544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6871687683231975e-05,
+      "loss": 0.4845,
+      "step": 1545
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6867837882295135e-05,
+      "loss": 0.4384,
+      "step": 1546
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6863986153787297e-05,
+      "loss": 0.4952,
+      "step": 1547
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.686013249878951e-05,
+      "loss": 0.475,
+      "step": 1548
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6856276918383368e-05,
+      "loss": 0.4936,
+      "step": 1549
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6852419413651003e-05,
+      "loss": 0.4954,
+      "step": 1550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.684855998567508e-05,
+      "loss": 0.4921,
+      "step": 1551
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.684469863553882e-05,
+      "loss": 0.5014,
+      "step": 1552
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.684083536432597e-05,
+      "loss": 0.4892,
+      "step": 1553
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6836970173120818e-05,
+      "loss": 0.4969,
+      "step": 1554
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6833103063008194e-05,
+      "loss": 0.4983,
+      "step": 1555
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6829234035073464e-05,
+      "loss": 0.5034,
+      "step": 1556
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.682536309040254e-05,
+      "loss": 0.4834,
+      "step": 1557
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6821490230081864e-05,
+      "loss": 0.464,
+      "step": 1558
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6817615455198414e-05,
+      "loss": 0.4887,
+      "step": 1559
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.681373876683971e-05,
+      "loss": 0.4895,
+      "step": 1560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.680986016609381e-05,
+      "loss": 0.4953,
+      "step": 1561
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6805979654049305e-05,
+      "loss": 0.4882,
+      "step": 1562
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.6802097231795324e-05,
+      "loss": 0.5271,
+      "step": 1563
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6798212900421536e-05,
+      "loss": 0.4953,
+      "step": 1564
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6794326661018136e-05,
+      "loss": 0.4628,
+      "step": 1565
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6790438514675865e-05,
+      "loss": 0.4971,
+      "step": 1566
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6786548462485994e-05,
+      "loss": 0.4879,
+      "step": 1567
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.678265650554033e-05,
+      "loss": 0.4589,
+      "step": 1568
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6778762644931213e-05,
+      "loss": 0.4851,
+      "step": 1569
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6774866881751518e-05,
+      "loss": 0.4966,
+      "step": 1570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6770969217094657e-05,
+      "loss": 0.5129,
+      "step": 1571
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.676706965205457e-05,
+      "loss": 0.5025,
+      "step": 1572
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6763168187725742e-05,
+      "loss": 0.5246,
+      "step": 1573
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6759264825203172e-05,
+      "loss": 0.4962,
+      "step": 1574
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6755359565582408e-05,
+      "loss": 0.4804,
+      "step": 1575
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.6751452409959527e-05,
+      "loss": 0.4765,
+      "step": 1576
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.674754335943113e-05,
+      "loss": 0.5119,
+      "step": 1577
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.674363241509436e-05,
+      "loss": 0.5444,
+      "step": 1578
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6739719578046888e-05,
+      "loss": 0.4934,
+      "step": 1579
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6735804849386914e-05,
+      "loss": 0.5107,
+      "step": 1580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6731888230213172e-05,
+      "loss": 0.5445,
+      "step": 1581
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6727969721624923e-05,
+      "loss": 0.5188,
+      "step": 1582
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.672404932472196e-05,
+      "loss": 0.5017,
+      "step": 1583
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.672012704060461e-05,
+      "loss": 0.5127,
+      "step": 1584
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6716202870373726e-05,
+      "loss": 0.4733,
+      "step": 1585
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6712276815130688e-05,
+      "loss": 0.5149,
+      "step": 1586
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.670834887597741e-05,
+      "loss": 0.4659,
+      "step": 1587
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.670441905401633e-05,
+      "loss": 0.4702,
+      "step": 1588
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6700487350350416e-05,
+      "loss": 0.4714,
+      "step": 1589
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6696553766083167e-05,
+      "loss": 0.48,
+      "step": 1590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.669261830231861e-05,
+      "loss": 0.4869,
+      "step": 1591
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.6688680960161292e-05,
+      "loss": 0.4932,
+      "step": 1592
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.66847417407163e-05,
+      "loss": 0.5002,
+      "step": 1593
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.668080064508923e-05,
+      "loss": 0.4699,
+      "step": 1594
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.667685767438622e-05,
+      "loss": 0.4734,
+      "step": 1595
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6672912829713936e-05,
+      "loss": 0.506,
+      "step": 1596
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.666896611217955e-05,
+      "loss": 0.5154,
+      "step": 1597
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6665017522890786e-05,
+      "loss": 0.4628,
+      "step": 1598
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6661067062955868e-05,
+      "loss": 0.4881,
+      "step": 1599
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6657114733483564e-05,
+      "loss": 0.5222,
+      "step": 1600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.665316053558316e-05,
+      "loss": 0.4997,
+      "step": 1601
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6649204470364467e-05,
+      "loss": 0.4897,
+      "step": 1602
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6645246538937815e-05,
+      "loss": 0.4963,
+      "step": 1603
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6641286742414066e-05,
+      "loss": 0.4787,
+      "step": 1604
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6637325081904595e-05,
+      "loss": 0.4995,
+      "step": 1605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.663336155852132e-05,
+      "loss": 0.5118,
+      "step": 1606
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.6629396173376656e-05,
+      "loss": 0.5022,
+      "step": 1607
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.662542892758356e-05,
+      "loss": 0.5062,
+      "step": 1608
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.66214598222555e-05,
+      "loss": 0.5097,
+      "step": 1609
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6617488858506478e-05,
+      "loss": 0.4886,
+      "step": 1610
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6613516037451e-05,
+      "loss": 0.4908,
+      "step": 1611
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6609541360204108e-05,
+      "loss": 0.4705,
+      "step": 1612
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.660556482788136e-05,
+      "loss": 0.5641,
+      "step": 1613
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6601586441598834e-05,
+      "loss": 0.501,
+      "step": 1614
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.659760620247313e-05,
+      "loss": 0.4935,
+      "step": 1615
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.659362411162137e-05,
+      "loss": 0.5196,
+      "step": 1616
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6589640170161188e-05,
+      "loss": 0.4981,
+      "step": 1617
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6585654379210745e-05,
+      "loss": 0.5202,
+      "step": 1618
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.6581666739888715e-05,
+      "loss": 0.4837,
+      "step": 1619
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.65776772533143e-05,
+      "loss": 0.4857,
+      "step": 1620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.657368592060721e-05,
+      "loss": 0.4868,
+      "step": 1621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.656969274288768e-05,
+      "loss": 0.5207,
+      "step": 1622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.656569772127646e-05,
+      "loss": 0.5117,
+      "step": 1623
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6561700856894817e-05,
+      "loss": 0.5011,
+      "step": 1624
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6557702150864538e-05,
+      "loss": 0.4876,
+      "step": 1625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6553701604307924e-05,
+      "loss": 0.5139,
+      "step": 1626
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6549699218347796e-05,
+      "loss": 0.4967,
+      "step": 1627
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.654569499410749e-05,
+      "loss": 0.4945,
+      "step": 1628
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.654168893271085e-05,
+      "loss": 0.498,
+      "step": 1629
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6537681035282247e-05,
+      "loss": 0.4923,
+      "step": 1630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6533671302946566e-05,
+      "loss": 0.5004,
+      "step": 1631
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6529659736829197e-05,
+      "loss": 0.5208,
+      "step": 1632
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.652564633805606e-05,
+      "loss": 0.4744,
+      "step": 1633
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6521631107753575e-05,
+      "loss": 0.5152,
+      "step": 1634
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6517614047048683e-05,
+      "loss": 0.5014,
+      "step": 1635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.6513595157068837e-05,
+      "loss": 0.4904,
+      "step": 1636
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.650957443894201e-05,
+      "loss": 0.4434,
+      "step": 1637
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6505551893796673e-05,
+      "loss": 0.481,
+      "step": 1638
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6501527522761828e-05,
+      "loss": 0.497,
+      "step": 1639
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6497501326966974e-05,
+      "loss": 0.5048,
+      "step": 1640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6493473307542132e-05,
+      "loss": 0.509,
+      "step": 1641
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6489443465617832e-05,
+      "loss": 0.4771,
+      "step": 1642
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.648541180232511e-05,
+      "loss": 0.5047,
+      "step": 1643
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6481378318795528e-05,
+      "loss": 0.4742,
+      "step": 1644
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6477343016161138e-05,
+      "loss": 0.4916,
+      "step": 1645
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6473305895554522e-05,
+      "loss": 0.4914,
+      "step": 1646
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6469266958108757e-05,
+      "loss": 0.4415,
+      "step": 1647
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6465226204957444e-05,
+      "loss": 0.4514,
+      "step": 1648
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.646118363723468e-05,
+      "loss": 0.5003,
+      "step": 1649
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6457139256075084e-05,
+      "loss": 0.4971,
+      "step": 1650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6453093062613774e-05,
+      "loss": 0.4783,
+      "step": 1651
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.6449045057986376e-05,
+      "loss": 0.4858,
+      "step": 1652
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.644499524332904e-05,
+      "loss": 0.5081,
+      "step": 1653
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6440943619778403e-05,
+      "loss": 0.5069,
+      "step": 1654
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6436890188471622e-05,
+      "loss": 0.5034,
+      "step": 1655
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.643283495054636e-05,
+      "loss": 0.4961,
+      "step": 1656
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.642877790714078e-05,
+      "loss": 0.4751,
+      "step": 1657
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.642471905939357e-05,
+      "loss": 0.4604,
+      "step": 1658
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6420658408443904e-05,
+      "loss": 0.4928,
+      "step": 1659
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6416595955431468e-05,
+      "loss": 0.4956,
+      "step": 1660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.641253170149646e-05,
+      "loss": 0.5046,
+      "step": 1661
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6408465647779578e-05,
+      "loss": 0.4758,
+      "step": 1662
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6404397795422024e-05,
+      "loss": 0.4873,
+      "step": 1663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.640032814556551e-05,
+      "loss": 0.4867,
+      "step": 1664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6396256699352252e-05,
+      "loss": 0.4854,
+      "step": 1665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6392183457924967e-05,
+      "loss": 0.4952,
+      "step": 1666
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6388108422426873e-05,
+      "loss": 0.4483,
+      "step": 1667
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6384031594001698e-05,
+      "loss": 0.4868,
+      "step": 1668
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.637995297379367e-05,
+      "loss": 0.5239,
+      "step": 1669
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6375872562947516e-05,
+      "loss": 0.5117,
+      "step": 1670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.637179036260848e-05,
+      "loss": 0.5197,
+      "step": 1671
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.636770637392229e-05,
+      "loss": 0.4781,
+      "step": 1672
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.636362059803519e-05,
+      "loss": 0.4938,
+      "step": 1673
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6359533036093915e-05,
+      "loss": 0.5139,
+      "step": 1674
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.635544368924571e-05,
+      "loss": 0.5368,
+      "step": 1675
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6351352558638313e-05,
+      "loss": 0.5025,
+      "step": 1676
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6347259645419966e-05,
+      "loss": 0.4668,
+      "step": 1677
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6343164950739417e-05,
+      "loss": 0.4899,
+      "step": 1678
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.633906847574591e-05,
+      "loss": 0.5279,
+      "step": 1679
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6334970221589182e-05,
+      "loss": 0.4785,
+      "step": 1680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6330870189419475e-05,
+      "loss": 0.5061,
+      "step": 1681
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6326768380387538e-05,
+      "loss": 0.4853,
+      "step": 1682
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6322664795644604e-05,
+      "loss": 0.495,
+      "step": 1683
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.631855943634241e-05,
+      "loss": 0.4349,
+      "step": 1684
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6314452303633193e-05,
+      "loss": 0.5487,
+      "step": 1685
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6310343398669693e-05,
+      "loss": 0.4936,
+      "step": 1686
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6306232722605133e-05,
+      "loss": 0.4722,
+      "step": 1687
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6302120276593248e-05,
+      "loss": 0.5042,
+      "step": 1688
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6298006061788264e-05,
+      "loss": 0.5306,
+      "step": 1689
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6293890079344892e-05,
+      "loss": 0.5045,
+      "step": 1690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6289772330418365e-05,
+      "loss": 0.4998,
+      "step": 1691
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6285652816164384e-05,
+      "loss": 0.5169,
+      "step": 1692
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6281531537739163e-05,
+      "loss": 0.5023,
+      "step": 1693
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6277408496299406e-05,
+      "loss": 0.4869,
+      "step": 1694
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6273283693002312e-05,
+      "loss": 0.5068,
+      "step": 1695
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6269157129005573e-05,
+      "loss": 0.4735,
+      "step": 1696
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.626502880546738e-05,
+      "loss": 0.48,
+      "step": 1697
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6260898723546416e-05,
+      "loss": 0.5318,
+      "step": 1698
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.625676688440185e-05,
+      "loss": 0.5035,
+      "step": 1699
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.625263328919335e-05,
+      "loss": 0.4679,
+      "step": 1700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6248497939081082e-05,
+      "loss": 0.5333,
+      "step": 1701
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6244360835225698e-05,
+      "loss": 0.4487,
+      "step": 1702
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.624022197878834e-05,
+      "loss": 0.4921,
+      "step": 1703
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.623608137093065e-05,
+      "loss": 0.4767,
+      "step": 1704
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6231939012814758e-05,
+      "loss": 0.4884,
+      "step": 1705
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.622779490560328e-05,
+      "loss": 0.4936,
+      "step": 1706
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6223649050459337e-05,
+      "loss": 0.5619,
+      "step": 1707
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.621950144854652e-05,
+      "loss": 0.5088,
+      "step": 1708
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6215352101028926e-05,
+      "loss": 0.5222,
+      "step": 1709
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6211201009071134e-05,
+      "loss": 0.4876,
+      "step": 1710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6207048173838226e-05,
+      "loss": 0.4759,
+      "step": 1711
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.620289359649575e-05,
+      "loss": 0.5471,
+      "step": 1712
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6198737278209763e-05,
+      "loss": 0.4909,
+      "step": 1713
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6194579220146806e-05,
+      "loss": 0.5141,
+      "step": 1714
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6190419423473897e-05,
+      "loss": 0.5057,
+      "step": 1715
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6186257889358557e-05,
+      "loss": 0.5086,
+      "step": 1716
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6182094618968793e-05,
+      "loss": 0.4666,
+      "step": 1717
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6177929613473088e-05,
+      "loss": 0.5046,
+      "step": 1718
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.617376287404042e-05,
+      "loss": 0.4704,
+      "step": 1719
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6169594401840255e-05,
+      "loss": 0.4911,
+      "step": 1720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6165424198042542e-05,
+      "loss": 0.4914,
+      "step": 1721
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6161252263817715e-05,
+      "loss": 0.4775,
+      "step": 1722
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6157078600336693e-05,
+      "loss": 0.4704,
+      "step": 1723
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6152903208770888e-05,
+      "loss": 0.4702,
+      "step": 1724
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6148726090292196e-05,
+      "loss": 0.5048,
+      "step": 1725
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6144547246072984e-05,
+      "loss": 0.4896,
+      "step": 1726
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.614036667728612e-05,
+      "loss": 0.506,
+      "step": 1727
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6136184385104945e-05,
+      "loss": 0.4839,
+      "step": 1728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6132000370703286e-05,
+      "loss": 0.4753,
+      "step": 1729
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6127814635255462e-05,
+      "loss": 0.4943,
+      "step": 1730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6123627179936262e-05,
+      "loss": 0.5303,
+      "step": 1731
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6119438005920968e-05,
+      "loss": 0.4707,
+      "step": 1732
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.611524711438533e-05,
+      "loss": 0.5114,
+      "step": 1733
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6111054506505607e-05,
+      "loss": 0.4873,
+      "step": 1734
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6106860183458514e-05,
+      "loss": 0.4573,
+      "step": 1735
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6102664146421255e-05,
+      "loss": 0.4946,
+      "step": 1736
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6098466396571514e-05,
+      "loss": 0.5046,
+      "step": 1737
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6094266935087467e-05,
+      "loss": 0.5017,
+      "step": 1738
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6090065763147755e-05,
+      "loss": 0.5371,
+      "step": 1739
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.608586288193151e-05,
+      "loss": 0.4745,
+      "step": 1740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.608165829261833e-05,
+      "loss": 0.4906,
+      "step": 1741
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6077451996388314e-05,
+      "loss": 0.4589,
+      "step": 1742
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6073243994422018e-05,
+      "loss": 0.4641,
+      "step": 1743
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6069034287900493e-05,
+      "loss": 0.4957,
+      "step": 1744
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6064822878005262e-05,
+      "loss": 0.479,
+      "step": 1745
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.606060976591832e-05,
+      "loss": 0.5074,
+      "step": 1746
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.605639495282215e-05,
+      "loss": 0.4891,
+      "step": 1747
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6052178439899712e-05,
+      "loss": 0.488,
+      "step": 1748
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6047960228334428e-05,
+      "loss": 0.4882,
+      "step": 1749
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6043740319310218e-05,
+      "loss": 0.4828,
+      "step": 1750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6039518714011465e-05,
+      "loss": 0.4835,
+      "step": 1751
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6035295413623032e-05,
+      "loss": 0.5265,
+      "step": 1752
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6031070419330258e-05,
+      "loss": 0.4545,
+      "step": 1753
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6026843732318958e-05,
+      "loss": 0.4948,
+      "step": 1754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.602261535377542e-05,
+      "loss": 0.5852,
+      "step": 1755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.60183852848864e-05,
+      "loss": 0.4922,
+      "step": 1756
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.601415352683915e-05,
+      "loss": 0.5126,
+      "step": 1757
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6009920080821365e-05,
+      "loss": 0.4729,
+      "step": 1758
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6005684948021248e-05,
+      "loss": 0.5047,
+      "step": 1759
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.600144812962745e-05,
+      "loss": 0.4839,
+      "step": 1760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5997209626829105e-05,
+      "loss": 0.4733,
+      "step": 1761
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5992969440815813e-05,
+      "loss": 0.4462,
+      "step": 1762
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.598872757277766e-05,
+      "loss": 0.4996,
+      "step": 1763
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5984484023905186e-05,
+      "loss": 0.4953,
+      "step": 1764
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5980238795389424e-05,
+      "loss": 0.4991,
+      "step": 1765
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5975991888421857e-05,
+      "loss": 0.4656,
+      "step": 1766
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.597174330419445e-05,
+      "loss": 0.4827,
+      "step": 1767
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5967493043899644e-05,
+      "loss": 0.5212,
+      "step": 1768
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5963241108730342e-05,
+      "loss": 0.5176,
+      "step": 1769
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.595898749987991e-05,
+      "loss": 0.478,
+      "step": 1770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5954732218542207e-05,
+      "loss": 0.4943,
+      "step": 1771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5950475265911537e-05,
+      "loss": 0.5017,
+      "step": 1772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5946216643182685e-05,
+      "loss": 0.5064,
+      "step": 1773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5941956351550908e-05,
+      "loss": 0.4653,
+      "step": 1774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5937694392211923e-05,
+      "loss": 0.5459,
+      "step": 1775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.593343076636192e-05,
+      "loss": 0.5163,
+      "step": 1776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5929165475197553e-05,
+      "loss": 0.4803,
+      "step": 1777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5924898519915947e-05,
+      "loss": 0.4616,
+      "step": 1778
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.592062990171469e-05,
+      "loss": 0.5039,
+      "step": 1779
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5916359621791847e-05,
+      "loss": 0.4811,
+      "step": 1780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5912087681345934e-05,
+      "loss": 0.488,
+      "step": 1781
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5907814081575943e-05,
+      "loss": 0.4915,
+      "step": 1782
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.590353882368133e-05,
+      "loss": 0.4763,
+      "step": 1783
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.589926190886202e-05,
+      "loss": 0.5249,
+      "step": 1784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5894983338318396e-05,
+      "loss": 0.4959,
+      "step": 1785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5890703113251305e-05,
+      "loss": 0.4974,
+      "step": 1786
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5886421234862072e-05,
+      "loss": 0.4823,
+      "step": 1787
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5882137704352466e-05,
+      "loss": 0.5042,
+      "step": 1788
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.5092,
+      "step": 1789
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.587356569178158e-05,
+      "loss": 0.5036,
+      "step": 1790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.586927721212618e-05,
+      "loss": 0.4688,
+      "step": 1791
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5864987085162155e-05,
+      "loss": 0.5075,
+      "step": 1792
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5860695312093608e-05,
+      "loss": 0.4574,
+      "step": 1793
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5856401894125095e-05,
+      "loss": 0.49,
+      "step": 1794
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.585210683246163e-05,
+      "loss": 0.4857,
+      "step": 1795
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5847810128308695e-05,
+      "loss": 0.5273,
+      "step": 1796
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5843511782872226e-05,
+      "loss": 0.4779,
+      "step": 1797
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.583921179735863e-05,
+      "loss": 0.5096,
+      "step": 1798
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5834910172974767e-05,
+      "loss": 0.5149,
+      "step": 1799
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5830606910927956e-05,
+      "loss": 0.5018,
+      "step": 1800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5826302012425977e-05,
+      "loss": 0.5022,
+      "step": 1801
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.582199547867707e-05,
+      "loss": 0.4903,
+      "step": 1802
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.581768731088994e-05,
+      "loss": 0.4917,
+      "step": 1803
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.581337751027374e-05,
+      "loss": 0.4503,
+      "step": 1804
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5809066078038082e-05,
+      "loss": 0.4935,
+      "step": 1805
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5804753015393045e-05,
+      "loss": 0.4633,
+      "step": 1806
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5800438323549167e-05,
+      "loss": 0.4962,
+      "step": 1807
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5796122003717424e-05,
+      "loss": 0.4695,
+      "step": 1808
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5791804057109266e-05,
+      "loss": 0.4752,
+      "step": 1809
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.57874844849366e-05,
+      "loss": 0.5203,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.578316328841178e-05,
+      "loss": 0.5127,
+      "step": 1811
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5778840468747628e-05,
+      "loss": 0.4468,
+      "step": 1812
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.57745160271574e-05,
+      "loss": 0.4872,
+      "step": 1813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5770189964854834e-05,
+      "loss": 0.4886,
+      "step": 1814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5765862283054105e-05,
+      "loss": 0.5162,
+      "step": 1815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.576153298296985e-05,
+      "loss": 0.4946,
+      "step": 1816
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.575720206581715e-05,
+      "loss": 0.5295,
+      "step": 1817
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5752869532811555e-05,
+      "loss": 0.513,
+      "step": 1818
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5748535385169062e-05,
+      "loss": 0.5128,
+      "step": 1819
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5744199624106115e-05,
+      "loss": 0.482,
+      "step": 1820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5739862250839623e-05,
+      "loss": 0.4437,
+      "step": 1821
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5735523266586935e-05,
+      "loss": 0.5026,
+      "step": 1822
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5731182672565865e-05,
+      "loss": 0.4651,
+      "step": 1823
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5726840469994658e-05,
+      "loss": 0.5238,
+      "step": 1824
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.572249666009204e-05,
+      "loss": 0.4972,
+      "step": 1825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5718151244077162e-05,
+      "loss": 0.4722,
+      "step": 1826
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.571380422316964e-05,
+      "loss": 0.5043,
+      "step": 1827
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5709455598589537e-05,
+      "loss": 0.436,
+      "step": 1828
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5705105371557362e-05,
+      "loss": 0.4849,
+      "step": 1829
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.570075354329408e-05,
+      "loss": 0.4757,
+      "step": 1830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5696400115021102e-05,
+      "loss": 0.4852,
+      "step": 1831
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5692045087960294e-05,
+      "loss": 0.4869,
+      "step": 1832
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5687688463333954e-05,
+      "loss": 0.5459,
+      "step": 1833
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.568333024236485e-05,
+      "loss": 0.4779,
+      "step": 1834
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5678970426276186e-05,
+      "loss": 0.508,
+      "step": 1835
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5674609016291613e-05,
+      "loss": 0.4839,
+      "step": 1836
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5670246013635232e-05,
+      "loss": 0.4917,
+      "step": 1837
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5665881419531593e-05,
+      "loss": 0.5003,
+      "step": 1838
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.566151523520569e-05,
+      "loss": 0.4802,
+      "step": 1839
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5657147461882965e-05,
+      "loss": 0.49,
+      "step": 1840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5652778100789304e-05,
+      "loss": 0.4841,
+      "step": 1841
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.564840715315104e-05,
+      "loss": 0.4745,
+      "step": 1842
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5644034620194953e-05,
+      "loss": 0.5214,
+      "step": 1843
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.563966050314826e-05,
+      "loss": 0.4715,
+      "step": 1844
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5635284803238632e-05,
+      "loss": 0.4653,
+      "step": 1845
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5630907521694184e-05,
+      "loss": 0.4968,
+      "step": 1846
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5626528659743466e-05,
+      "loss": 0.507,
+      "step": 1847
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.562214821861548e-05,
+      "loss": 0.4981,
+      "step": 1848
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.561776619953967e-05,
+      "loss": 0.4915,
+      "step": 1849
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5613382603745918e-05,
+      "loss": 0.5093,
+      "step": 1850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.560899743246455e-05,
+      "loss": 0.49,
+      "step": 1851
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5604610686926346e-05,
+      "loss": 0.4908,
+      "step": 1852
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5600222368362506e-05,
+      "loss": 0.5194,
+      "step": 1853
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5595832478004685e-05,
+      "loss": 0.4776,
+      "step": 1854
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.559144101708499e-05,
+      "loss": 0.4756,
+      "step": 1855
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5587047986835942e-05,
+      "loss": 0.4733,
+      "step": 1856
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.558265338849052e-05,
+      "loss": 0.5164,
+      "step": 1857
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5578257223282146e-05,
+      "loss": 0.5059,
+      "step": 1858
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5573859492444672e-05,
+      "loss": 0.4619,
+      "step": 1859
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.556946019721239e-05,
+      "loss": 0.4917,
+      "step": 1860
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.556505933882004e-05,
+      "loss": 0.5133,
+      "step": 1861
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5560656918502787e-05,
+      "loss": 0.5143,
+      "step": 1862
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.555625293749625e-05,
+      "loss": 0.4908,
+      "step": 1863
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5551847397036476e-05,
+      "loss": 0.465,
+      "step": 1864
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5547440298359948e-05,
+      "loss": 0.5334,
+      "step": 1865
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5543031642703594e-05,
+      "loss": 0.495,
+      "step": 1866
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.553862143130478e-05,
+      "loss": 0.4917,
+      "step": 1867
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.553420966540129e-05,
+      "loss": 0.5022,
+      "step": 1868
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5529796346231376e-05,
+      "loss": 0.4532,
+      "step": 1869
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5525381475033692e-05,
+      "loss": 0.4638,
+      "step": 1870
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5520965053047353e-05,
+      "loss": 0.5026,
+      "step": 1871
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.55165470815119e-05,
+      "loss": 0.4906,
+      "step": 1872
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5512127561667304e-05,
+      "loss": 0.4698,
+      "step": 1873
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.550770649475398e-05,
+      "loss": 0.4757,
+      "step": 1874
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.550328388201277e-05,
+      "loss": 0.4805,
+      "step": 1875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5498859724684953e-05,
+      "loss": 0.4983,
+      "step": 1876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5494434024012247e-05,
+      "loss": 0.5818,
+      "step": 1877
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5490006781236785e-05,
+      "loss": 0.4802,
+      "step": 1878
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5485577997601158e-05,
+      "loss": 0.4864,
+      "step": 1879
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5481147674348366e-05,
+      "loss": 0.4753,
+      "step": 1880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.547671581272186e-05,
+      "loss": 0.5381,
+      "step": 1881
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5472282413965508e-05,
+      "loss": 0.5472,
+      "step": 1882
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5467847479323622e-05,
+      "loss": 0.5016,
+      "step": 1883
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.546341101004093e-05,
+      "loss": 0.4962,
+      "step": 1884
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.545897300736261e-05,
+      "loss": 0.514,
+      "step": 1885
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5454533472534253e-05,
+      "loss": 0.5045,
+      "step": 1886
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5450092406801892e-05,
+      "loss": 0.464,
+      "step": 1887
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.544564981141198e-05,
+      "loss": 0.4967,
+      "step": 1888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5441205687611403e-05,
+      "loss": 0.5189,
+      "step": 1889
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5436760036647485e-05,
+      "loss": 0.4891,
+      "step": 1890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5432312859767963e-05,
+      "loss": 0.483,
+      "step": 1891
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5427864158221015e-05,
+      "loss": 0.5056,
+      "step": 1892
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5423413933255237e-05,
+      "loss": 0.4923,
+      "step": 1893
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.541896218611966e-05,
+      "loss": 0.497,
+      "step": 1894
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.541450891806374e-05,
+      "loss": 0.5341,
+      "step": 1895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5410054130337358e-05,
+      "loss": 0.5075,
+      "step": 1896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5405597824190822e-05,
+      "loss": 0.5031,
+      "step": 1897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5401140000874873e-05,
+      "loss": 0.5042,
+      "step": 1898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5396680661640667e-05,
+      "loss": 0.4972,
+      "step": 1899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.539221980773979e-05,
+      "loss": 0.487,
+      "step": 1900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.538775744042426e-05,
+      "loss": 0.4908,
+      "step": 1901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5383293560946505e-05,
+      "loss": 0.5037,
+      "step": 1902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5378828170559387e-05,
+      "loss": 0.5053,
+      "step": 1903
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5374361270516197e-05,
+      "loss": 0.467,
+      "step": 1904
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5369892862070636e-05,
+      "loss": 0.5163,
+      "step": 1905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5365422946476842e-05,
+      "loss": 0.4656,
+      "step": 1906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5360951524989367e-05,
+      "loss": 0.4932,
+      "step": 1907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5356478598863187e-05,
+      "loss": 0.4975,
+      "step": 1908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5352004169353706e-05,
+      "loss": 0.4963,
+      "step": 1909
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5347528237716742e-05,
+      "loss": 0.453,
+      "step": 1910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5343050805208543e-05,
+      "loss": 0.4866,
+      "step": 1911
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.533857187308577e-05,
+      "loss": 0.4702,
+      "step": 1912
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.533409144260551e-05,
+      "loss": 0.5095,
+      "step": 1913
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5329609515025262e-05,
+      "loss": 0.4992,
+      "step": 1914
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5325126091602965e-05,
+      "loss": 0.4683,
+      "step": 1915
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.532064117359696e-05,
+      "loss": 0.4832,
+      "step": 1916
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5316154762266008e-05,
+      "loss": 0.4995,
+      "step": 1917
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5311666858869296e-05,
+      "loss": 0.4624,
+      "step": 1918
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.530717746466643e-05,
+      "loss": 0.4787,
+      "step": 1919
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5302686580917428e-05,
+      "loss": 0.4613,
+      "step": 1920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5298194208882735e-05,
+      "loss": 0.4764,
+      "step": 1921
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5293700349823203e-05,
+      "loss": 0.4643,
+      "step": 1922
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5289205005000113e-05,
+      "loss": 0.4731,
+      "step": 1923
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5284708175675153e-05,
+      "loss": 0.485,
+      "step": 1924
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.528020986311043e-05,
+      "loss": 0.4789,
+      "step": 1925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5275710068568477e-05,
+      "loss": 0.4775,
+      "step": 1926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5271208793312226e-05,
+      "loss": 0.4853,
+      "step": 1927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5266706038605038e-05,
+      "loss": 0.5434,
+      "step": 1928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5262201805710683e-05,
+      "loss": 0.5327,
+      "step": 1929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.525769609589335e-05,
+      "loss": 0.4669,
+      "step": 1930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5253188910417636e-05,
+      "loss": 0.466,
+      "step": 1931
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5248680250548558e-05,
+      "loss": 0.4469,
+      "step": 1932
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.524417011755155e-05,
+      "loss": 0.5053,
+      "step": 1933
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5239658512692447e-05,
+      "loss": 0.4796,
+      "step": 1934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.523514543723751e-05,
+      "loss": 0.4903,
+      "step": 1935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5230630892453407e-05,
+      "loss": 0.4798,
+      "step": 1936
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5226114879607215e-05,
+      "loss": 0.4407,
+      "step": 1937
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.522159739996643e-05,
+      "loss": 0.5244,
+      "step": 1938
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5217078454798952e-05,
+      "loss": 0.4863,
+      "step": 1939
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5212558045373106e-05,
+      "loss": 0.473,
+      "step": 1940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5208036172957612e-05,
+      "loss": 0.5057,
+      "step": 1941
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5203512838821609e-05,
+      "loss": 0.4693,
+      "step": 1942
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5198988044234644e-05,
+      "loss": 0.503,
+      "step": 1943
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5194461790466674e-05,
+      "loss": 0.502,
+      "step": 1944
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5189934078788069e-05,
+      "loss": 0.493,
+      "step": 1945
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5185404910469604e-05,
+      "loss": 0.4808,
+      "step": 1946
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5180874286782464e-05,
+      "loss": 0.4626,
+      "step": 1947
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.517634220899824e-05,
+      "loss": 0.4917,
+      "step": 1948
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5171808678388934e-05,
+      "loss": 0.4725,
+      "step": 1949
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5167273696226965e-05,
+      "loss": 0.4932,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.516273726378514e-05,
+      "loss": 0.4965,
+      "step": 1951
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5158199382336678e-05,
+      "loss": 0.5235,
+      "step": 1952
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5153660053155227e-05,
+      "loss": 0.5067,
+      "step": 1953
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5149119277514808e-05,
+      "loss": 0.4946,
+      "step": 1954
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5144577056689872e-05,
+      "loss": 0.5235,
+      "step": 1955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5140033391955266e-05,
+      "loss": 0.4944,
+      "step": 1956
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5135488284586241e-05,
+      "loss": 0.4987,
+      "step": 1957
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.513094173585846e-05,
+      "loss": 0.5327,
+      "step": 1958
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5126393747047983e-05,
+      "loss": 0.5004,
+      "step": 1959
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.512184431943128e-05,
+      "loss": 0.4803,
+      "step": 1960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5117293454285217e-05,
+      "loss": 0.5076,
+      "step": 1961
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5112741152887078e-05,
+      "loss": 0.5194,
+      "step": 1962
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5108187416514533e-05,
+      "loss": 0.4837,
+      "step": 1963
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5103632246445666e-05,
+      "loss": 0.4671,
+      "step": 1964
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5099075643958959e-05,
+      "loss": 0.5293,
+      "step": 1965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5094517610333294e-05,
+      "loss": 0.5148,
+      "step": 1966
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5089958146847965e-05,
+      "loss": 0.5091,
+      "step": 1967
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5085397254782655e-05,
+      "loss": 0.4895,
+      "step": 1968
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.508083493541745e-05,
+      "loss": 0.5222,
+      "step": 1969
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5076271190032845e-05,
+      "loss": 0.488,
+      "step": 1970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.507170601990973e-05,
+      "loss": 0.492,
+      "step": 1971
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5067139426329389e-05,
+      "loss": 0.466,
+      "step": 1972
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5062571410573515e-05,
+      "loss": 0.5013,
+      "step": 1973
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5058001973924197e-05,
+      "loss": 0.5407,
+      "step": 1974
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5053431117663922e-05,
+      "loss": 0.5054,
+      "step": 1975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5048858843075573e-05,
+      "loss": 0.4888,
+      "step": 1976
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5044285151442437e-05,
+      "loss": 0.4669,
+      "step": 1977
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5039710044048192e-05,
+      "loss": 0.5075,
+      "step": 1978
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5035133522176916e-05,
+      "loss": 0.5005,
+      "step": 1979
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5030555587113091e-05,
+      "loss": 0.5604,
+      "step": 1980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5025976240141585e-05,
+      "loss": 0.5057,
+      "step": 1981
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5021395482547665e-05,
+      "loss": 0.494,
+      "step": 1982
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5016813315616998e-05,
+      "loss": 0.4726,
+      "step": 1983
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5012229740635644e-05,
+      "loss": 0.4818,
+      "step": 1984
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5007644758890059e-05,
+      "loss": 0.4719,
+      "step": 1985
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5003058371667087e-05,
+      "loss": 0.52,
+      "step": 1986
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4998470580253981e-05,
+      "loss": 0.4593,
+      "step": 1987
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4993881385938376e-05,
+      "loss": 0.5065,
+      "step": 1988
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4989290790008304e-05,
+      "loss": 0.4921,
+      "step": 1989
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4984698793752193e-05,
+      "loss": 0.5069,
+      "step": 1990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.498010539845886e-05,
+      "loss": 0.4687,
+      "step": 1991
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4975510605417514e-05,
+      "loss": 0.4661,
+      "step": 1992
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4970914415917764e-05,
+      "loss": 0.5055,
+      "step": 1993
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4966316831249601e-05,
+      "loss": 0.4628,
+      "step": 1994
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4961717852703417e-05,
+      "loss": 0.5247,
+      "step": 1995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4957117481569987e-05,
+      "loss": 0.4945,
+      "step": 1996
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4952515719140482e-05,
+      "loss": 0.5304,
+      "step": 1997
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4947912566706459e-05,
+      "loss": 0.5036,
+      "step": 1998
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4943308025559871e-05,
+      "loss": 0.488,
+      "step": 1999
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4938702096993057e-05,
+      "loss": 0.4976,
+      "step": 2000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4934094782298747e-05,
+      "loss": 0.4887,
+      "step": 2001
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4929486082770059e-05,
+      "loss": 0.5295,
+      "step": 2002
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4924875999700499e-05,
+      "loss": 0.4886,
+      "step": 2003
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4920264534383962e-05,
+      "loss": 0.4646,
+      "step": 2004
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4915651688114733e-05,
+      "loss": 0.5137,
+      "step": 2005
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.491103746218748e-05,
+      "loss": 0.4779,
+      "step": 2006
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.490642185789726e-05,
+      "loss": 0.4943,
+      "step": 2007
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4901804876539522e-05,
+      "loss": 0.52,
+      "step": 2008
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4897186519410095e-05,
+      "loss": 0.4914,
+      "step": 2009
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.48925667878052e-05,
+      "loss": 0.4862,
+      "step": 2010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4887945683021436e-05,
+      "loss": 0.5101,
+      "step": 2011
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4883323206355791e-05,
+      "loss": 0.4578,
+      "step": 2012
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4878699359105641e-05,
+      "loss": 0.4595,
+      "step": 2013
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4874074142568741e-05,
+      "loss": 0.5145,
+      "step": 2014
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.486944755804324e-05,
+      "loss": 0.4861,
+      "step": 2015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4864819606827664e-05,
+      "loss": 0.5072,
+      "step": 2016
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4860190290220913e-05,
+      "loss": 0.5119,
+      "step": 2017
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4855559609522292e-05,
+      "loss": 0.4695,
+      "step": 2018
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4850927566031472e-05,
+      "loss": 0.4962,
+      "step": 2019
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.484629416104851e-05,
+      "loss": 0.4925,
+      "step": 2020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4841659395873852e-05,
+      "loss": 0.4892,
+      "step": 2021
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4837023271808317e-05,
+      "loss": 0.5147,
+      "step": 2022
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.483238579015311e-05,
+      "loss": 0.4766,
+      "step": 2023
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4827746952209816e-05,
+      "loss": 0.5117,
+      "step": 2024
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4823106759280404e-05,
+      "loss": 0.4925,
+      "step": 2025
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4818465212667213e-05,
+      "loss": 0.5164,
+      "step": 2026
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4813822313672974e-05,
+      "loss": 0.4803,
+      "step": 2027
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.480917806360079e-05,
+      "loss": 0.4733,
+      "step": 2028
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4804532463754148e-05,
+      "loss": 0.4636,
+      "step": 2029
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4799885515436912e-05,
+      "loss": 0.4766,
+      "step": 2030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4795237219953323e-05,
+      "loss": 0.497,
+      "step": 2031
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4790587578607998e-05,
+      "loss": 0.5008,
+      "step": 2032
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4785936592705938e-05,
+      "loss": 0.4939,
+      "step": 2033
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.478128426355252e-05,
+      "loss": 0.5068,
+      "step": 2034
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4776630592453492e-05,
+      "loss": 0.4975,
+      "step": 2035
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4771975580714986e-05,
+      "loss": 0.5264,
+      "step": 2036
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4767319229643506e-05,
+      "loss": 0.4815,
+      "step": 2037
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4762661540545932e-05,
+      "loss": 0.528,
+      "step": 2038
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4758002514729524e-05,
+      "loss": 0.4937,
+      "step": 2039
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4753342153501913e-05,
+      "loss": 0.48,
+      "step": 2040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4748680458171099e-05,
+      "loss": 0.4883,
+      "step": 2041
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4744017430045473e-05,
+      "loss": 0.4962,
+      "step": 2042
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4739353070433784e-05,
+      "loss": 0.5208,
+      "step": 2043
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.473468738064516e-05,
+      "loss": 0.4529,
+      "step": 2044
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4730020361989108e-05,
+      "loss": 0.5103,
+      "step": 2045
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.47253520157755e-05,
+      "loss": 0.5091,
+      "step": 2046
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.472068234331458e-05,
+      "loss": 0.5402,
+      "step": 2047
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4716011345916976e-05,
+      "loss": 0.5004,
+      "step": 2048
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4711339024893674e-05,
+      "loss": 0.4977,
+      "step": 2049
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.470666538155604e-05,
+      "loss": 0.4738,
+      "step": 2050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4701990417215807e-05,
+      "loss": 0.4804,
+      "step": 2051
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4697314133185083e-05,
+      "loss": 0.4805,
+      "step": 2052
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4692636530776336e-05,
+      "loss": 0.5,
+      "step": 2053
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.468795761130242e-05,
+      "loss": 0.5291,
+      "step": 2054
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4683277376076548e-05,
+      "loss": 0.5034,
+      "step": 2055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4678595826412303e-05,
+      "loss": 0.4834,
+      "step": 2056
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4673912963623637e-05,
+      "loss": 0.5055,
+      "step": 2057
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4669228789024877e-05,
+      "loss": 0.4725,
+      "step": 2058
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.466454330393071e-05,
+      "loss": 0.4926,
+      "step": 2059
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4659856509656194e-05,
+      "loss": 0.5579,
+      "step": 2060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4655168407516754e-05,
+      "loss": 0.5123,
+      "step": 2061
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.465047899882818e-05,
+      "loss": 0.4895,
+      "step": 2062
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4645788284906639e-05,
+      "loss": 0.5141,
+      "step": 2063
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.464109626706865e-05,
+      "loss": 0.4875,
+      "step": 2064
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4636402946631108e-05,
+      "loss": 0.5015,
+      "step": 2065
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4631708324911269e-05,
+      "loss": 0.4835,
+      "step": 2066
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4627012403226752e-05,
+      "loss": 0.5068,
+      "step": 2067
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.462231518289555e-05,
+      "loss": 0.5033,
+      "step": 2068
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.461761666523601e-05,
+      "loss": 0.5302,
+      "step": 2069
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4612916851566851e-05,
+      "loss": 0.4861,
+      "step": 2070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4608215743207153e-05,
+      "loss": 0.4787,
+      "step": 2071
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4603513341476354e-05,
+      "loss": 0.5375,
+      "step": 2072
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4598809647694266e-05,
+      "loss": 0.509,
+      "step": 2073
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4594104663181052e-05,
+      "loss": 0.4946,
+      "step": 2074
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4589398389257246e-05,
+      "loss": 0.5032,
+      "step": 2075
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.458469082724374e-05,
+      "loss": 0.4754,
+      "step": 2076
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4579981978461792e-05,
+      "loss": 0.4844,
+      "step": 2077
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.457527184423301e-05,
+      "loss": 0.4729,
+      "step": 2078
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4570560425879377e-05,
+      "loss": 0.4808,
+      "step": 2079
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4565847724723225e-05,
+      "loss": 0.4565,
+      "step": 2080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4561133742087253e-05,
+      "loss": 0.4898,
+      "step": 2081
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4556418479294514e-05,
+      "loss": 0.5159,
+      "step": 2082
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.455170193766843e-05,
+      "loss": 0.5023,
+      "step": 2083
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.454698411853277e-05,
+      "loss": 0.5002,
+      "step": 2084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.454226502321167e-05,
+      "loss": 0.4662,
+      "step": 2085
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.453754465302962e-05,
+      "loss": 0.4774,
+      "step": 2086
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.453282300931147e-05,
+      "loss": 0.5091,
+      "step": 2087
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4528100093382422e-05,
+      "loss": 0.4902,
+      "step": 2088
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4523375906568048e-05,
+      "loss": 0.4714,
+      "step": 2089
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4518650450194261e-05,
+      "loss": 0.4562,
+      "step": 2090
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.451392372558734e-05,
+      "loss": 0.5023,
+      "step": 2091
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4509195734073917e-05,
+      "loss": 0.5238,
+      "step": 2092
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4504466476980983e-05,
+      "loss": 0.4943,
+      "step": 2093
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4499735955635882e-05,
+      "loss": 0.5044,
+      "step": 2094
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4495004171366302e-05,
+      "loss": 0.4935,
+      "step": 2095
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4490271125500306e-05,
+      "loss": 0.5045,
+      "step": 2096
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4485536819366299e-05,
+      "loss": 0.4973,
+      "step": 2097
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.448080125429304e-05,
+      "loss": 0.5382,
+      "step": 2098
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4476064431609641e-05,
+      "loss": 0.4492,
+      "step": 2099
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4471326352645573e-05,
+      "loss": 0.4704,
+      "step": 2100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4466587018730654e-05,
+      "loss": 0.4742,
+      "step": 2101
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4461846431195053e-05,
+      "loss": 0.4793,
+      "step": 2102
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4457104591369293e-05,
+      "loss": 0.4901,
+      "step": 2103
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4452361500584252e-05,
+      "loss": 0.4717,
+      "step": 2104
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4447617160171154e-05,
+      "loss": 0.486,
+      "step": 2105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.444287157146158e-05,
+      "loss": 0.4732,
+      "step": 2106
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.443812473578745e-05,
+      "loss": 0.504,
+      "step": 2107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4433376654481046e-05,
+      "loss": 0.4926,
+      "step": 2108
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.442862732887499e-05,
+      "loss": 0.4825,
+      "step": 2109
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4423876760302266e-05,
+      "loss": 0.5016,
+      "step": 2110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4419124950096192e-05,
+      "loss": 0.4784,
+      "step": 2111
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4414371899590445e-05,
+      "loss": 0.5022,
+      "step": 2112
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4409617610119041e-05,
+      "loss": 0.539,
+      "step": 2113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4404862083016355e-05,
+      "loss": 0.4778,
+      "step": 2114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4400105319617102e-05,
+      "loss": 0.4949,
+      "step": 2115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.439534732125634e-05,
+      "loss": 0.5068,
+      "step": 2116
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4390588089269488e-05,
+      "loss": 0.5148,
+      "step": 2117
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4385827624992297e-05,
+      "loss": 0.4851,
+      "step": 2118
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4381065929760867e-05,
+      "loss": 0.4907,
+      "step": 2119
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4376303004911654e-05,
+      "loss": 0.4712,
+      "step": 2120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.437153885178144e-05,
+      "loss": 0.4832,
+      "step": 2121
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4366773471707368e-05,
+      "loss": 0.4634,
+      "step": 2122
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4362006866026921e-05,
+      "loss": 0.4701,
+      "step": 2123
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4357239036077924e-05,
+      "loss": 0.4728,
+      "step": 2124
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4352469983198542e-05,
+      "loss": 0.4946,
+      "step": 2125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4347699708727299e-05,
+      "loss": 0.4711,
+      "step": 2126
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4342928214003038e-05,
+      "loss": 0.4568,
+      "step": 2127
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.433815550036496e-05,
+      "loss": 0.4914,
+      "step": 2128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4333381569152612e-05,
+      "loss": 0.4917,
+      "step": 2129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4328606421705868e-05,
+      "loss": 0.4972,
+      "step": 2130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4323830059364953e-05,
+      "loss": 0.4808,
+      "step": 2131
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4319052483470437e-05,
+      "loss": 0.495,
+      "step": 2132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4314273695363216e-05,
+      "loss": 0.4679,
+      "step": 2133
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4309493696384543e-05,
+      "loss": 0.5073,
+      "step": 2134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4304712487875999e-05,
+      "loss": 0.5183,
+      "step": 2135
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4299930071179505e-05,
+      "loss": 0.5257,
+      "step": 2136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4295146447637325e-05,
+      "loss": 0.4997,
+      "step": 2137
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4290361618592071e-05,
+      "loss": 0.4862,
+      "step": 2138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4285575585386673e-05,
+      "loss": 0.4498,
+      "step": 2139
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4280788349364414e-05,
+      "loss": 0.5013,
+      "step": 2140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4275999911868912e-05,
+      "loss": 0.4696,
+      "step": 2141
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4271210274244114e-05,
+      "loss": 0.5032,
+      "step": 2142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4266419437834312e-05,
+      "loss": 0.4575,
+      "step": 2143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4261627403984136e-05,
+      "loss": 0.5013,
+      "step": 2144
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4256834174038545e-05,
+      "loss": 0.4897,
+      "step": 2145
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4252039749342844e-05,
+      "loss": 0.4883,
+      "step": 2146
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4247244131242656e-05,
+      "loss": 0.5088,
+      "step": 2147
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4242447321083959e-05,
+      "loss": 0.4663,
+      "step": 2148
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4237649320213053e-05,
+      "loss": 0.4839,
+      "step": 2149
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4232850129976573e-05,
+      "loss": 0.5,
+      "step": 2150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.422804975172149e-05,
+      "loss": 0.477,
+      "step": 2151
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4223248186795116e-05,
+      "loss": 0.4794,
+      "step": 2152
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4218445436545081e-05,
+      "loss": 0.4905,
+      "step": 2153
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4213641502319358e-05,
+      "loss": 0.4793,
+      "step": 2154
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.420883638546625e-05,
+      "loss": 0.459,
+      "step": 2155
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4204030087334391e-05,
+      "loss": 0.4852,
+      "step": 2156
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4199222609272747e-05,
+      "loss": 0.4884,
+      "step": 2157
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4194413952630615e-05,
+      "loss": 0.5076,
+      "step": 2158
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4189604118757626e-05,
+      "loss": 0.4751,
+      "step": 2159
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4184793109003734e-05,
+      "loss": 0.5153,
+      "step": 2160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4179980924719231e-05,
+      "loss": 0.5292,
+      "step": 2161
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4175167567254735e-05,
+      "loss": 0.5047,
+      "step": 2162
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4170353037961193e-05,
+      "loss": 0.53,
+      "step": 2163
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.416553733818988e-05,
+      "loss": 0.4982,
+      "step": 2164
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4160720469292402e-05,
+      "loss": 0.4829,
+      "step": 2165
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4155902432620692e-05,
+      "loss": 0.4918,
+      "step": 2166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.415108322952701e-05,
+      "loss": 0.4779,
+      "step": 2167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4146262861363945e-05,
+      "loss": 0.5003,
+      "step": 2168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4141441329484414e-05,
+      "loss": 0.5143,
+      "step": 2169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4136618635241655e-05,
+      "loss": 0.5223,
+      "step": 2170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4131794779989233e-05,
+      "loss": 0.4915,
+      "step": 2171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4126969765081055e-05,
+      "loss": 0.4968,
+      "step": 2172
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4122143591871327e-05,
+      "loss": 0.5001,
+      "step": 2173
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.41173162617146e-05,
+      "loss": 0.4703,
+      "step": 2174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4112487775965741e-05,
+      "loss": 0.5016,
+      "step": 2175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4107658135979944e-05,
+      "loss": 0.4968,
+      "step": 2176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4102827343112725e-05,
+      "loss": 0.507,
+      "step": 2177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4097995398719929e-05,
+      "loss": 0.4762,
+      "step": 2178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4093162304157715e-05,
+      "loss": 0.4871,
+      "step": 2179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4088328060782573e-05,
+      "loss": 0.5035,
+      "step": 2180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4083492669951312e-05,
+      "loss": 0.5147,
+      "step": 2181
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4078656133021065e-05,
+      "loss": 0.4859,
+      "step": 2182
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4073818451349283e-05,
+      "loss": 0.5278,
+      "step": 2183
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.406897962629374e-05,
+      "loss": 0.5022,
+      "step": 2184
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4064139659212534e-05,
+      "loss": 0.4675,
+      "step": 2185
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4059298551464077e-05,
+      "loss": 0.5268,
+      "step": 2186
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4054456304407111e-05,
+      "loss": 0.5141,
+      "step": 2187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4049612919400687e-05,
+      "loss": 0.5162,
+      "step": 2188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4044768397804181e-05,
+      "loss": 0.494,
+      "step": 2189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.403992274097729e-05,
+      "loss": 0.492,
+      "step": 2190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4035075950280021e-05,
+      "loss": 0.4818,
+      "step": 2191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.403022802707271e-05,
+      "loss": 0.4826,
+      "step": 2192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4025378972716004e-05,
+      "loss": 0.5209,
+      "step": 2193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4020528788570871e-05,
+      "loss": 0.4911,
+      "step": 2194
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4015677475998595e-05,
+      "loss": 0.4891,
+      "step": 2195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4010825036360771e-05,
+      "loss": 0.5338,
+      "step": 2196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.400597147101932e-05,
+      "loss": 0.4825,
+      "step": 2197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4001116781336475e-05,
+      "loss": 0.471,
+      "step": 2198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.399626096867478e-05,
+      "loss": 0.4719,
+      "step": 2199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.3991404034397102e-05,
+      "loss": 0.4844,
+      "step": 2200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.398654597986662e-05,
+      "loss": 0.4912,
+      "step": 2201
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.3981686806446822e-05,
+      "loss": 0.509,
+      "step": 2202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3976826515501518e-05,
+      "loss": 0.5028,
+      "step": 2203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3971965108394827e-05,
+      "loss": 0.4754,
+      "step": 2204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3967102586491179e-05,
+      "loss": 0.5052,
+      "step": 2205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3962238951155325e-05,
+      "loss": 0.4803,
+      "step": 2206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3957374203752325e-05,
+      "loss": 0.5064,
+      "step": 2207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3952508345647546e-05,
+      "loss": 0.5004,
+      "step": 2208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3947641378206671e-05,
+      "loss": 0.5101,
+      "step": 2209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3942773302795697e-05,
+      "loss": 0.4889,
+      "step": 2210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3937904120780925e-05,
+      "loss": 0.509,
+      "step": 2211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3933033833528971e-05,
+      "loss": 0.5285,
+      "step": 2212
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3928162442406765e-05,
+      "loss": 0.5387,
+      "step": 2213
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3923289948781538e-05,
+      "loss": 0.5058,
+      "step": 2214
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3918416354020836e-05,
+      "loss": 0.4837,
+      "step": 2215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3913541659492516e-05,
+      "loss": 0.4815,
+      "step": 2216
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.3908665866564736e-05,
+      "loss": 0.4681,
+      "step": 2217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3903788976605972e-05,
+      "loss": 0.506,
+      "step": 2218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3898910990985e-05,
+      "loss": 0.5058,
+      "step": 2219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3894031911070904e-05,
+      "loss": 0.4847,
+      "step": 2220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3889151738233087e-05,
+      "loss": 0.4951,
+      "step": 2221
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3884270473841235e-05,
+      "loss": 0.4927,
+      "step": 2222
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3879388119265366e-05,
+      "loss": 0.4699,
+      "step": 2223
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3874504675875791e-05,
+      "loss": 0.5349,
+      "step": 2224
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3869620145043123e-05,
+      "loss": 0.4957,
+      "step": 2225
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3864734528138288e-05,
+      "loss": 0.4657,
+      "step": 2226
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3859847826532518e-05,
+      "loss": 0.511,
+      "step": 2227
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.385496004159734e-05,
+      "loss": 0.5271,
+      "step": 2228
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3850071174704596e-05,
+      "loss": 0.4646,
+      "step": 2229
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3845181227226423e-05,
+      "loss": 0.4827,
+      "step": 2230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3840290200535265e-05,
+      "loss": 0.4871,
+      "step": 2231
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.3835398096003869e-05,
+      "loss": 0.4928,
+      "step": 2232
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3830504915005285e-05,
+      "loss": 0.4904,
+      "step": 2233
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3825610658912863e-05,
+      "loss": 0.4594,
+      "step": 2234
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3820715329100256e-05,
+      "loss": 0.4881,
+      "step": 2235
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3815818926941422e-05,
+      "loss": 0.4578,
+      "step": 2236
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3810921453810611e-05,
+      "loss": 0.4822,
+      "step": 2237
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.380602291108238e-05,
+      "loss": 0.5512,
+      "step": 2238
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3801123300131587e-05,
+      "loss": 0.5119,
+      "step": 2239
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3796222622333389e-05,
+      "loss": 0.4778,
+      "step": 2240
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.379132087906324e-05,
+      "loss": 0.4658,
+      "step": 2241
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.378641807169689e-05,
+      "loss": 0.4784,
+      "step": 2242
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3781514201610398e-05,
+      "loss": 0.4842,
+      "step": 2243
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3776609270180118e-05,
+      "loss": 0.4689,
+      "step": 2244
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3771703278782689e-05,
+      "loss": 0.4997,
+      "step": 2245
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.3766796228795064e-05,
+      "loss": 0.5217,
+      "step": 2246
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3761888121594487e-05,
+      "loss": 0.4725,
+      "step": 2247
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3756978958558496e-05,
+      "loss": 0.4786,
+      "step": 2248
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.375206874106493e-05,
+      "loss": 0.4814,
+      "step": 2249
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3747157470491923e-05,
+      "loss": 0.4969,
+      "step": 2250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3742245148217901e-05,
+      "loss": 0.4791,
+      "step": 2251
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3737331775621588e-05,
+      "loss": 0.4764,
+      "step": 2252
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3732417354082002e-05,
+      "loss": 0.4959,
+      "step": 2253
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3727501884978456e-05,
+      "loss": 0.4977,
+      "step": 2254
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.372258536969056e-05,
+      "loss": 0.5343,
+      "step": 2255
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.371766780959821e-05,
+      "loss": 0.4699,
+      "step": 2256
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.37127492060816e-05,
+      "loss": 0.503,
+      "step": 2257
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3707829560521219e-05,
+      "loss": 0.4918,
+      "step": 2258
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3702908874297846e-05,
+      "loss": 0.4999,
+      "step": 2259
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3697987148792546e-05,
+      "loss": 0.5087,
+      "step": 2260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.369306438538669e-05,
+      "loss": 0.4966,
+      "step": 2261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.368814058546193e-05,
+      "loss": 0.4839,
+      "step": 2262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3683215750400207e-05,
+      "loss": 0.4963,
+      "step": 2263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3678289881583759e-05,
+      "loss": 0.4981,
+      "step": 2264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3673362980395115e-05,
+      "loss": 0.4385,
+      "step": 2265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3668435048217083e-05,
+      "loss": 0.5018,
+      "step": 2266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3663506086432774e-05,
+      "loss": 0.4818,
+      "step": 2267
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3658576096425578e-05,
+      "loss": 0.4613,
+      "step": 2268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3653645079579178e-05,
+      "loss": 0.4918,
+      "step": 2269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.364871303727755e-05,
+      "loss": 0.5107,
+      "step": 2270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3643779970904942e-05,
+      "loss": 0.5079,
+      "step": 2271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3638845881845909e-05,
+      "loss": 0.5121,
+      "step": 2272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3633910771485276e-05,
+      "loss": 0.4599,
+      "step": 2273
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3628974641208168e-05,
+      "loss": 0.488,
+      "step": 2274
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3624037492399991e-05,
+      "loss": 0.5232,
+      "step": 2275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3619099326446434e-05,
+      "loss": 0.5337,
+      "step": 2276
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3614160144733475e-05,
+      "loss": 0.4987,
+      "step": 2277
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3609219948647377e-05,
+      "loss": 0.4844,
+      "step": 2278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3604278739574683e-05,
+      "loss": 0.5125,
+      "step": 2279
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3599336518902228e-05,
+      "loss": 0.5085,
+      "step": 2280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3594393288017122e-05,
+      "loss": 0.4843,
+      "step": 2281
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3589449048306773e-05,
+      "loss": 0.4926,
+      "step": 2282
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3584503801158854e-05,
+      "loss": 0.491,
+      "step": 2283
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3579557547961335e-05,
+      "loss": 0.4813,
+      "step": 2284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3574610290102462e-05,
+      "loss": 0.498,
+      "step": 2285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3569662028970759e-05,
+      "loss": 0.4855,
+      "step": 2286
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3564712765955038e-05,
+      "loss": 0.4984,
+      "step": 2287
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3559762502444396e-05,
+      "loss": 0.5005,
+      "step": 2288
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3554811239828198e-05,
+      "loss": 0.5207,
+      "step": 2289
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3549858979496104e-05,
+      "loss": 0.4841,
+      "step": 2290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3544905722838042e-05,
+      "loss": 0.5085,
+      "step": 2291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3539951471244224e-05,
+      "loss": 0.4746,
+      "step": 2292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3534996226105144e-05,
+      "loss": 0.4904,
+      "step": 2293
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3530039988811573e-05,
+      "loss": 0.5245,
+      "step": 2294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3525082760754557e-05,
+      "loss": 0.4997,
+      "step": 2295
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.352012454332543e-05,
+      "loss": 0.4993,
+      "step": 2296
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3515165337915788e-05,
+      "loss": 0.5101,
+      "step": 2297
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3510205145917517e-05,
+      "loss": 0.4784,
+      "step": 2298
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.350524396872278e-05,
+      "loss": 0.4903,
+      "step": 2299
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3500281807724003e-05,
+      "loss": 0.4962,
+      "step": 2300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3495318664313904e-05,
+      "loss": 0.475,
+      "step": 2301
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3490354539885473e-05,
+      "loss": 0.4922,
+      "step": 2302
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3485389435831966e-05,
+      "loss": 0.5075,
+      "step": 2303
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3480423353546926e-05,
+      "loss": 0.5166,
+      "step": 2304
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3475456294424163e-05,
+      "loss": 0.4945,
+      "step": 2305
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3470488259857763e-05,
+      "loss": 0.523,
+      "step": 2306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3465519251242085e-05,
+      "loss": 0.5399,
+      "step": 2307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3460549269971765e-05,
+      "loss": 0.4964,
+      "step": 2308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.345557831744171e-05,
+      "loss": 0.5056,
+      "step": 2309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3450606395047094e-05,
+      "loss": 0.4744,
+      "step": 2310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3445633504183376e-05,
+      "loss": 0.5038,
+      "step": 2311
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3440659646246275e-05,
+      "loss": 0.5182,
+      "step": 2312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3435684822631783e-05,
+      "loss": 0.4954,
+      "step": 2313
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3430709034736169e-05,
+      "loss": 0.5193,
+      "step": 2314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3425732283955968e-05,
+      "loss": 0.5161,
+      "step": 2315
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3420754571687989e-05,
+      "loss": 0.4841,
+      "step": 2316
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3415775899329306e-05,
+      "loss": 0.5132,
+      "step": 2317
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3410796268277264e-05,
+      "loss": 0.4653,
+      "step": 2318
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.340581567992948e-05,
+      "loss": 0.5226,
+      "step": 2319
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3400834135683836e-05,
+      "loss": 0.5066,
+      "step": 2320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3395851636938484e-05,
+      "loss": 0.5038,
+      "step": 2321
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3390868185091844e-05,
+      "loss": 0.4913,
+      "step": 2322
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3385883781542601e-05,
+      "loss": 0.5052,
+      "step": 2323
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3380898427689717e-05,
+      "loss": 0.4717,
+      "step": 2324
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3375912124932406e-05,
+      "loss": 0.4844,
+      "step": 2325
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3370924874670155e-05,
+      "loss": 0.5088,
+      "step": 2326
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3365936678302722e-05,
+      "loss": 0.5426,
+      "step": 2327
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3360947537230123e-05,
+      "loss": 0.506,
+      "step": 2328
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.335595745285264e-05,
+      "loss": 0.4665,
+      "step": 2329
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3350966426570825e-05,
+      "loss": 0.4601,
+      "step": 2330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3345974459785494e-05,
+      "loss": 0.4649,
+      "step": 2331
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3340981553897719e-05,
+      "loss": 0.4749,
+      "step": 2332
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.333598771030884e-05,
+      "loss": 0.4608,
+      "step": 2333
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3330992930420466e-05,
+      "loss": 0.5071,
+      "step": 2334
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3325997215634457e-05,
+      "loss": 0.4862,
+      "step": 2335
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3321000567352944e-05,
+      "loss": 0.4901,
+      "step": 2336
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3316002986978326e-05,
+      "loss": 0.5135,
+      "step": 2337
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3311004475913247e-05,
+      "loss": 0.5051,
+      "step": 2338
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3306005035560623e-05,
+      "loss": 0.4989,
+      "step": 2339
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.330100466732363e-05,
+      "loss": 0.4646,
+      "step": 2340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.32960033726057e-05,
+      "loss": 0.4715,
+      "step": 2341
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.329100115281053e-05,
+      "loss": 0.4844,
+      "step": 2342
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3285998009342076e-05,
+      "loss": 0.5051,
+      "step": 2343
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.328099394360455e-05,
+      "loss": 0.5271,
+      "step": 2344
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3275988957002429e-05,
+      "loss": 0.4777,
+      "step": 2345
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3270983050940435e-05,
+      "loss": 0.4982,
+      "step": 2346
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3265976226823569e-05,
+      "loss": 0.4717,
+      "step": 2347
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3260968486057068e-05,
+      "loss": 0.4823,
+      "step": 2348
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3255959830046446e-05,
+      "loss": 0.496,
+      "step": 2349
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3250950260197455e-05,
+      "loss": 0.4878,
+      "step": 2350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.324593977791612e-05,
+      "loss": 0.4919,
+      "step": 2351
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3240928384608712e-05,
+      "loss": 0.4997,
+      "step": 2352
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3235916081681757e-05,
+      "loss": 0.4921,
+      "step": 2353
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3230902870542046e-05,
+      "loss": 0.5137,
+      "step": 2354
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3225888752596614e-05,
+      "loss": 0.4929,
+      "step": 2355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3220873729252752e-05,
+      "loss": 0.4413,
+      "step": 2356
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.321585780191802e-05,
+      "loss": 0.4459,
+      "step": 2357
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3210840972000209e-05,
+      "loss": 0.4613,
+      "step": 2358
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3205823240907381e-05,
+      "loss": 0.5138,
+      "step": 2359
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3200804610047842e-05,
+      "loss": 0.5126,
+      "step": 2360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.319578508083015e-05,
+      "loss": 0.5372,
+      "step": 2361
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3190764654663122e-05,
+      "loss": 0.496,
+      "step": 2362
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3185743332955818e-05,
+      "loss": 0.4629,
+      "step": 2363
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3180721117117562e-05,
+      "loss": 0.5045,
+      "step": 2364
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3175698008557914e-05,
+      "loss": 0.4811,
+      "step": 2365
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3170674008686692e-05,
+      "loss": 0.4964,
+      "step": 2366
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3165649118913968e-05,
+      "loss": 0.5173,
+      "step": 2367
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3160623340650058e-05,
+      "loss": 0.4976,
+      "step": 2368
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3155596675305522e-05,
+      "loss": 0.4787,
+      "step": 2369
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3150569124291186e-05,
+      "loss": 0.4784,
+      "step": 2370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.314554068901811e-05,
+      "loss": 0.5282,
+      "step": 2371
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3140511370897607e-05,
+      "loss": 0.5015,
+      "step": 2372
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3135481171341233e-05,
+      "loss": 0.4699,
+      "step": 2373
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3130450091760804e-05,
+      "loss": 0.4843,
+      "step": 2374
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3125418133568369e-05,
+      "loss": 0.4747,
+      "step": 2375
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.312038529817623e-05,
+      "loss": 0.5123,
+      "step": 2376
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3115351586996937e-05,
+      "loss": 0.503,
+      "step": 2377
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3110317001443278e-05,
+      "loss": 0.4729,
+      "step": 2378
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3105281542928296e-05,
+      "loss": 0.497,
+      "step": 2379
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3100245212865279e-05,
+      "loss": 0.5098,
+      "step": 2380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3095208012667747e-05,
+      "loss": 0.5135,
+      "step": 2381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.4688,
+      "step": 2382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3085131007524483e-05,
+      "loss": 0.4836,
+      "step": 2383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3080091205407027e-05,
+      "loss": 0.4571,
+      "step": 2384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3075050538811611e-05,
+      "loss": 0.4976,
+      "step": 2385
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3070009009152984e-05,
+      "loss": 0.5032,
+      "step": 2386
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3064966617846126e-05,
+      "loss": 0.4935,
+      "step": 2387
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3059923366306272e-05,
+      "loss": 0.5023,
+      "step": 2388
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3054879255948896e-05,
+      "loss": 0.4684,
+      "step": 2389
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3049834288189702e-05,
+      "loss": 0.5184,
+      "step": 2390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3044788464444648e-05,
+      "loss": 0.4983,
+      "step": 2391
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3039741786129927e-05,
+      "loss": 0.514,
+      "step": 2392
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3034694254661971e-05,
+      "loss": 0.4756,
+      "step": 2393
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.302964587145745e-05,
+      "loss": 0.4788,
+      "step": 2394
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3024596637933277e-05,
+      "loss": 0.4821,
+      "step": 2395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3019546555506603e-05,
+      "loss": 0.5358,
+      "step": 2396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3014495625594815e-05,
+      "loss": 0.5028,
+      "step": 2397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3009443849615538e-05,
+      "loss": 0.48,
+      "step": 2398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3004391228986637e-05,
+      "loss": 0.4808,
+      "step": 2399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2999337765126217e-05,
+      "loss": 0.4664,
+      "step": 2400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2994283459452606e-05,
+      "loss": 0.5013,
+      "step": 2401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2989228313384383e-05,
+      "loss": 0.497,
+      "step": 2402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2984172328340357e-05,
+      "loss": 0.5056,
+      "step": 2403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2979115505739573e-05,
+      "loss": 0.4779,
+      "step": 2404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2974057847001306e-05,
+      "loss": 0.4619,
+      "step": 2405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2968999353545082e-05,
+      "loss": 0.4889,
+      "step": 2406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.296394002679064e-05,
+      "loss": 0.5278,
+      "step": 2407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2958879868157966e-05,
+      "loss": 0.501,
+      "step": 2408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2953818879067276e-05,
+      "loss": 0.481,
+      "step": 2409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.2948757060939019e-05,
+      "loss": 0.5152,
+      "step": 2410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2943694415193875e-05,
+      "loss": 0.5048,
+      "step": 2411
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2938630943252765e-05,
+      "loss": 0.5087,
+      "step": 2412
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.293356664653683e-05,
+      "loss": 0.5102,
+      "step": 2413
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2928501526467448e-05,
+      "loss": 0.4895,
+      "step": 2414
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2923435584466228e-05,
+      "loss": 0.5067,
+      "step": 2415
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.291836882195501e-05,
+      "loss": 0.5439,
+      "step": 2416
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2913301240355861e-05,
+      "loss": 0.473,
+      "step": 2417
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2908232841091088e-05,
+      "loss": 0.4874,
+      "step": 2418
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2903163625583213e-05,
+      "loss": 0.5007,
+      "step": 2419
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2898093595254998e-05,
+      "loss": 0.5122,
+      "step": 2420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2893022751529425e-05,
+      "loss": 0.4922,
+      "step": 2421
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2887951095829713e-05,
+      "loss": 0.4741,
+      "step": 2422
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.288287862957931e-05,
+      "loss": 0.4794,
+      "step": 2423
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2877805354201876e-05,
+      "loss": 0.5161,
+      "step": 2424
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.2872731271121315e-05,
+      "loss": 0.4921,
+      "step": 2425
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.286765638176175e-05,
+      "loss": 0.4682,
+      "step": 2426
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2862580687547534e-05,
+      "loss": 0.5074,
+      "step": 2427
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2857504189903241e-05,
+      "loss": 0.5238,
+      "step": 2428
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2852426890253676e-05,
+      "loss": 0.4943,
+      "step": 2429
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2847348790023858e-05,
+      "loss": 0.4989,
+      "step": 2430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.284226989063905e-05,
+      "loss": 0.4927,
+      "step": 2431
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.283719019352472e-05,
+      "loss": 0.5004,
+      "step": 2432
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2832109700106572e-05,
+      "loss": 0.4837,
+      "step": 2433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2827028411810527e-05,
+      "loss": 0.4997,
+      "step": 2434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2821946330062738e-05,
+      "loss": 0.4895,
+      "step": 2435
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2816863456289565e-05,
+      "loss": 0.5072,
+      "step": 2436
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2811779791917605e-05,
+      "loss": 0.4691,
+      "step": 2437
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2806695338373669e-05,
+      "loss": 0.4917,
+      "step": 2438
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.2801610097084796e-05,
+      "loss": 0.4832,
+      "step": 2439
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2796524069478243e-05,
+      "loss": 0.5057,
+      "step": 2440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2791437256981479e-05,
+      "loss": 0.4779,
+      "step": 2441
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2786349661022205e-05,
+      "loss": 0.5197,
+      "step": 2442
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2781261283028344e-05,
+      "loss": 0.4913,
+      "step": 2443
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2776172124428023e-05,
+      "loss": 0.4637,
+      "step": 2444
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.27710821866496e-05,
+      "loss": 0.4851,
+      "step": 2445
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2765991471121655e-05,
+      "loss": 0.4761,
+      "step": 2446
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2760899979272977e-05,
+      "loss": 0.4816,
+      "step": 2447
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.275580771253257e-05,
+      "loss": 0.4659,
+      "step": 2448
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2750714672329676e-05,
+      "loss": 0.463,
+      "step": 2449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2745620860093726e-05,
+      "loss": 0.4985,
+      "step": 2450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.274052627725439e-05,
+      "loss": 0.4847,
+      "step": 2451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.2735430925241543e-05,
+      "loss": 0.4654,
+      "step": 2452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.273033480548528e-05,
+      "loss": 0.4958,
+      "step": 2453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.272523791941591e-05,
+      "loss": 0.5283,
+      "step": 2454
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2720140268463958e-05,
+      "loss": 0.4846,
+      "step": 2455
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2715041854060161e-05,
+      "loss": 0.502,
+      "step": 2456
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2709942677635474e-05,
+      "loss": 0.4987,
+      "step": 2457
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2704842740621061e-05,
+      "loss": 0.48,
+      "step": 2458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2699742044448311e-05,
+      "loss": 0.45,
+      "step": 2459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2694640590548808e-05,
+      "loss": 0.5029,
+      "step": 2460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2689538380354367e-05,
+      "loss": 0.4948,
+      "step": 2461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2684435415297e-05,
+      "loss": 0.5004,
+      "step": 2462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.267933169680894e-05,
+      "loss": 0.518,
+      "step": 2463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2674227226322633e-05,
+      "loss": 0.4908,
+      "step": 2464
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2669122005270724e-05,
+      "loss": 0.4942,
+      "step": 2465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2664016035086081e-05,
+      "loss": 0.4937,
+      "step": 2466
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2658909317201782e-05,
+      "loss": 0.512,
+      "step": 2467
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2653801853051105e-05,
+      "loss": 0.4871,
+      "step": 2468
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.2648693644067545e-05,
+      "loss": 0.4573,
+      "step": 2469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2643584691684802e-05,
+      "loss": 0.5149,
+      "step": 2470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2638474997336794e-05,
+      "loss": 0.4781,
+      "step": 2471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2633364562457632e-05,
+      "loss": 0.4893,
+      "step": 2472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.262825338848165e-05,
+      "loss": 0.4776,
+      "step": 2473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.262314147684338e-05,
+      "loss": 0.5026,
+      "step": 2474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2618028828977563e-05,
+      "loss": 0.5018,
+      "step": 2475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2612915446319148e-05,
+      "loss": 0.5081,
+      "step": 2476
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2607801330303287e-05,
+      "loss": 0.4972,
+      "step": 2477
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2602686482365343e-05,
+      "loss": 0.5139,
+      "step": 2478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2597570903940879e-05,
+      "loss": 0.5296,
+      "step": 2479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.259245459646567e-05,
+      "loss": 0.4634,
+      "step": 2480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.258733756137569e-05,
+      "loss": 0.5005,
+      "step": 2481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2582219800107113e-05,
+      "loss": 0.5058,
+      "step": 2482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.257710131409633e-05,
+      "loss": 0.4789,
+      "step": 2483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.2571982104779927e-05,
+      "loss": 0.497,
+      "step": 2484
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2566862173594689e-05,
+      "loss": 0.5065,
+      "step": 2485
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.256174152197761e-05,
+      "loss": 0.526,
+      "step": 2486
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2556620151365886e-05,
+      "loss": 0.5361,
+      "step": 2487
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2551498063196914e-05,
+      "loss": 0.4911,
+      "step": 2488
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.254637525890829e-05,
+      "loss": 0.4768,
+      "step": 2489
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2541251739937814e-05,
+      "loss": 0.4887,
+      "step": 2490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2536127507723486e-05,
+      "loss": 0.4734,
+      "step": 2491
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2531002563703502e-05,
+      "loss": 0.453,
+      "step": 2492
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2525876909316263e-05,
+      "loss": 0.4697,
+      "step": 2493
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.252075054600037e-05,
+      "loss": 0.4998,
+      "step": 2494
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2515623475194623e-05,
+      "loss": 0.4756,
+      "step": 2495
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.251049569833801e-05,
+      "loss": 0.4939,
+      "step": 2496
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.250536721686973e-05,
+      "loss": 0.5076,
+      "step": 2497
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2500238032229177e-05,
+      "loss": 0.5119,
+      "step": 2498
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.2495108145855938e-05,
+      "loss": 0.512,
+      "step": 2499
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2489977559189796e-05,
+      "loss": 0.5,
+      "step": 2500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2484846273670745e-05,
+      "loss": 0.514,
+      "step": 2501
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2479714290738953e-05,
+      "loss": 0.4768,
+      "step": 2502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.24745816118348e-05,
+      "loss": 0.4813,
+      "step": 2503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2469448238398859e-05,
+      "loss": 0.4583,
+      "step": 2504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2464314171871888e-05,
+      "loss": 0.5163,
+      "step": 2505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2459179413694849e-05,
+      "loss": 0.4898,
+      "step": 2506
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.24540439653089e-05,
+      "loss": 0.4777,
+      "step": 2507
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2448907828155383e-05,
+      "loss": 0.5537,
+      "step": 2508
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2443771003675842e-05,
+      "loss": 0.4661,
+      "step": 2509
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2438633493312016e-05,
+      "loss": 0.5147,
+      "step": 2510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.243349529850582e-05,
+      "loss": 0.4929,
+      "step": 2511
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2428356420699378e-05,
+      "loss": 0.4882,
+      "step": 2512
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2423216861335004e-05,
+      "loss": 0.5021,
+      "step": 2513
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2418076621855191e-05,
+      "loss": 0.4877,
+      "step": 2514
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.241293570370264e-05,
+      "loss": 0.4847,
+      "step": 2515
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2407794108320229e-05,
+      "loss": 0.478,
+      "step": 2516
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2402651837151029e-05,
+      "loss": 0.5041,
+      "step": 2517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2397508891638308e-05,
+      "loss": 0.4899,
+      "step": 2518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2392365273225513e-05,
+      "loss": 0.4775,
+      "step": 2519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2387220983356283e-05,
+      "loss": 0.4811,
+      "step": 2520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2382076023474455e-05,
+      "loss": 0.5075,
+      "step": 2521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2376930395024039e-05,
+      "loss": 0.4862,
+      "step": 2522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2371784099449242e-05,
+      "loss": 0.5041,
+      "step": 2523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.236663713819446e-05,
+      "loss": 0.5246,
+      "step": 2524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2361489512704264e-05,
+      "loss": 0.508,
+      "step": 2525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2356341224423422e-05,
+      "loss": 0.4847,
+      "step": 2526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.235119227479689e-05,
+      "loss": 0.4775,
+      "step": 2527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2346042665269799e-05,
+      "loss": 0.5094,
+      "step": 2528
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2340892397287475e-05,
+      "loss": 0.5002,
+      "step": 2529
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2335741472295426e-05,
+      "loss": 0.4888,
+      "step": 2530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2330589891739338e-05,
+      "loss": 0.4987,
+      "step": 2531
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2325437657065089e-05,
+      "loss": 0.4593,
+      "step": 2532
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2320284769718739e-05,
+      "loss": 0.495,
+      "step": 2533
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2315131231146526e-05,
+      "loss": 0.4926,
+      "step": 2534
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.230997704279488e-05,
+      "loss": 0.4855,
+      "step": 2535
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2304822206110409e-05,
+      "loss": 0.4989,
+      "step": 2536
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2299666722539897e-05,
+      "loss": 0.5409,
+      "step": 2537
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2294510593530318e-05,
+      "loss": 0.4749,
+      "step": 2538
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2289353820528825e-05,
+      "loss": 0.4767,
+      "step": 2539
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2284196404982746e-05,
+      "loss": 0.5033,
+      "step": 2540
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2279038348339595e-05,
+      "loss": 0.4805,
+      "step": 2541
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.227387965204707e-05,
+      "loss": 0.4998,
+      "step": 2542
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.226872031755304e-05,
+      "loss": 0.4781,
+      "step": 2543
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2263560346305556e-05,
+      "loss": 0.4978,
+      "step": 2544
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2258399739752848e-05,
+      "loss": 0.484,
+      "step": 2545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2253238499343328e-05,
+      "loss": 0.4384,
+      "step": 2546
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2248076626525578e-05,
+      "loss": 0.462,
+      "step": 2547
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2242914122748363e-05,
+      "loss": 0.496,
+      "step": 2548
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.223775098946063e-05,
+      "loss": 0.4909,
+      "step": 2549
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.223258722811149e-05,
+      "loss": 0.4811,
+      "step": 2550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2227422840150238e-05,
+      "loss": 0.5114,
+      "step": 2551
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2222257827026345e-05,
+      "loss": 0.4714,
+      "step": 2552
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.221709219018946e-05,
+      "loss": 0.4699,
+      "step": 2553
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2211925931089394e-05,
+      "loss": 0.4859,
+      "step": 2554
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2206759051176151e-05,
+      "loss": 0.4844,
+      "step": 2555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2201591551899898e-05,
+      "loss": 0.4592,
+      "step": 2556
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2196423434710978e-05,
+      "loss": 0.5196,
+      "step": 2557
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2191254701059904e-05,
+      "loss": 0.4786,
+      "step": 2558
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2186085352397374e-05,
+      "loss": 0.4673,
+      "step": 2559
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.218091539017424e-05,
+      "loss": 0.4997,
+      "step": 2560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2175744815841543e-05,
+      "loss": 0.5088,
+      "step": 2561
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2170573630850486e-05,
+      "loss": 0.476,
+      "step": 2562
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.216540183665245e-05,
+      "loss": 0.4563,
+      "step": 2563
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2160229434698984e-05,
+      "loss": 0.4688,
+      "step": 2564
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2155056426441803e-05,
+      "loss": 0.5236,
+      "step": 2565
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2149882813332796e-05,
+      "loss": 0.486,
+      "step": 2566
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2144708596824027e-05,
+      "loss": 0.4992,
+      "step": 2567
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.213953377836772e-05,
+      "loss": 0.5141,
+      "step": 2568
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2134358359416277e-05,
+      "loss": 0.4961,
+      "step": 2569
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.212918234142226e-05,
+      "loss": 0.4442,
+      "step": 2570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2124005725838404e-05,
+      "loss": 0.5016,
+      "step": 2571
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2118828514117611e-05,
+      "loss": 0.5002,
+      "step": 2572
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2113650707712952e-05,
+      "loss": 0.486,
+      "step": 2573
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2108472308077661e-05,
+      "loss": 0.4673,
+      "step": 2574
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2103293316665137e-05,
+      "loss": 0.4508,
+      "step": 2575
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2098113734928957e-05,
+      "loss": 0.4631,
+      "step": 2576
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.209293356432285e-05,
+      "loss": 0.4737,
+      "step": 2577
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2087752806300715e-05,
+      "loss": 0.4912,
+      "step": 2578
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.208257146231662e-05,
+      "loss": 0.4755,
+      "step": 2579
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2077389533824789e-05,
+      "loss": 0.5224,
+      "step": 2580
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2072207022279618e-05,
+      "loss": 0.4868,
+      "step": 2581
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2067023929135664e-05,
+      "loss": 0.5011,
+      "step": 2582
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2061840255847645e-05,
+      "loss": 0.5079,
+      "step": 2583
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2056656003870448e-05,
+      "loss": 0.4912,
+      "step": 2584
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2051471174659116e-05,
+      "loss": 0.5013,
+      "step": 2585
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.204628576966885e-05,
+      "loss": 0.4873,
+      "step": 2586
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2041099790355025e-05,
+      "loss": 0.4853,
+      "step": 2587
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.203591323817317e-05,
+      "loss": 0.481,
+      "step": 2588
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2030726114578974e-05,
+      "loss": 0.5212,
+      "step": 2589
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2025538421028293e-05,
+      "loss": 0.4934,
+      "step": 2590
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2020350158977131e-05,
+      "loss": 0.5001,
+      "step": 2591
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2015161329881663e-05,
+      "loss": 0.4796,
+      "step": 2592
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.200997193519822e-05,
+      "loss": 0.4958,
+      "step": 2593
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2004781976383286e-05,
+      "loss": 0.5291,
+      "step": 2594
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1999591454893511e-05,
+      "loss": 0.5025,
+      "step": 2595
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1994400372185698e-05,
+      "loss": 0.4827,
+      "step": 2596
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1989208729716808e-05,
+      "loss": 0.4829,
+      "step": 2597
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1984016528943962e-05,
+      "loss": 0.485,
+      "step": 2598
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1978823771324439e-05,
+      "loss": 0.5281,
+      "step": 2599
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1973630458315667e-05,
+      "loss": 0.4634,
+      "step": 2600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1968436591375236e-05,
+      "loss": 0.4909,
+      "step": 2601
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.196324217196089e-05,
+      "loss": 0.5222,
+      "step": 2602
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.1958047201530527e-05,
+      "loss": 0.4826,
+      "step": 2603
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1952851681542201e-05,
+      "loss": 0.4825,
+      "step": 2604
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.194765561345412e-05,
+      "loss": 0.475,
+      "step": 2605
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1942458998724642e-05,
+      "loss": 0.5081,
+      "step": 2606
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1937261838812286e-05,
+      "loss": 0.4609,
+      "step": 2607
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1932064135175723e-05,
+      "loss": 0.5037,
+      "step": 2608
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1926865889273762e-05,
+      "loss": 0.4807,
+      "step": 2609
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1921667102565384e-05,
+      "loss": 0.492,
+      "step": 2610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1916467776509717e-05,
+      "loss": 0.5251,
+      "step": 2611
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1911267912566027e-05,
+      "loss": 0.4634,
+      "step": 2612
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1906067512193748e-05,
+      "loss": 0.4855,
+      "step": 2613
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.190086657685246e-05,
+      "loss": 0.4897,
+      "step": 2614
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1895665108001879e-05,
+      "loss": 0.4853,
+      "step": 2615
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1890463107101891e-05,
+      "loss": 0.4809,
+      "step": 2616
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.188526057561252e-05,
+      "loss": 0.5015,
+      "step": 2617
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.1880057514993944e-05,
+      "loss": 0.5009,
+      "step": 2618
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1874853926706486e-05,
+      "loss": 0.4776,
+      "step": 2619
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1869649812210618e-05,
+      "loss": 0.5275,
+      "step": 2620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1864445172966956e-05,
+      "loss": 0.4887,
+      "step": 2621
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1859240010436272e-05,
+      "loss": 0.5075,
+      "step": 2622
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.185403432607948e-05,
+      "loss": 0.4988,
+      "step": 2623
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1848828121357637e-05,
+      "loss": 0.4991,
+      "step": 2624
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1843621397731954e-05,
+      "loss": 0.5394,
+      "step": 2625
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1838414156663778e-05,
+      "loss": 0.5116,
+      "step": 2626
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.183320639961461e-05,
+      "loss": 0.4602,
+      "step": 2627
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1827998128046091e-05,
+      "loss": 0.4982,
+      "step": 2628
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1822789343420008e-05,
+      "loss": 0.4962,
+      "step": 2629
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1817580047198287e-05,
+      "loss": 0.4969,
+      "step": 2630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1812370240843012e-05,
+      "loss": 0.4923,
+      "step": 2631
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1807159925816389e-05,
+      "loss": 0.5061,
+      "step": 2632
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1801949103580788e-05,
+      "loss": 0.5,
+      "step": 2633
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1796737775598708e-05,
+      "loss": 0.5,
+      "step": 2634
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.179152594333279e-05,
+      "loss": 0.5057,
+      "step": 2635
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1786313608245823e-05,
+      "loss": 0.4722,
+      "step": 2636
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1781100771800733e-05,
+      "loss": 0.4782,
+      "step": 2637
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1775887435460588e-05,
+      "loss": 0.502,
+      "step": 2638
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.17706736006886e-05,
+      "loss": 0.4728,
+      "step": 2639
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1765459268948111e-05,
+      "loss": 0.4592,
+      "step": 2640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1760244441702612e-05,
+      "loss": 0.4893,
+      "step": 2641
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1755029120415728e-05,
+      "loss": 0.4818,
+      "step": 2642
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1749813306551221e-05,
+      "loss": 0.4849,
+      "step": 2643
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1744597001573002e-05,
+      "loss": 0.5335,
+      "step": 2644
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1739380206945108e-05,
+      "loss": 0.4826,
+      "step": 2645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1734162924131719e-05,
+      "loss": 0.4944,
+      "step": 2646
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1728945154597149e-05,
+      "loss": 0.4772,
+      "step": 2647
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1723726899805851e-05,
+      "loss": 0.4972,
+      "step": 2648
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1718508161222415e-05,
+      "loss": 0.4766,
+      "step": 2649
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1713288940311562e-05,
+      "loss": 0.4663,
+      "step": 2650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1708069238538153e-05,
+      "loss": 0.4894,
+      "step": 2651
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1702849057367185e-05,
+      "loss": 0.468,
+      "step": 2652
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1697628398263785e-05,
+      "loss": 0.5206,
+      "step": 2653
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1692407262693219e-05,
+      "loss": 0.4589,
+      "step": 2654
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.168718565212088e-05,
+      "loss": 0.496,
+      "step": 2655
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.16819635680123e-05,
+      "loss": 0.4686,
+      "step": 2656
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1676741011833142e-05,
+      "loss": 0.4786,
+      "step": 2657
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1671517985049204e-05,
+      "loss": 0.4619,
+      "step": 2658
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1666294489126411e-05,
+      "loss": 0.5025,
+      "step": 2659
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1661070525530827e-05,
+      "loss": 0.4791,
+      "step": 2660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1655846095728638e-05,
+      "loss": 0.4788,
+      "step": 2661
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.1650621201186165e-05,
+      "loss": 0.4947,
+      "step": 2662
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1645395843369867e-05,
+      "loss": 0.4753,
+      "step": 2663
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1640170023746315e-05,
+      "loss": 0.4809,
+      "step": 2664
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1634943743782235e-05,
+      "loss": 0.4601,
+      "step": 2665
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1629717004944455e-05,
+      "loss": 0.4867,
+      "step": 2666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.162448980869995e-05,
+      "loss": 0.5017,
+      "step": 2667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.161926215651582e-05,
+      "loss": 0.461,
+      "step": 2668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1614034049859289e-05,
+      "loss": 0.4891,
+      "step": 2669
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.160880549019771e-05,
+      "loss": 0.4865,
+      "step": 2670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1603576478998561e-05,
+      "loss": 0.4866,
+      "step": 2671
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1598347017729457e-05,
+      "loss": 0.4877,
+      "step": 2672
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1593117107858127e-05,
+      "loss": 0.5105,
+      "step": 2673
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1587886750852434e-05,
+      "loss": 0.4858,
+      "step": 2674
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1582655948180357e-05,
+      "loss": 0.4396,
+      "step": 2675
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.157742470131001e-05,
+      "loss": 0.5135,
+      "step": 2676
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1572193011709627e-05,
+      "loss": 0.4782,
+      "step": 2677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1566960880847572e-05,
+      "loss": 0.5253,
+      "step": 2678
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.156172831019232e-05,
+      "loss": 0.4964,
+      "step": 2679
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1556495301212485e-05,
+      "loss": 0.4945,
+      "step": 2680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1551261855376792e-05,
+      "loss": 0.5082,
+      "step": 2681
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1546027974154095e-05,
+      "loss": 0.4884,
+      "step": 2682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.154079365901337e-05,
+      "loss": 0.4731,
+      "step": 2683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1535558911423706e-05,
+      "loss": 0.4869,
+      "step": 2684
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1530323732854326e-05,
+      "loss": 0.5016,
+      "step": 2685
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.152508812477457e-05,
+      "loss": 0.5469,
+      "step": 2686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1519852088653895e-05,
+      "loss": 0.5322,
+      "step": 2687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1514615625961877e-05,
+      "loss": 0.4848,
+      "step": 2688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1509378738168224e-05,
+      "loss": 0.4862,
+      "step": 2689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1504141426742744e-05,
+      "loss": 0.4539,
+      "step": 2690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.149890369315538e-05,
+      "loss": 0.5259,
+      "step": 2691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1493665538876182e-05,
+      "loss": 0.5112,
+      "step": 2692
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.148842696537533e-05,
+      "loss": 0.5013,
+      "step": 2693
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1483187974123116e-05,
+      "loss": 0.5112,
+      "step": 2694
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.147794856658994e-05,
+      "loss": 0.4677,
+      "step": 2695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1472708744246334e-05,
+      "loss": 0.5068,
+      "step": 2696
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1467468508562943e-05,
+      "loss": 0.5124,
+      "step": 2697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1462227861010513e-05,
+      "loss": 0.478,
+      "step": 2698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1456986803059927e-05,
+      "loss": 0.4768,
+      "step": 2699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1451745336182173e-05,
+      "loss": 0.4729,
+      "step": 2700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1446503461848353e-05,
+      "loss": 0.4853,
+      "step": 2701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1441261181529684e-05,
+      "loss": 0.5183,
+      "step": 2702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.14360184966975e-05,
+      "loss": 0.4719,
+      "step": 2703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1430775408823243e-05,
+      "loss": 0.5446,
+      "step": 2704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1425531919378469e-05,
+      "loss": 0.5112,
+      "step": 2705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1420288029834855e-05,
+      "loss": 0.4901,
+      "step": 2706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1415043741664184e-05,
+      "loss": 0.5144,
+      "step": 2707
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.140979905633835e-05,
+      "loss": 0.4611,
+      "step": 2708
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1404553975329357e-05,
+      "loss": 0.533,
+      "step": 2709
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1399308500109326e-05,
+      "loss": 0.4642,
+      "step": 2710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1394062632150483e-05,
+      "loss": 0.4878,
+      "step": 2711
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1388816372925166e-05,
+      "loss": 0.5049,
+      "step": 2712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.138356972390583e-05,
+      "loss": 0.4818,
+      "step": 2713
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1378322686565026e-05,
+      "loss": 0.5138,
+      "step": 2714
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1373075262375421e-05,
+      "loss": 0.4992,
+      "step": 2715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1367827452809794e-05,
+      "loss": 0.487,
+      "step": 2716
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1362579259341029e-05,
+      "loss": 0.4777,
+      "step": 2717
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1357330683442114e-05,
+      "loss": 0.4653,
+      "step": 2718
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1352081726586144e-05,
+      "loss": 0.4782,
+      "step": 2719
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1346832390246334e-05,
+      "loss": 0.4701,
+      "step": 2720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.134158267589599e-05,
+      "loss": 0.5024,
+      "step": 2721
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1336332585008532e-05,
+      "loss": 0.501,
+      "step": 2722
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1331082119057485e-05,
+      "loss": 0.5276,
+      "step": 2723
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1325831279516475e-05,
+      "loss": 0.5013,
+      "step": 2724
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.132058006785924e-05,
+      "loss": 0.4997,
+      "step": 2725
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.131532848555961e-05,
+      "loss": 0.4982,
+      "step": 2726
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1310076534091539e-05,
+      "loss": 0.4904,
+      "step": 2727
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1304824214929064e-05,
+      "loss": 0.5161,
+      "step": 2728
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1299571529546342e-05,
+      "loss": 0.4886,
+      "step": 2729
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1294318479417618e-05,
+      "loss": 0.508,
+      "step": 2730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1289065066017249e-05,
+      "loss": 0.4876,
+      "step": 2731
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1283811290819693e-05,
+      "loss": 0.4646,
+      "step": 2732
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1278557155299506e-05,
+      "loss": 0.5344,
+      "step": 2733
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1273302660931345e-05,
+      "loss": 0.522,
+      "step": 2734
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1268047809189976e-05,
+      "loss": 0.4928,
+      "step": 2735
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1262792601550254e-05,
+      "loss": 0.4954,
+      "step": 2736
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1257537039487141e-05,
+      "loss": 0.498,
+      "step": 2737
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1252281124475695e-05,
+      "loss": 0.5206,
+      "step": 2738
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1247024857991075e-05,
+      "loss": 0.4763,
+      "step": 2739
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1241768241508537e-05,
+      "loss": 0.4737,
+      "step": 2740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.123651127650344e-05,
+      "loss": 0.4942,
+      "step": 2741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1231253964451235e-05,
+      "loss": 0.4861,
+      "step": 2742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1225996306827471e-05,
+      "loss": 0.4762,
+      "step": 2743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.12207383051078e-05,
+      "loss": 0.4883,
+      "step": 2744
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1215479960767958e-05,
+      "loss": 0.5143,
+      "step": 2745
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1210221275283794e-05,
+      "loss": 0.4959,
+      "step": 2746
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.120496225013124e-05,
+      "loss": 0.5203,
+      "step": 2747
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1199702886786327e-05,
+      "loss": 0.4639,
+      "step": 2748
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1194443186725186e-05,
+      "loss": 0.5363,
+      "step": 2749
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.118918315142403e-05,
+      "loss": 0.5006,
+      "step": 2750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.118392278235918e-05,
+      "loss": 0.5088,
+      "step": 2751
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1178662081007044e-05,
+      "loss": 0.4637,
+      "step": 2752
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.117340104884412e-05,
+      "loss": 0.4974,
+      "step": 2753
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1168139687347003e-05,
+      "loss": 0.4883,
+      "step": 2754
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1162877997992389e-05,
+      "loss": 0.5157,
+      "step": 2755
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1157615982257047e-05,
+      "loss": 0.5178,
+      "step": 2756
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.115235364161785e-05,
+      "loss": 0.5356,
+      "step": 2757
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1147090977551764e-05,
+      "loss": 0.4933,
+      "step": 2758
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.114182799153584e-05,
+      "loss": 0.5038,
+      "step": 2759
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1136564685047213e-05,
+      "loss": 0.472,
+      "step": 2760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1131301059563129e-05,
+      "loss": 0.4746,
+      "step": 2761
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1126037116560905e-05,
+      "loss": 0.469,
+      "step": 2762
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1120772857517947e-05,
+      "loss": 0.4833,
+      "step": 2763
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1115508283911767e-05,
+      "loss": 0.4831,
+      "step": 2764
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1110243397219945e-05,
+      "loss": 0.4783,
+      "step": 2765
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1104978198920158e-05,
+      "loss": 0.4802,
+      "step": 2766
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1099712690490172e-05,
+      "loss": 0.4894,
+      "step": 2767
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1094446873407838e-05,
+      "loss": 0.502,
+      "step": 2768
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1089180749151098e-05,
+      "loss": 0.5038,
+      "step": 2769
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1083914319197967e-05,
+      "loss": 0.5063,
+      "step": 2770
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.107864758502656e-05,
+      "loss": 0.4997,
+      "step": 2771
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1073380548115074e-05,
+      "loss": 0.4957,
+      "step": 2772
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.106811320994178e-05,
+      "loss": 0.4841,
+      "step": 2773
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1062845571985052e-05,
+      "loss": 0.4778,
+      "step": 2774
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1057577635723337e-05,
+      "loss": 0.4859,
+      "step": 2775
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1052309402635164e-05,
+      "loss": 0.4793,
+      "step": 2776
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1047040874199151e-05,
+      "loss": 0.4921,
+      "step": 2777
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1041772051894e-05,
+      "loss": 0.4774,
+      "step": 2778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1036502937198484e-05,
+      "loss": 0.5116,
+      "step": 2779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1031233531591471e-05,
+      "loss": 0.4943,
+      "step": 2780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1025963836551907e-05,
+      "loss": 0.4552,
+      "step": 2781
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1020693853558815e-05,
+      "loss": 0.4817,
+      "step": 2782
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1015423584091306e-05,
+      "loss": 0.4796,
+      "step": 2783
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1010153029628563e-05,
+      "loss": 0.4945,
+      "step": 2784
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1004882191649857e-05,
+      "loss": 0.4959,
+      "step": 2785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.099961107163453e-05,
+      "loss": 0.5319,
+      "step": 2786
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0994339671062012e-05,
+      "loss": 0.5032,
+      "step": 2787
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0989067991411808e-05,
+      "loss": 0.4865,
+      "step": 2788
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.09837960341635e-05,
+      "loss": 0.4712,
+      "step": 2789
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0978523800796747e-05,
+      "loss": 0.5002,
+      "step": 2790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0973251292791292e-05,
+      "loss": 0.5202,
+      "step": 2791
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0967978511626947e-05,
+      "loss": 0.4463,
+      "step": 2792
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0962705458783605e-05,
+      "loss": 0.498,
+      "step": 2793
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.095743213574123e-05,
+      "loss": 0.4906,
+      "step": 2794
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0952158543979878e-05,
+      "loss": 0.4979,
+      "step": 2795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.0946884684979659e-05,
+      "loss": 0.483,
+      "step": 2796
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.094161056022077e-05,
+      "loss": 0.4596,
+      "step": 2797
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0936336171183484e-05,
+      "loss": 0.497,
+      "step": 2798
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0931061519348139e-05,
+      "loss": 0.5555,
+      "step": 2799
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0925786606195153e-05,
+      "loss": 0.4815,
+      "step": 2800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.092051143320502e-05,
+      "loss": 0.4933,
+      "step": 2801
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.09152360018583e-05,
+      "loss": 0.5072,
+      "step": 2802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0909960313635632e-05,
+      "loss": 0.4998,
+      "step": 2803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0904684370017726e-05,
+      "loss": 0.4853,
+      "step": 2804
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0899408172485357e-05,
+      "loss": 0.4921,
+      "step": 2805
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0894131722519376e-05,
+      "loss": 0.4851,
+      "step": 2806
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0888855021600711e-05,
+      "loss": 0.4909,
+      "step": 2807
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0883578071210348e-05,
+      "loss": 0.5264,
+      "step": 2808
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.087830087282935e-05,
+      "loss": 0.4648,
+      "step": 2809
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0873023427938855e-05,
+      "loss": 0.4745,
+      "step": 2810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0867745738020058e-05,
+      "loss": 0.4899,
+      "step": 2811
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0862467804554231e-05,
+      "loss": 0.4364,
+      "step": 2812
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0857189629022713e-05,
+      "loss": 0.496,
+      "step": 2813
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0851911212906909e-05,
+      "loss": 0.4562,
+      "step": 2814
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0846632557688295e-05,
+      "loss": 0.4876,
+      "step": 2815
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0841353664848406e-05,
+      "loss": 0.4894,
+      "step": 2816
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0836074535868857e-05,
+      "loss": 0.5028,
+      "step": 2817
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0830795172231322e-05,
+      "loss": 0.5217,
+      "step": 2818
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.082551557541753e-05,
+      "loss": 0.4886,
+      "step": 2819
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.08202357469093e-05,
+      "loss": 0.485,
+      "step": 2820
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.081495568818849e-05,
+      "loss": 0.4923,
+      "step": 2821
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0809675400737045e-05,
+      "loss": 0.4748,
+      "step": 2822
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0804394886036959e-05,
+      "loss": 0.5513,
+      "step": 2823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0799114145570298e-05,
+      "loss": 0.4667,
+      "step": 2824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0793833180819183e-05,
+      "loss": 0.5125,
+      "step": 2825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0788551993265804e-05,
+      "loss": 0.4919,
+      "step": 2826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0783270584392418e-05,
+      "loss": 0.4575,
+      "step": 2827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0777988955681331e-05,
+      "loss": 0.4957,
+      "step": 2828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0772707108614923e-05,
+      "loss": 0.5028,
+      "step": 2829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0767425044675634e-05,
+      "loss": 0.5076,
+      "step": 2830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0762142765345955e-05,
+      "loss": 0.508,
+      "step": 2831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.075686027210845e-05,
+      "loss": 0.525,
+      "step": 2832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0751577566445732e-05,
+      "loss": 0.497,
+      "step": 2833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0746294649840481e-05,
+      "loss": 0.4831,
+      "step": 2834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0741011523775433e-05,
+      "loss": 0.4795,
+      "step": 2835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0735728189733386e-05,
+      "loss": 0.4814,
+      "step": 2836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0730444649197191e-05,
+      "loss": 0.497,
+      "step": 2837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0725160903649765e-05,
+      "loss": 0.4884,
+      "step": 2838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0719876954574071e-05,
+      "loss": 0.4925,
+      "step": 2839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0714592803453138e-05,
+      "loss": 0.5302,
+      "step": 2840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0709308451770053e-05,
+      "loss": 0.4754,
+      "step": 2841
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.070402390100795e-05,
+      "loss": 0.4796,
+      "step": 2842
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0698739152650031e-05,
+      "loss": 0.4843,
+      "step": 2843
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0693454208179544e-05,
+      "loss": 0.4846,
+      "step": 2844
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0688169069079793e-05,
+      "loss": 0.4903,
+      "step": 2845
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.068288373683414e-05,
+      "loss": 0.5108,
+      "step": 2846
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0677598212926001e-05,
+      "loss": 0.5432,
+      "step": 2847
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0672312498838844e-05,
+      "loss": 0.4934,
+      "step": 2848
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0667026596056186e-05,
+      "loss": 0.5237,
+      "step": 2849
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0661740506061616e-05,
+      "loss": 0.5274,
+      "step": 2850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.065645423033875e-05,
+      "loss": 0.5007,
+      "step": 2851
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0651167770371267e-05,
+      "loss": 0.5235,
+      "step": 2852
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0645881127642907e-05,
+      "loss": 0.4926,
+      "step": 2853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0640594303637444e-05,
+      "loss": 0.5101,
+      "step": 2854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0635307299838715e-05,
+      "loss": 0.4413,
+      "step": 2855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0630020117730606e-05,
+      "loss": 0.4903,
+      "step": 2856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.062473275879705e-05,
+      "loss": 0.4597,
+      "step": 2857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0619445224522027e-05,
+      "loss": 0.528,
+      "step": 2858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0614157516389579e-05,
+      "loss": 0.4897,
+      "step": 2859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0608869635883776e-05,
+      "loss": 0.4912,
+      "step": 2860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0603581584488755e-05,
+      "loss": 0.4837,
+      "step": 2861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0598293363688694e-05,
+      "loss": 0.5153,
+      "step": 2862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0593004974967817e-05,
+      "loss": 0.4801,
+      "step": 2863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.05877164198104e-05,
+      "loss": 0.508,
+      "step": 2864
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0582427699700759e-05,
+      "loss": 0.5199,
+      "step": 2865
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.057713881612326e-05,
+      "loss": 0.4843,
+      "step": 2866
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0571849770562316e-05,
+      "loss": 0.4659,
+      "step": 2867
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0566560564502384e-05,
+      "loss": 0.5031,
+      "step": 2868
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0561271199427965e-05,
+      "loss": 0.4719,
+      "step": 2869
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0555981676823606e-05,
+      "loss": 0.5044,
+      "step": 2870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0550691998173897e-05,
+      "loss": 0.4975,
+      "step": 2871
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0545402164963476e-05,
+      "loss": 0.4609,
+      "step": 2872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0540112178677022e-05,
+      "loss": 0.5131,
+      "step": 2873
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.053482204079925e-05,
+      "loss": 0.4446,
+      "step": 2874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0529531752814928e-05,
+      "loss": 0.4816,
+      "step": 2875
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.052424131620886e-05,
+      "loss": 0.4872,
+      "step": 2876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0518950732465895e-05,
+      "loss": 0.52,
+      "step": 2877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0513660003070924e-05,
+      "loss": 0.5158,
+      "step": 2878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0508369129508876e-05,
+      "loss": 0.4921,
+      "step": 2879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0503078113264715e-05,
+      "loss": 0.4898,
+      "step": 2880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0497786955823457e-05,
+      "loss": 0.4705,
+      "step": 2881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0492495658670151e-05,
+      "loss": 0.4777,
+      "step": 2882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0487204223289882e-05,
+      "loss": 0.502,
+      "step": 2883
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0481912651167784e-05,
+      "loss": 0.4947,
+      "step": 2884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0476620943789021e-05,
+      "loss": 0.5034,
+      "step": 2885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0471329102638799e-05,
+      "loss": 0.5059,
+      "step": 2886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0466037129202356e-05,
+      "loss": 0.4792,
+      "step": 2887
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.046074502496497e-05,
+      "loss": 0.4651,
+      "step": 2888
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0455452791411962e-05,
+      "loss": 0.521,
+      "step": 2889
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0450160430028679e-05,
+      "loss": 0.4928,
+      "step": 2890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0444867942300512e-05,
+      "loss": 0.4638,
+      "step": 2891
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0439575329712883e-05,
+      "loss": 0.4608,
+      "step": 2892
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.043428259375125e-05,
+      "loss": 0.4649,
+      "step": 2893
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0428989735901104e-05,
+      "loss": 0.4693,
+      "step": 2894
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0423696757647977e-05,
+      "loss": 0.5068,
+      "step": 2895
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0418403660477425e-05,
+      "loss": 0.4996,
+      "step": 2896
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0413110445875046e-05,
+      "loss": 0.4853,
+      "step": 2897
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0407817115326463e-05,
+      "loss": 0.4898,
+      "step": 2898
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0402523670317343e-05,
+      "loss": 0.5222,
+      "step": 2899
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.039723011233337e-05,
+      "loss": 0.4907,
+      "step": 2900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0391936442860271e-05,
+      "loss": 0.5157,
+      "step": 2901
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0386642663383802e-05,
+      "loss": 0.4984,
+      "step": 2902
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0381348775389745e-05,
+      "loss": 0.4725,
+      "step": 2903
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0376054780363917e-05,
+      "loss": 0.4468,
+      "step": 2904
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0370760679792173e-05,
+      "loss": 0.4809,
+      "step": 2905
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0365466475160377e-05,
+      "loss": 0.4874,
+      "step": 2906
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0360172167954439e-05,
+      "loss": 0.4814,
+      "step": 2907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0354877759660296e-05,
+      "loss": 0.4635,
+      "step": 2908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0349583251763905e-05,
+      "loss": 0.5117,
+      "step": 2909
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0344288645751257e-05,
+      "loss": 0.4821,
+      "step": 2910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.033899394310837e-05,
+      "loss": 0.4777,
+      "step": 2911
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0333699145321294e-05,
+      "loss": 0.4786,
+      "step": 2912
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0328404253876096e-05,
+      "loss": 0.4784,
+      "step": 2913
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0323109270258873e-05,
+      "loss": 0.5113,
+      "step": 2914
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.031781419595575e-05,
+      "loss": 0.4958,
+      "step": 2915
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0312519032452877e-05,
+      "loss": 0.4805,
+      "step": 2916
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0307223781236424e-05,
+      "loss": 0.4957,
+      "step": 2917
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0301928443792598e-05,
+      "loss": 0.4952,
+      "step": 2918
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0296633021607617e-05,
+      "loss": 0.4686,
+      "step": 2919
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0291337516167725e-05,
+      "loss": 0.4754,
+      "step": 2920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0286041928959197e-05,
+      "loss": 0.4714,
+      "step": 2921
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0280746261468326e-05,
+      "loss": 0.4981,
+      "step": 2922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0275450515181424e-05,
+      "loss": 0.5099,
+      "step": 2923
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0270154691584833e-05,
+      "loss": 0.5042,
+      "step": 2924
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0264858792164908e-05,
+      "loss": 0.5074,
+      "step": 2925
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0259562818408033e-05,
+      "loss": 0.4982,
+      "step": 2926
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0254266771800609e-05,
+      "loss": 0.5062,
+      "step": 2927
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0248970653829063e-05,
+      "loss": 0.5191,
+      "step": 2928
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0243674465979825e-05,
+      "loss": 0.541,
+      "step": 2929
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0238378209739366e-05,
+      "loss": 0.5299,
+      "step": 2930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0233081886594165e-05,
+      "loss": 0.4885,
+      "step": 2931
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0227785498030722e-05,
+      "loss": 0.484,
+      "step": 2932
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0222489045535553e-05,
+      "loss": 0.4875,
+      "step": 2933
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0217192530595196e-05,
+      "loss": 0.5089,
+      "step": 2934
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0211895954696204e-05,
+      "loss": 0.4779,
+      "step": 2935
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0206599319325148e-05,
+      "loss": 0.4459,
+      "step": 2936
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0201302625968616e-05,
+      "loss": 0.5009,
+      "step": 2937
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0196005876113209e-05,
+      "loss": 0.4415,
+      "step": 2938
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0190709071245547e-05,
+      "loss": 0.5037,
+      "step": 2939
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0185412212852268e-05,
+      "loss": 0.4729,
+      "step": 2940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0180115302420019e-05,
+      "loss": 0.4909,
+      "step": 2941
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0174818341435466e-05,
+      "loss": 0.51,
+      "step": 2942
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0169521331385287e-05,
+      "loss": 0.5205,
+      "step": 2943
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0164224273756172e-05,
+      "loss": 0.4777,
+      "step": 2944
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0158927170034831e-05,
+      "loss": 0.4962,
+      "step": 2945
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0153630021707982e-05,
+      "loss": 0.5115,
+      "step": 2946
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0148332830262352e-05,
+      "loss": 0.4814,
+      "step": 2947
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0143035597184691e-05,
+      "loss": 0.4373,
+      "step": 2948
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.013773832396175e-05,
+      "loss": 0.491,
+      "step": 2949
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0132441012080296e-05,
+      "loss": 0.5256,
+      "step": 2950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0127143663027106e-05,
+      "loss": 0.5094,
+      "step": 2951
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.012184627828897e-05,
+      "loss": 0.5509,
+      "step": 2952
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0116548859352682e-05,
+      "loss": 0.4828,
+      "step": 2953
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0111251407705052e-05,
+      "loss": 0.502,
+      "step": 2954
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0105953924832894e-05,
+      "loss": 0.5207,
+      "step": 2955
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0100656412223035e-05,
+      "loss": 0.518,
+      "step": 2956
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.009535887136231e-05,
+      "loss": 0.4915,
+      "step": 2957
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0090061303737555e-05,
+      "loss": 0.4972,
+      "step": 2958
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0084763710835624e-05,
+      "loss": 0.5316,
+      "step": 2959
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0079466094143373e-05,
+      "loss": 0.4677,
+      "step": 2960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0074168455147662e-05,
+      "loss": 0.5158,
+      "step": 2961
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.006887079533536e-05,
+      "loss": 0.5006,
+      "step": 2962
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0063573116193346e-05,
+      "loss": 0.4925,
+      "step": 2963
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0058275419208496e-05,
+      "loss": 0.5025,
+      "step": 2964
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0052977705867697e-05,
+      "loss": 0.5037,
+      "step": 2965
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0047679977657836e-05,
+      "loss": 0.5,
+      "step": 2966
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0042382236065814e-05,
+      "loss": 0.5016,
+      "step": 2967
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0037084482578523e-05,
+      "loss": 0.4967,
+      "step": 2968
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0031786718682866e-05,
+      "loss": 0.457,
+      "step": 2969
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0026488945865744e-05,
+      "loss": 0.4925,
+      "step": 2970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.002119116561407e-05,
+      "loss": 0.4812,
+      "step": 2971
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0015893379414744e-05,
+      "loss": 0.4598,
+      "step": 2972
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0010595588754683e-05,
+      "loss": 0.506,
+      "step": 2973
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.00052977951208e-05,
+      "loss": 0.4896,
+      "step": 2974
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1e-05,
+      "loss": 0.5029,
+      "step": 2975
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.994702204879203e-06,
+      "loss": 0.4762,
+      "step": 2976
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.989404411245316e-06,
+      "loss": 0.4894,
+      "step": 2977
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.984106620585258e-06,
+      "loss": 0.5094,
+      "step": 2978
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.978808834385936e-06,
+      "loss": 0.4987,
+      "step": 2979
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.973511054134259e-06,
+      "loss": 0.5028,
+      "step": 2980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96821328131714e-06,
+      "loss": 0.4956,
+      "step": 2981
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96291551742148e-06,
+      "loss": 0.5067,
+      "step": 2982
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.957617763934188e-06,
+      "loss": 0.5024,
+      "step": 2983
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.952320022342165e-06,
+      "loss": 0.5173,
+      "step": 2984
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.947022294132306e-06,
+      "loss": 0.4876,
+      "step": 2985
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.941724580791507e-06,
+      "loss": 0.4784,
+      "step": 2986
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.936426883806657e-06,
+      "loss": 0.4727,
+      "step": 2987
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.93112920466464e-06,
+      "loss": 0.48,
+      "step": 2988
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.92583154485234e-06,
+      "loss": 0.4937,
+      "step": 2989
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.920533905856634e-06,
+      "loss": 0.522,
+      "step": 2990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.915236289164381e-06,
+      "loss": 0.4809,
+      "step": 2991
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.909938696262447e-06,
+      "loss": 0.4693,
+      "step": 2992
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.904641128637693e-06,
+      "loss": 0.4934,
+      "step": 2993
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.899343587776966e-06,
+      "loss": 0.5,
+      "step": 2994
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.894046075167106e-06,
+      "loss": 0.4683,
+      "step": 2995
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.888748592294953e-06,
+      "loss": 0.4824,
+      "step": 2996
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.883451140647323e-06,
+      "loss": 0.49,
+      "step": 2997
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.878153721711034e-06,
+      "loss": 0.5117,
+      "step": 2998
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.872856336972896e-06,
+      "loss": 0.4748,
+      "step": 2999
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.867558987919704e-06,
+      "loss": 0.5105,
+      "step": 3000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.862261676038254e-06,
+      "loss": 0.4925,
+      "step": 3001
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.856964402815312e-06,
+      "loss": 0.5202,
+      "step": 3002
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.851667169737651e-06,
+      "loss": 0.4646,
+      "step": 3003
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.846369978292022e-06,
+      "loss": 0.4966,
+      "step": 3004
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.841072829965172e-06,
+      "loss": 0.4671,
+      "step": 3005
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.83577572624383e-06,
+      "loss": 0.5169,
+      "step": 3006
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.830478668614718e-06,
+      "loss": 0.4912,
+      "step": 3007
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.825181658564539e-06,
+      "loss": 0.4678,
+      "step": 3008
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.819884697579984e-06,
+      "loss": 0.5064,
+      "step": 3009
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.814587787147735e-06,
+      "loss": 0.48,
+      "step": 3010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.809290928754454e-06,
+      "loss": 0.501,
+      "step": 3011
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.803994123886793e-06,
+      "loss": 0.4783,
+      "step": 3012
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.798697374031388e-06,
+      "loss": 0.4799,
+      "step": 3013
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.793400680674853e-06,
+      "loss": 0.4619,
+      "step": 3014
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.788104045303797e-06,
+      "loss": 0.4934,
+      "step": 3015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.782807469404805e-06,
+      "loss": 0.5111,
+      "step": 3016
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.777510954464448e-06,
+      "loss": 0.5035,
+      "step": 3017
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.77221450196928e-06,
+      "loss": 0.5074,
+      "step": 3018
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.766918113405838e-06,
+      "loss": 0.4702,
+      "step": 3019
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.761621790260636e-06,
+      "loss": 0.456,
+      "step": 3020
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.756325534020177e-06,
+      "loss": 0.469,
+      "step": 3021
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.751029346170942e-06,
+      "loss": 0.4753,
+      "step": 3022
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.74573322819939e-06,
+      "loss": 0.4956,
+      "step": 3023
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.740437181591967e-06,
+      "loss": 0.437,
+      "step": 3024
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.735141207835095e-06,
+      "loss": 0.4146,
+      "step": 3025
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.72984530841517e-06,
+      "loss": 0.4427,
+      "step": 3026
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.724549484818578e-06,
+      "loss": 0.4721,
+      "step": 3027
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.719253738531676e-06,
+      "loss": 0.439,
+      "step": 3028
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.713958071040803e-06,
+      "loss": 0.4536,
+      "step": 3029
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.708662483832279e-06,
+      "loss": 0.4351,
+      "step": 3030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.703366978392388e-06,
+      "loss": 0.4254,
+      "step": 3031
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.698071556207407e-06,
+      "loss": 0.4023,
+      "step": 3032
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.69277621876358e-06,
+      "loss": 0.4288,
+      "step": 3033
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.687480967547127e-06,
+      "loss": 0.4396,
+      "step": 3034
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.682185804044252e-06,
+      "loss": 0.4131,
+      "step": 3035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.676890729741134e-06,
+      "loss": 0.4208,
+      "step": 3036
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.67159574612391e-06,
+      "loss": 0.4434,
+      "step": 3037
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.66630085467871e-06,
+      "loss": 0.4389,
+      "step": 3038
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.661006056891631e-06,
+      "loss": 0.4429,
+      "step": 3039
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.655711354248747e-06,
+      "loss": 0.4069,
+      "step": 3040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.650416748236099e-06,
+      "loss": 0.4562,
+      "step": 3041
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.645122240339709e-06,
+      "loss": 0.4161,
+      "step": 3042
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.639827832045564e-06,
+      "loss": 0.4611,
+      "step": 3043
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.634533524839626e-06,
+      "loss": 0.4519,
+      "step": 3044
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.62923932020783e-06,
+      "loss": 0.4385,
+      "step": 3045
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.623945219636081e-06,
+      "loss": 0.4605,
+      "step": 3046
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.618651224610257e-06,
+      "loss": 0.4065,
+      "step": 3047
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.613357336616203e-06,
+      "loss": 0.4553,
+      "step": 3048
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.608063557139732e-06,
+      "loss": 0.4096,
+      "step": 3049
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.602769887666633e-06,
+      "loss": 0.4256,
+      "step": 3050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.59747632968266e-06,
+      "loss": 0.4498,
+      "step": 3051
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.592182884673536e-06,
+      "loss": 0.4204,
+      "step": 3052
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.586889554124957e-06,
+      "loss": 0.4516,
+      "step": 3053
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.581596339522576e-06,
+      "loss": 0.4111,
+      "step": 3054
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.576303242352025e-06,
+      "loss": 0.4505,
+      "step": 3055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.571010264098897e-06,
+      "loss": 0.4205,
+      "step": 3056
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.565717406248752e-06,
+      "loss": 0.417,
+      "step": 3057
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.560424670287119e-06,
+      "loss": 0.441,
+      "step": 3058
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.555132057699493e-06,
+      "loss": 0.41,
+      "step": 3059
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.549839569971323e-06,
+      "loss": 0.4201,
+      "step": 3060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.54454720858804e-06,
+      "loss": 0.403,
+      "step": 3061
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.539254975035031e-06,
+      "loss": 0.3843,
+      "step": 3062
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.533962870797646e-06,
+      "loss": 0.413,
+      "step": 3063
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.528670897361203e-06,
+      "loss": 0.4564,
+      "step": 3064
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.523379056210982e-06,
+      "loss": 0.4398,
+      "step": 3065
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.518087348832219e-06,
+      "loss": 0.4408,
+      "step": 3066
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.512795776710122e-06,
+      "loss": 0.4469,
+      "step": 3067
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.507504341329852e-06,
+      "loss": 0.4652,
+      "step": 3068
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.502213044176545e-06,
+      "loss": 0.4121,
+      "step": 3069
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.496921886735287e-06,
+      "loss": 0.4466,
+      "step": 3070
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.491630870491131e-06,
+      "loss": 0.4442,
+      "step": 3071
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.486339996929079e-06,
+      "loss": 0.4356,
+      "step": 3072
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.481049267534106e-06,
+      "loss": 0.442,
+      "step": 3073
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.475758683791142e-06,
+      "loss": 0.4661,
+      "step": 3074
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.470468247185076e-06,
+      "loss": 0.4441,
+      "step": 3075
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.465177959200756e-06,
+      "loss": 0.4245,
+      "step": 3076
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.459887821322983e-06,
+      "loss": 0.4399,
+      "step": 3077
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.454597835036527e-06,
+      "loss": 0.4381,
+      "step": 3078
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.449308001826104e-06,
+      "loss": 0.4601,
+      "step": 3079
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.444018323176399e-06,
+      "loss": 0.4666,
+      "step": 3080
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.43872880057204e-06,
+      "loss": 0.4482,
+      "step": 3081
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.433439435497621e-06,
+      "loss": 0.4147,
+      "step": 3082
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.428150229437689e-06,
+      "loss": 0.4019,
+      "step": 3083
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.422861183876742e-06,
+      "loss": 0.4396,
+      "step": 3084
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.417572300299244e-06,
+      "loss": 0.4445,
+      "step": 3085
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.412283580189601e-06,
+      "loss": 0.4139,
+      "step": 3086
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.406995025032183e-06,
+      "loss": 0.4168,
+      "step": 3087
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.40170663631131e-06,
+      "loss": 0.4385,
+      "step": 3088
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.396418415511248e-06,
+      "loss": 0.4235,
+      "step": 3089
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.391130364116226e-06,
+      "loss": 0.4682,
+      "step": 3090
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.385842483610426e-06,
+      "loss": 0.4274,
+      "step": 3091
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.380554775477974e-06,
+      "loss": 0.4169,
+      "step": 3092
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.375267241202952e-06,
+      "loss": 0.3905,
+      "step": 3093
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.369979882269397e-06,
+      "loss": 0.4587,
+      "step": 3094
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.364692700161287e-06,
+      "loss": 0.4684,
+      "step": 3095
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.35940569636256e-06,
+      "loss": 0.4058,
+      "step": 3096
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.354118872357096e-06,
+      "loss": 0.4391,
+      "step": 3097
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.348832229628733e-06,
+      "loss": 0.4388,
+      "step": 3098
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.343545769661252e-06,
+      "loss": 0.4554,
+      "step": 3099
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.33825949393839e-06,
+      "loss": 0.412,
+      "step": 3100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.332973403943815e-06,
+      "loss": 0.448,
+      "step": 3101
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.327687501161158e-06,
+      "loss": 0.4424,
+      "step": 3102
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.322401787074e-06,
+      "loss": 0.4538,
+      "step": 3103
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.317116263165862e-06,
+      "loss": 0.4263,
+      "step": 3104
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.311830930920214e-06,
+      "loss": 0.4219,
+      "step": 3105
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.306545791820461e-06,
+      "loss": 0.4664,
+      "step": 3106
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.301260847349974e-06,
+      "loss": 0.4363,
+      "step": 3107
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.295976098992053e-06,
+      "loss": 0.4299,
+      "step": 3108
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.29069154822995e-06,
+      "loss": 0.4489,
+      "step": 3109
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.285407196546862e-06,
+      "loss": 0.4411,
+      "step": 3110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.280123045425936e-06,
+      "loss": 0.4333,
+      "step": 3111
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.274839096350241e-06,
+      "loss": 0.4369,
+      "step": 3112
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.269555350802812e-06,
+      "loss": 0.435,
+      "step": 3113
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.264271810266618e-06,
+      "loss": 0.4532,
+      "step": 3114
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.25898847622457e-06,
+      "loss": 0.4396,
+      "step": 3115
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.253705350159522e-06,
+      "loss": 0.4343,
+      "step": 3116
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.248422433554273e-06,
+      "loss": 0.4988,
+      "step": 3117
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.243139727891554e-06,
+      "loss": 0.4265,
+      "step": 3118
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.237857234654048e-06,
+      "loss": 0.4469,
+      "step": 3119
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.232574955324369e-06,
+      "loss": 0.4516,
+      "step": 3120
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.227292891385078e-06,
+      "loss": 0.4299,
+      "step": 3121
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.22201104431867e-06,
+      "loss": 0.4532,
+      "step": 3122
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.216729415607588e-06,
+      "loss": 0.4405,
+      "step": 3123
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.211448006734199e-06,
+      "loss": 0.4121,
+      "step": 3124
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.206166819180822e-06,
+      "loss": 0.4103,
+      "step": 3125
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.200885854429706e-06,
+      "loss": 0.4354,
+      "step": 3126
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.195605113963042e-06,
+      "loss": 0.4422,
+      "step": 3127
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.190324599262957e-06,
+      "loss": 0.4214,
+      "step": 3128
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.185044311811511e-06,
+      "loss": 0.4335,
+      "step": 3129
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.179764253090703e-06,
+      "loss": 0.4208,
+      "step": 3130
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.174484424582471e-06,
+      "loss": 0.4444,
+      "step": 3131
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.169204827768683e-06,
+      "loss": 0.4358,
+      "step": 3132
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.163925464131143e-06,
+      "loss": 0.4224,
+      "step": 3133
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.158646335151598e-06,
+      "loss": 0.4395,
+      "step": 3134
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.153367442311712e-06,
+      "loss": 0.4383,
+      "step": 3135
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.148088787093093e-06,
+      "loss": 0.4676,
+      "step": 3136
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.142810370977289e-06,
+      "loss": 0.4364,
+      "step": 3137
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.137532195445769e-06,
+      "loss": 0.4792,
+      "step": 3138
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.132254261979943e-06,
+      "loss": 0.4508,
+      "step": 3139
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.12697657206115e-06,
+      "loss": 0.45,
+      "step": 3140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.121699127170652e-06,
+      "loss": 0.4413,
+      "step": 3141
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.116421928789655e-06,
+      "loss": 0.4299,
+      "step": 3142
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.111144978399292e-06,
+      "loss": 0.4424,
+      "step": 3143
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.105868277480622e-06,
+      "loss": 0.4621,
+      "step": 3144
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.100591827514643e-06,
+      "loss": 0.445,
+      "step": 3145
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.09531562998228e-06,
+      "loss": 0.4271,
+      "step": 3146
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.09003968636437e-06,
+      "loss": 0.4619,
+      "step": 3147
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.084763998141703e-06,
+      "loss": 0.4598,
+      "step": 3148
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.079488566794984e-06,
+      "loss": 0.4174,
+      "step": 3149
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.07421339380485e-06,
+      "loss": 0.4865,
+      "step": 3150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.068938480651868e-06,
+      "loss": 0.4655,
+      "step": 3151
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.063663828816523e-06,
+      "loss": 0.4233,
+      "step": 3152
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.058389439779233e-06,
+      "loss": 0.4591,
+      "step": 3153
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.053115315020344e-06,
+      "loss": 0.4367,
+      "step": 3154
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.047841456020125e-06,
+      "loss": 0.4348,
+      "step": 3155
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.042567864258768e-06,
+      "loss": 0.4011,
+      "step": 3156
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.0372945412164e-06,
+      "loss": 0.4606,
+      "step": 3157
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.032021488373058e-06,
+      "loss": 0.3948,
+      "step": 3158
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.026748707208712e-06,
+      "loss": 0.492,
+      "step": 3159
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.021476199203255e-06,
+      "loss": 0.4212,
+      "step": 3160
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.016203965836503e-06,
+      "loss": 0.4398,
+      "step": 3161
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.010932008588194e-06,
+      "loss": 0.4554,
+      "step": 3162
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.00566032893799e-06,
+      "loss": 0.4753,
+      "step": 3163
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.000388928365473e-06,
+      "loss": 0.4076,
+      "step": 3164
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.995117808350146e-06,
+      "loss": 0.4389,
+      "step": 3165
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.989846970371438e-06,
+      "loss": 0.4405,
+      "step": 3166
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.984576415908696e-06,
+      "loss": 0.4117,
+      "step": 3167
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.979306146441185e-06,
+      "loss": 0.4344,
+      "step": 3168
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.974036163448098e-06,
+      "loss": 0.4781,
+      "step": 3169
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.968766468408532e-06,
+      "loss": 0.4679,
+      "step": 3170
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.96349706280152e-06,
+      "loss": 0.4317,
+      "step": 3171
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.958227948106005e-06,
+      "loss": 0.4041,
+      "step": 3172
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.95295912580085e-06,
+      "loss": 0.4189,
+      "step": 3173
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.947690597364836e-06,
+      "loss": 0.4271,
+      "step": 3174
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.942422364276668e-06,
+      "loss": 0.4367,
+      "step": 3175
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.937154428014951e-06,
+      "loss": 0.4442,
+      "step": 3176
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.931886790058223e-06,
+      "loss": 0.4431,
+      "step": 3177
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.92661945188493e-06,
+      "loss": 0.469,
+      "step": 3178
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.921352414973441e-06,
+      "loss": 0.4279,
+      "step": 3179
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.916085680802038e-06,
+      "loss": 0.4393,
+      "step": 3180
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.910819250848907e-06,
+      "loss": 0.4693,
+      "step": 3181
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.905553126592164e-06,
+      "loss": 0.4108,
+      "step": 3182
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.900287309509831e-06,
+      "loss": 0.4587,
+      "step": 3183
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.895021801079846e-06,
+      "loss": 0.4543,
+      "step": 3184
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.889756602780059e-06,
+      "loss": 0.4242,
+      "step": 3185
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.884491716088238e-06,
+      "loss": 0.4155,
+      "step": 3186
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.879227142482055e-06,
+      "loss": 0.433,
+      "step": 3187
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.8739628834391e-06,
+      "loss": 0.4293,
+      "step": 3188
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.868698940436874e-06,
+      "loss": 0.4508,
+      "step": 3189
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.863435314952787e-06,
+      "loss": 0.4317,
+      "step": 3190
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.858172008464164e-06,
+      "loss": 0.4433,
+      "step": 3191
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.852909022448239e-06,
+      "loss": 0.4192,
+      "step": 3192
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.847646358382153e-06,
+      "loss": 0.4316,
+      "step": 3193
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.842384017742956e-06,
+      "loss": 0.4411,
+      "step": 3194
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.837122002007614e-06,
+      "loss": 0.412,
+      "step": 3195
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.831860312652995e-06,
+      "loss": 0.4164,
+      "step": 3196
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.82659895115588e-06,
+      "loss": 0.4399,
+      "step": 3197
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.821337918992961e-06,
+      "loss": 0.4652,
+      "step": 3198
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.816077217640822e-06,
+      "loss": 0.4325,
+      "step": 3199
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.810816848575971e-06,
+      "loss": 0.4143,
+      "step": 3200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.805556813274817e-06,
+      "loss": 0.4277,
+      "step": 3201
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.800297113213673e-06,
+      "loss": 0.4151,
+      "step": 3202
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.795037749868764e-06,
+      "loss": 0.4328,
+      "step": 3203
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.789778724716209e-06,
+      "loss": 0.4202,
+      "step": 3204
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.784520039232044e-06,
+      "loss": 0.4346,
+      "step": 3205
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.779261694892205e-06,
+      "loss": 0.4752,
+      "step": 3206
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.77400369317253e-06,
+      "loss": 0.4587,
+      "step": 3207
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.768746035548767e-06,
+      "loss": 0.4165,
+      "step": 3208
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.763488723496565e-06,
+      "loss": 0.4291,
+      "step": 3209
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.758231758491467e-06,
+      "loss": 0.4129,
+      "step": 3210
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.752975142008928e-06,
+      "loss": 0.4158,
+      "step": 3211
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.747718875524307e-06,
+      "loss": 0.4357,
+      "step": 3212
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.74246296051286e-06,
+      "loss": 0.4941,
+      "step": 3213
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.737207398449746e-06,
+      "loss": 0.4942,
+      "step": 3214
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.731952190810029e-06,
+      "loss": 0.4383,
+      "step": 3215
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.726697339068657e-06,
+      "loss": 0.4342,
+      "step": 3216
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.721442844700499e-06,
+      "loss": 0.4577,
+      "step": 3217
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.71618870918031e-06,
+      "loss": 0.4291,
+      "step": 3218
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.71093493398275e-06,
+      "loss": 0.4078,
+      "step": 3219
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.705681520582382e-06,
+      "loss": 0.4073,
+      "step": 3220
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.700428470453663e-06,
+      "loss": 0.4608,
+      "step": 3221
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.695175785070938e-06,
+      "loss": 0.413,
+      "step": 3222
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.689923465908464e-06,
+      "loss": 0.4612,
+      "step": 3223
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.684671514440391e-06,
+      "loss": 0.4355,
+      "step": 3224
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.679419932140765e-06,
+      "loss": 0.4477,
+      "step": 3225
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.67416872048353e-06,
+      "loss": 0.4294,
+      "step": 3226
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.66891788094252e-06,
+      "loss": 0.4199,
+      "step": 3227
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.66366741499147e-06,
+      "loss": 0.4324,
+      "step": 3228
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.658417324104011e-06,
+      "loss": 0.421,
+      "step": 3229
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.653167609753667e-06,
+      "loss": 0.4279,
+      "step": 3230
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.647918273413856e-06,
+      "loss": 0.4307,
+      "step": 3231
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.642669316557893e-06,
+      "loss": 0.4243,
+      "step": 3232
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.637420740658976e-06,
+      "loss": 0.4207,
+      "step": 3233
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.632172547190208e-06,
+      "loss": 0.4335,
+      "step": 3234
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.62692473762458e-06,
+      "loss": 0.4559,
+      "step": 3235
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.621677313434977e-06,
+      "loss": 0.469,
+      "step": 3236
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.616430276094172e-06,
+      "loss": 0.4419,
+      "step": 3237
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.611183627074835e-06,
+      "loss": 0.4237,
+      "step": 3238
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.60593736784952e-06,
+      "loss": 0.4596,
+      "step": 3239
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.600691499890677e-06,
+      "loss": 0.4473,
+      "step": 3240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.595446024670644e-06,
+      "loss": 0.4466,
+      "step": 3241
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.590200943661652e-06,
+      "loss": 0.4342,
+      "step": 3242
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.584956258335816e-06,
+      "loss": 0.4702,
+      "step": 3243
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.579711970165148e-06,
+      "loss": 0.4397,
+      "step": 3244
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.574468080621533e-06,
+      "loss": 0.4034,
+      "step": 3245
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.56922459117676e-06,
+      "loss": 0.4269,
+      "step": 3246
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.563981503302503e-06,
+      "loss": 0.4427,
+      "step": 3247
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.558738818470317e-06,
+      "loss": 0.4177,
+      "step": 3248
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.553496538151647e-06,
+      "loss": 0.4476,
+      "step": 3249
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.54825466381783e-06,
+      "loss": 0.4454,
+      "step": 3250
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.543013196940075e-06,
+      "loss": 0.4845,
+      "step": 3251
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.53777213898949e-06,
+      "loss": 0.4478,
+      "step": 3252
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.532531491437062e-06,
+      "loss": 0.4501,
+      "step": 3253
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.527291255753666e-06,
+      "loss": 0.4996,
+      "step": 3254
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.522051433410064e-06,
+      "loss": 0.4275,
+      "step": 3255
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.516812025876891e-06,
+      "loss": 0.3928,
+      "step": 3256
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.511573034624673e-06,
+      "loss": 0.4554,
+      "step": 3257
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.50633446112382e-06,
+      "loss": 0.4638,
+      "step": 3258
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.501096306844624e-06,
+      "loss": 0.421,
+      "step": 3259
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.495858573257258e-06,
+      "loss": 0.4767,
+      "step": 3260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.490621261831781e-06,
+      "loss": 0.4308,
+      "step": 3261
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.485384374038124e-06,
+      "loss": 0.4348,
+      "step": 3262
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.480147911346108e-06,
+      "loss": 0.4315,
+      "step": 3263
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.474911875225432e-06,
+      "loss": 0.4272,
+      "step": 3264
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.469676267145674e-06,
+      "loss": 0.4492,
+      "step": 3265
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.464441088576296e-06,
+      "loss": 0.4666,
+      "step": 3266
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.459206340986637e-06,
+      "loss": 0.4391,
+      "step": 3267
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.453972025845908e-06,
+      "loss": 0.4728,
+      "step": 3268
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.448738144623212e-06,
+      "loss": 0.4391,
+      "step": 3269
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.443504698787517e-06,
+      "loss": 0.4268,
+      "step": 3270
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.438271689807682e-06,
+      "loss": 0.4355,
+      "step": 3271
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.43303911915243e-06,
+      "loss": 0.4216,
+      "step": 3272
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.427806988290374e-06,
+      "loss": 0.44,
+      "step": 3273
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.422575298689993e-06,
+      "loss": 0.4194,
+      "step": 3274
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.417344051819646e-06,
+      "loss": 0.4382,
+      "step": 3275
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.412113249147571e-06,
+      "loss": 0.4844,
+      "step": 3276
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.406882892141875e-06,
+      "loss": 0.4389,
+      "step": 3277
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.401652982270548e-06,
+      "loss": 0.4001,
+      "step": 3278
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.39642352100144e-06,
+      "loss": 0.4499,
+      "step": 3279
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.391194509802294e-06,
+      "loss": 0.4405,
+      "step": 3280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.385965950140714e-06,
+      "loss": 0.4058,
+      "step": 3281
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.380737843484181e-06,
+      "loss": 0.4037,
+      "step": 3282
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.37551019130005e-06,
+      "loss": 0.4304,
+      "step": 3283
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.37028299505555e-06,
+      "loss": 0.4666,
+      "step": 3284
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.365056256217772e-06,
+      "loss": 0.4508,
+      "step": 3285
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.359829976253687e-06,
+      "loss": 0.4296,
+      "step": 3286
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.354604156630136e-06,
+      "loss": 0.4795,
+      "step": 3287
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.349378798813835e-06,
+      "loss": 0.4067,
+      "step": 3288
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.344153904271363e-06,
+      "loss": 0.438,
+      "step": 3289
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.338929474469177e-06,
+      "loss": 0.4478,
+      "step": 3290
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.33370551087359e-06,
+      "loss": 0.4555,
+      "step": 3291
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.328482014950798e-06,
+      "loss": 0.4411,
+      "step": 3292
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.32325898816686e-06,
+      "loss": 0.4222,
+      "step": 3293
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.318036431987703e-06,
+      "loss": 0.4431,
+      "step": 3294
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.312814347879121e-06,
+      "loss": 0.4481,
+      "step": 3295
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.307592737306786e-06,
+      "loss": 0.4643,
+      "step": 3296
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.302371601736218e-06,
+      "loss": 0.4421,
+      "step": 3297
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.297150942632818e-06,
+      "loss": 0.4551,
+      "step": 3298
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.29193076146185e-06,
+      "loss": 0.4231,
+      "step": 3299
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.286711059688441e-06,
+      "loss": 0.4674,
+      "step": 3300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.28149183877759e-06,
+      "loss": 0.4272,
+      "step": 3301
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.276273100194154e-06,
+      "loss": 0.4478,
+      "step": 3302
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.271054845402855e-06,
+      "loss": 0.4773,
+      "step": 3303
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.265837075868283e-06,
+      "loss": 0.4381,
+      "step": 3304
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.260619793054894e-06,
+      "loss": 0.4767,
+      "step": 3305
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.255402998427e-06,
+      "loss": 0.4362,
+      "step": 3306
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.250186693448782e-06,
+      "loss": 0.4714,
+      "step": 3307
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.244970879584277e-06,
+      "loss": 0.4361,
+      "step": 3308
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.239755558297392e-06,
+      "loss": 0.4301,
+      "step": 3309
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.234540731051892e-06,
+      "loss": 0.4509,
+      "step": 3310
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.229326399311403e-06,
+      "loss": 0.4405,
+      "step": 3311
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.224112564539413e-06,
+      "loss": 0.4602,
+      "step": 3312
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.21889922819927e-06,
+      "loss": 0.464,
+      "step": 3313
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.21368639175418e-06,
+      "loss": 0.4464,
+      "step": 3314
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.208474056667212e-06,
+      "loss": 0.4818,
+      "step": 3315
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.203262224401295e-06,
+      "loss": 0.4171,
+      "step": 3316
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.198050896419214e-06,
+      "loss": 0.4621,
+      "step": 3317
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.19284007418361e-06,
+      "loss": 0.4563,
+      "step": 3318
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.187629759156994e-06,
+      "loss": 0.4325,
+      "step": 3319
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.182419952801716e-06,
+      "loss": 0.4425,
+      "step": 3320
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.177210656579996e-06,
+      "loss": 0.4053,
+      "step": 3321
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.172001871953912e-06,
+      "loss": 0.4533,
+      "step": 3322
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.166793600385391e-06,
+      "loss": 0.4385,
+      "step": 3323
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.161585843336227e-06,
+      "loss": 0.448,
+      "step": 3324
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.15637860226805e-06,
+      "loss": 0.3988,
+      "step": 3325
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.151171878642365e-06,
+      "loss": 0.3908,
+      "step": 3326
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.145965673920523e-06,
+      "loss": 0.4152,
+      "step": 3327
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.14075998956373e-06,
+      "loss": 0.4291,
+      "step": 3328
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.135554827033044e-06,
+      "loss": 0.4169,
+      "step": 3329
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.130350187789387e-06,
+      "loss": 0.4492,
+      "step": 3330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.125146073293517e-06,
+      "loss": 0.4439,
+      "step": 3331
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.119942485006058e-06,
+      "loss": 0.4429,
+      "step": 3332
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.114739424387481e-06,
+      "loss": 0.4352,
+      "step": 3333
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.10953689289811e-06,
+      "loss": 0.436,
+      "step": 3334
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.104334891998124e-06,
+      "loss": 0.4581,
+      "step": 3335
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.099133423147547e-06,
+      "loss": 0.458,
+      "step": 3336
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.093932487806254e-06,
+      "loss": 0.4513,
+      "step": 3337
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.088732087433975e-06,
+      "loss": 0.4382,
+      "step": 3338
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.083532223490288e-06,
+      "loss": 0.3819,
+      "step": 3339
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.078332897434617e-06,
+      "loss": 0.4486,
+      "step": 3340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.07313411072624e-06,
+      "loss": 0.4267,
+      "step": 3341
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.067935864824283e-06,
+      "loss": 0.452,
+      "step": 3342
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.062738161187716e-06,
+      "loss": 0.4281,
+      "step": 3343
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.05754100127536e-06,
+      "loss": 0.4447,
+      "step": 3344
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.052344386545882e-06,
+      "loss": 0.465,
+      "step": 3345
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.0471483184578e-06,
+      "loss": 0.433,
+      "step": 3346
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.041952798469473e-06,
+      "loss": 0.4692,
+      "step": 3347
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.036757828039112e-06,
+      "loss": 0.4626,
+      "step": 3348
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.031563408624767e-06,
+      "loss": 0.4259,
+      "step": 3349
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.026369541684334e-06,
+      "loss": 0.4502,
+      "step": 3350
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.021176228675563e-06,
+      "loss": 0.4452,
+      "step": 3351
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.015983471056038e-06,
+      "loss": 0.4518,
+      "step": 3352
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.010791270283197e-06,
+      "loss": 0.4352,
+      "step": 3353
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.00559962781431e-06,
+      "loss": 0.4588,
+      "step": 3354
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.000408545106492e-06,
+      "loss": 0.4438,
+      "step": 3355
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.995218023616715e-06,
+      "loss": 0.4948,
+      "step": 3356
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.990028064801781e-06,
+      "loss": 0.432,
+      "step": 3357
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.984838670118335e-06,
+      "loss": 0.4285,
+      "step": 3358
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.979649841022872e-06,
+      "loss": 0.4376,
+      "step": 3359
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.97446157897171e-06,
+      "loss": 0.4321,
+      "step": 3360
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.969273885421027e-06,
+      "loss": 0.3945,
+      "step": 3361
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.964086761826832e-06,
+      "loss": 0.4434,
+      "step": 3362
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.958900209644975e-06,
+      "loss": 0.4359,
+      "step": 3363
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.953714230331152e-06,
+      "loss": 0.4034,
+      "step": 3364
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.948528825340891e-06,
+      "loss": 0.4401,
+      "step": 3365
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.943343996129555e-06,
+      "loss": 0.4259,
+      "step": 3366
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.938159744152357e-06,
+      "loss": 0.4389,
+      "step": 3367
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.932976070864338e-06,
+      "loss": 0.4086,
+      "step": 3368
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.927792977720384e-06,
+      "loss": 0.4316,
+      "step": 3369
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.922610466175213e-06,
+      "loss": 0.4205,
+      "step": 3370
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.917428537683386e-06,
+      "loss": 0.4412,
+      "step": 3371
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.912247193699288e-06,
+      "loss": 0.443,
+      "step": 3372
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.907066435677154e-06,
+      "loss": 0.4321,
+      "step": 3373
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.901886265071046e-06,
+      "loss": 0.4128,
+      "step": 3374
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.896706683334863e-06,
+      "loss": 0.4476,
+      "step": 3375
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.891527691922346e-06,
+      "loss": 0.4417,
+      "step": 3376
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.886349292287052e-06,
+      "loss": 0.4693,
+      "step": 3377
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.881171485882392e-06,
+      "loss": 0.4304,
+      "step": 3378
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.8759942741616e-06,
+      "loss": 0.4458,
+      "step": 3379
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.870817658577743e-06,
+      "loss": 0.4159,
+      "step": 3380
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.865641640583725e-06,
+      "loss": 0.4161,
+      "step": 3381
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.860466221632282e-06,
+      "loss": 0.4145,
+      "step": 3382
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.855291403175976e-06,
+      "loss": 0.4316,
+      "step": 3383
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.850117186667206e-06,
+      "loss": 0.4341,
+      "step": 3384
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.844943573558202e-06,
+      "loss": 0.4535,
+      "step": 3385
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.839770565301019e-06,
+      "loss": 0.4604,
+      "step": 3386
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.83459816334755e-06,
+      "loss": 0.4484,
+      "step": 3387
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.829426369149518e-06,
+      "loss": 0.498,
+      "step": 3388
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.82425518415846e-06,
+      "loss": 0.4422,
+      "step": 3389
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.819084609825762e-06,
+      "loss": 0.474,
+      "step": 3390
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.81391464760263e-06,
+      "loss": 0.4473,
+      "step": 3391
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.808745298940096e-06,
+      "loss": 0.459,
+      "step": 3392
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.803576565289023e-06,
+      "loss": 0.405,
+      "step": 3393
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.798408448100106e-06,
+      "loss": 0.4761,
+      "step": 3394
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.793240948823852e-06,
+      "loss": 0.457,
+      "step": 3395
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.788074068910609e-06,
+      "loss": 0.4435,
+      "step": 3396
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.782907809810542e-06,
+      "loss": 0.4636,
+      "step": 3397
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.777742172973655e-06,
+      "loss": 0.4407,
+      "step": 3398
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.772577159849767e-06,
+      "loss": 0.4257,
+      "step": 3399
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.767412771888515e-06,
+      "loss": 0.4465,
+      "step": 3400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.762249010539372e-06,
+      "loss": 0.4883,
+      "step": 3401
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.757085877251638e-06,
+      "loss": 0.4491,
+      "step": 3402
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.751923373474425e-06,
+      "loss": 0.3983,
+      "step": 3403
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.746761500656676e-06,
+      "loss": 0.4486,
+      "step": 3404
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.741600260247155e-06,
+      "loss": 0.4098,
+      "step": 3405
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.73643965369445e-06,
+      "loss": 0.4879,
+      "step": 3406
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.731279682446964e-06,
+      "loss": 0.4209,
+      "step": 3407
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.726120347952932e-06,
+      "loss": 0.4568,
+      "step": 3408
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.720961651660406e-06,
+      "loss": 0.4212,
+      "step": 3409
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.715803595017257e-06,
+      "loss": 0.4247,
+      "step": 3410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.71064617947118e-06,
+      "loss": 0.4458,
+      "step": 3411
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.705489406469684e-06,
+      "loss": 0.4574,
+      "step": 3412
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.700333277460104e-06,
+      "loss": 0.4325,
+      "step": 3413
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.695177793889593e-06,
+      "loss": 0.392,
+      "step": 3414
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.69002295720512e-06,
+      "loss": 0.4819,
+      "step": 3415
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.684868768853472e-06,
+      "loss": 0.4743,
+      "step": 3416
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.679715230281265e-06,
+      "loss": 0.432,
+      "step": 3417
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.674562342934914e-06,
+      "loss": 0.4601,
+      "step": 3418
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.669410108260664e-06,
+      "loss": 0.4758,
+      "step": 3419
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.664258527704576e-06,
+      "loss": 0.4441,
+      "step": 3420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.659107602712524e-06,
+      "loss": 0.4276,
+      "step": 3421
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.6539573347302e-06,
+      "loss": 0.4361,
+      "step": 3422
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.648807725203112e-06,
+      "loss": 0.441,
+      "step": 3423
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.64365877557658e-06,
+      "loss": 0.4189,
+      "step": 3424
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.638510487295738e-06,
+      "loss": 0.4356,
+      "step": 3425
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.633362861805544e-06,
+      "loss": 0.4215,
+      "step": 3426
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.628215900550758e-06,
+      "loss": 0.4181,
+      "step": 3427
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.623069604975966e-06,
+      "loss": 0.4294,
+      "step": 3428
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.617923976525549e-06,
+      "loss": 0.443,
+      "step": 3429
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.61277901664372e-06,
+      "loss": 0.4298,
+      "step": 3430
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.607634726774491e-06,
+      "loss": 0.4353,
+      "step": 3431
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.602491108361695e-06,
+      "loss": 0.4528,
+      "step": 3432
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.597348162848972e-06,
+      "loss": 0.4203,
+      "step": 3433
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.592205891679777e-06,
+      "loss": 0.4689,
+      "step": 3434
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.587064296297364e-06,
+      "loss": 0.4104,
+      "step": 3435
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.5819233781448105e-06,
+      "loss": 0.4375,
+      "step": 3436
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.576783138665e-06,
+      "loss": 0.4357,
+      "step": 3437
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.571643579300622e-06,
+      "loss": 0.4485,
+      "step": 3438
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.5665047014941805e-06,
+      "loss": 0.4608,
+      "step": 3439
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.56136650668799e-06,
+      "loss": 0.4587,
+      "step": 3440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.55622899632416e-06,
+      "loss": 0.4229,
+      "step": 3441
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.55109217184462e-06,
+      "loss": 0.4576,
+      "step": 3442
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.545956034691104e-06,
+      "loss": 0.4087,
+      "step": 3443
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.540820586305153e-06,
+      "loss": 0.4538,
+      "step": 3444
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.535685828128117e-06,
+      "loss": 0.4152,
+      "step": 3445
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.530551761601147e-06,
+      "loss": 0.4219,
+      "step": 3446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.525418388165202e-06,
+      "loss": 0.4365,
+      "step": 3447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.520285709261049e-06,
+      "loss": 0.4476,
+      "step": 3448
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.5151537263292575e-06,
+      "loss": 0.4355,
+      "step": 3449
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.510022440810203e-06,
+      "loss": 0.4305,
+      "step": 3450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.504891854144066e-06,
+      "loss": 0.436,
+      "step": 3451
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.499761967770827e-06,
+      "loss": 0.4559,
+      "step": 3452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.494632783130271e-06,
+      "loss": 0.4461,
+      "step": 3453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.489504301661992e-06,
+      "loss": 0.4525,
+      "step": 3454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.48437652480538e-06,
+      "loss": 0.4435,
+      "step": 3455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.4792494539996285e-06,
+      "loss": 0.4356,
+      "step": 3456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.474123090683738e-06,
+      "loss": 0.431,
+      "step": 3457
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.468997436296501e-06,
+      "loss": 0.4583,
+      "step": 3458
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.4638724922765185e-06,
+      "loss": 0.3965,
+      "step": 3459
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.458748260062187e-06,
+      "loss": 0.4467,
+      "step": 3460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.453624741091712e-06,
+      "loss": 0.4076,
+      "step": 3461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.448501936803087e-06,
+      "loss": 0.438,
+      "step": 3462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.443379848634118e-06,
+      "loss": 0.4097,
+      "step": 3463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.438258478022393e-06,
+      "loss": 0.4785,
+      "step": 3464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.433137826405314e-06,
+      "loss": 0.4716,
+      "step": 3465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.428017895220076e-06,
+      "loss": 0.4587,
+      "step": 3466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.42289868590367e-06,
+      "loss": 0.4538,
+      "step": 3467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.4177801998928864e-06,
+      "loss": 0.4278,
+      "step": 3468
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.412662438624316e-06,
+      "loss": 0.4138,
+      "step": 3469
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.407545403534334e-06,
+      "loss": 0.4657,
+      "step": 3470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.402429096059125e-06,
+      "loss": 0.4418,
+      "step": 3471
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.397313517634659e-06,
+      "loss": 0.4446,
+      "step": 3472
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.392198669696715e-06,
+      "loss": 0.4365,
+      "step": 3473
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.387084553680859e-06,
+      "loss": 0.4334,
+      "step": 3474
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.381971171022442e-06,
+      "loss": 0.4196,
+      "step": 3475
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.376858523156624e-06,
+      "loss": 0.4396,
+      "step": 3476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.371746611518352e-06,
+      "loss": 0.4371,
+      "step": 3477
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.366635437542368e-06,
+      "loss": 0.436,
+      "step": 3478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.361525002663209e-06,
+      "loss": 0.412,
+      "step": 3479
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.356415308315201e-06,
+      "loss": 0.4762,
+      "step": 3480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.35130635593246e-06,
+      "loss": 0.4307,
+      "step": 3481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.3461981469489e-06,
+      "loss": 0.4331,
+      "step": 3482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.341090682798222e-06,
+      "loss": 0.4192,
+      "step": 3483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.335983964913919e-06,
+      "loss": 0.4222,
+      "step": 3484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.3308779947292776e-06,
+      "loss": 0.4051,
+      "step": 3485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.325772773677372e-06,
+      "loss": 0.4489,
+      "step": 3486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.320668303191062e-06,
+      "loss": 0.4374,
+      "step": 3487
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.315564584703002e-06,
+      "loss": 0.4563,
+      "step": 3488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.310461619645634e-06,
+      "loss": 0.4302,
+      "step": 3489
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.305359409451192e-06,
+      "loss": 0.4415,
+      "step": 3490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.300257955551691e-06,
+      "loss": 0.4531,
+      "step": 3491
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.29515725937894e-06,
+      "loss": 0.4458,
+      "step": 3492
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.29005732236453e-06,
+      "loss": 0.4209,
+      "step": 3493
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.284958145939842e-06,
+      "loss": 0.4142,
+      "step": 3494
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.279859731536045e-06,
+      "loss": 0.4405,
+      "step": 3495
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.274762080584091e-06,
+      "loss": 0.4735,
+      "step": 3496
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.269665194514721e-06,
+      "loss": 0.4689,
+      "step": 3497
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.26456907475846e-06,
+      "loss": 0.4187,
+      "step": 3498
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.2594737227456125e-06,
+      "loss": 0.4564,
+      "step": 3499
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.2543791399062755e-06,
+      "loss": 0.4354,
+      "step": 3500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.2492853276703275e-06,
+      "loss": 0.4134,
+      "step": 3501
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.244192287467429e-06,
+      "loss": 0.4516,
+      "step": 3502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.23910002072703e-06,
+      "loss": 0.489,
+      "step": 3503
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.2340085288783504e-06,
+      "loss": 0.4368,
+      "step": 3504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.228917813350404e-06,
+      "loss": 0.45,
+      "step": 3505
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.223827875571981e-06,
+      "loss": 0.4866,
+      "step": 3506
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.21873871697166e-06,
+      "loss": 0.4565,
+      "step": 3507
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.213650338977795e-06,
+      "loss": 0.4598,
+      "step": 3508
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.208562743018525e-06,
+      "loss": 0.4587,
+      "step": 3509
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.203475930521764e-06,
+      "loss": 0.4419,
+      "step": 3510
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.198389902915206e-06,
+      "loss": 0.4119,
+      "step": 3511
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.193304661626333e-06,
+      "loss": 0.4405,
+      "step": 3512
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.188220208082398e-06,
+      "loss": 0.4706,
+      "step": 3513
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.183136543710436e-06,
+      "loss": 0.4529,
+      "step": 3514
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.1780536699372685e-06,
+      "loss": 0.4359,
+      "step": 3515
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.172971588189475e-06,
+      "loss": 0.4417,
+      "step": 3516
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.167890299893432e-06,
+      "loss": 0.4259,
+      "step": 3517
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.162809806475283e-06,
+      "loss": 0.4407,
+      "step": 3518
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.157730109360954e-06,
+      "loss": 0.4335,
+      "step": 3519
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.1526512099761424e-06,
+      "loss": 0.4171,
+      "step": 3520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.147573109746331e-06,
+      "loss": 0.4404,
+      "step": 3521
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.142495810096762e-06,
+      "loss": 0.4391,
+      "step": 3522
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.137419312452469e-06,
+      "loss": 0.4373,
+      "step": 3523
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.132343618238251e-06,
+      "loss": 0.4633,
+      "step": 3524
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.127268728878687e-06,
+      "loss": 0.4483,
+      "step": 3525
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.122194645798128e-06,
+      "loss": 0.3956,
+      "step": 3526
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.117121370420695e-06,
+      "loss": 0.412,
+      "step": 3527
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.112048904170288e-06,
+      "loss": 0.4419,
+      "step": 3528
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.106977248470577e-06,
+      "loss": 0.4381,
+      "step": 3529
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.101906404745006e-06,
+      "loss": 0.4403,
+      "step": 3530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.096836374416789e-06,
+      "loss": 0.4569,
+      "step": 3531
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.091767158908916e-06,
+      "loss": 0.4104,
+      "step": 3532
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.0866987596441394e-06,
+      "loss": 0.4476,
+      "step": 3533
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.081631178044992e-06,
+      "loss": 0.4346,
+      "step": 3534
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.076564415533774e-06,
+      "loss": 0.4343,
+      "step": 3535
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.071498473532554e-06,
+      "loss": 0.4373,
+      "step": 3536
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.066433353463171e-06,
+      "loss": 0.452,
+      "step": 3537
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.061369056747239e-06,
+      "loss": 0.4414,
+      "step": 3538
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.056305584806127e-06,
+      "loss": 0.4686,
+      "step": 3539
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.0512429390609825e-06,
+      "loss": 0.4706,
+      "step": 3540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.046181120932726e-06,
+      "loss": 0.4547,
+      "step": 3541
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.041120131842035e-06,
+      "loss": 0.4455,
+      "step": 3542
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.03605997320936e-06,
+      "loss": 0.4433,
+      "step": 3543
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.031000646454922e-06,
+      "loss": 0.4661,
+      "step": 3544
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.0259421529986946e-06,
+      "loss": 0.4327,
+      "step": 3545
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.020884494260431e-06,
+      "loss": 0.4314,
+      "step": 3546
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.015827671659647e-06,
+      "loss": 0.4583,
+      "step": 3547
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.010771686615618e-06,
+      "loss": 0.4411,
+      "step": 3548
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.005716540547399e-06,
+      "loss": 0.4061,
+      "step": 3549
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.00066223487379e-06,
+      "loss": 0.4522,
+      "step": 3550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.995608771013365e-06,
+      "loss": 0.4227,
+      "step": 3551
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.990556150384464e-06,
+      "loss": 0.426,
+      "step": 3552
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.985504374405188e-06,
+      "loss": 0.4526,
+      "step": 3553
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.9804534444934005e-06,
+      "loss": 0.4475,
+      "step": 3554
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.975403362066727e-06,
+      "loss": 0.4351,
+      "step": 3555
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.970354128542553e-06,
+      "loss": 0.4514,
+      "step": 3556
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.965305745338033e-06,
+      "loss": 0.4737,
+      "step": 3557
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.960258213870075e-06,
+      "loss": 0.4709,
+      "step": 3558
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.955211535555353e-06,
+      "loss": 0.4494,
+      "step": 3559
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.9501657118102994e-06,
+      "loss": 0.4359,
+      "step": 3560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.945120744051108e-06,
+      "loss": 0.4241,
+      "step": 3561
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.9400766336937286e-06,
+      "loss": 0.4718,
+      "step": 3562
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.935033382153875e-06,
+      "loss": 0.4306,
+      "step": 3563
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.92999099084702e-06,
+      "loss": 0.4353,
+      "step": 3564
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.92494946118839e-06,
+      "loss": 0.449,
+      "step": 3565
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.919908794592973e-06,
+      "loss": 0.4529,
+      "step": 3566
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.914868992475521e-06,
+      "loss": 0.445,
+      "step": 3567
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.459,
+      "step": 3568
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.904791987332256e-06,
+      "loss": 0.4115,
+      "step": 3569
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.899754787134725e-06,
+      "loss": 0.4252,
+      "step": 3570
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.894718457071703e-06,
+      "loss": 0.4236,
+      "step": 3571
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.889682998556724e-06,
+      "loss": 0.4085,
+      "step": 3572
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.88464841300307e-06,
+      "loss": 0.4205,
+      "step": 3573
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.8796147018237734e-06,
+      "loss": 0.421,
+      "step": 3574
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.874581866431633e-06,
+      "loss": 0.4488,
+      "step": 3575
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.869549908239198e-06,
+      "loss": 0.4126,
+      "step": 3576
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.864518828658767e-06,
+      "loss": 0.4676,
+      "step": 3577
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.8594886291024e-06,
+      "loss": 0.4283,
+      "step": 3578
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.854459310981894e-06,
+      "loss": 0.4777,
+      "step": 3579
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.849430875708818e-06,
+      "loss": 0.4633,
+      "step": 3580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.84440332469448e-06,
+      "loss": 0.4546,
+      "step": 3581
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.839376659349945e-06,
+      "loss": 0.4471,
+      "step": 3582
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.8343508810860325e-06,
+      "loss": 0.4466,
+      "step": 3583
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.8293259913133115e-06,
+      "loss": 0.4365,
+      "step": 3584
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.82430199144209e-06,
+      "loss": 0.4517,
+      "step": 3585
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.819278882882442e-06,
+      "loss": 0.4448,
+      "step": 3586
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.8142566670441835e-06,
+      "loss": 0.4237,
+      "step": 3587
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.809235345336882e-06,
+      "loss": 0.4277,
+      "step": 3588
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.8042149191698526e-06,
+      "loss": 0.492,
+      "step": 3589
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.799195389952163e-06,
+      "loss": 0.4245,
+      "step": 3590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.794176759092622e-06,
+      "loss": 0.4467,
+      "step": 3591
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.789159027999792e-06,
+      "loss": 0.4284,
+      "step": 3592
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.784142198081983e-06,
+      "loss": 0.4248,
+      "step": 3593
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.779126270747247e-06,
+      "loss": 0.4149,
+      "step": 3594
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.77411124740339e-06,
+      "loss": 0.4378,
+      "step": 3595
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.7690971294579596e-06,
+      "loss": 0.4382,
+      "step": 3596
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.764083918318246e-06,
+      "loss": 0.4614,
+      "step": 3597
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.759071615391293e-06,
+      "loss": 0.4236,
+      "step": 3598
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.754060222083883e-06,
+      "loss": 0.4642,
+      "step": 3599
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.7490497398025444e-06,
+      "loss": 0.4343,
+      "step": 3600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.744040169953559e-06,
+      "loss": 0.462,
+      "step": 3601
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.739031513942933e-06,
+      "loss": 0.4474,
+      "step": 3602
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.734023773176434e-06,
+      "loss": 0.4472,
+      "step": 3603
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.729016949059566e-06,
+      "loss": 0.4427,
+      "step": 3604
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.724011042997576e-06,
+      "loss": 0.4836,
+      "step": 3605
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.719006056395452e-06,
+      "loss": 0.4896,
+      "step": 3606
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.7140019906579305e-06,
+      "loss": 0.4458,
+      "step": 3607
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.708998847189474e-06,
+      "loss": 0.4098,
+      "step": 3608
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.703996627394303e-06,
+      "loss": 0.4333,
+      "step": 3609
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.698995332676375e-06,
+      "loss": 0.4708,
+      "step": 3610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.693994964439379e-06,
+      "loss": 0.4483,
+      "step": 3611
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.688995524086754e-06,
+      "loss": 0.4463,
+      "step": 3612
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.683997013021679e-06,
+      "loss": 0.4482,
+      "step": 3613
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.678999432647057e-06,
+      "loss": 0.4296,
+      "step": 3614
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.674002784365547e-06,
+      "loss": 0.4126,
+      "step": 3615
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.669007069579538e-06,
+      "loss": 0.4426,
+      "step": 3616
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.664012289691162e-06,
+      "loss": 0.4256,
+      "step": 3617
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.659018446102283e-06,
+      "loss": 0.4596,
+      "step": 3618
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.65402554021451e-06,
+      "loss": 0.4325,
+      "step": 3619
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.649033573429178e-06,
+      "loss": 0.4374,
+      "step": 3620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.644042547147362e-06,
+      "loss": 0.4303,
+      "step": 3621
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.639052462769881e-06,
+      "loss": 0.4271,
+      "step": 3622
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.634063321697282e-06,
+      "loss": 0.403,
+      "step": 3623
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.629075125329849e-06,
+      "loss": 0.428,
+      "step": 3624
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.6240878750676e-06,
+      "loss": 0.4293,
+      "step": 3625
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.619101572310287e-06,
+      "loss": 0.4489,
+      "step": 3626
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.6141162184574e-06,
+      "loss": 0.4596,
+      "step": 3627
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.6091318149081595e-06,
+      "loss": 0.4322,
+      "step": 3628
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.60414836306152e-06,
+      "loss": 0.442,
+      "step": 3629
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.5991658643161696e-06,
+      "loss": 0.4025,
+      "step": 3630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.594184320070525e-06,
+      "loss": 0.4346,
+      "step": 3631
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.58920373172274e-06,
+      "loss": 0.4604,
+      "step": 3632
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.584224100670698e-06,
+      "loss": 0.4403,
+      "step": 3633
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.5792454283120136e-06,
+      "loss": 0.4435,
+      "step": 3634
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.574267716044033e-06,
+      "loss": 0.431,
+      "step": 3635
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.569290965263835e-06,
+      "loss": 0.4122,
+      "step": 3636
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.564315177368222e-06,
+      "loss": 0.4402,
+      "step": 3637
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.559340353753729e-06,
+      "loss": 0.4413,
+      "step": 3638
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.5543664958166266e-06,
+      "loss": 0.4892,
+      "step": 3639
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.549393604952906e-06,
+      "loss": 0.4502,
+      "step": 3640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.544421682558293e-06,
+      "loss": 0.4277,
+      "step": 3641
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.539450730028238e-06,
+      "loss": 0.4171,
+      "step": 3642
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.534480748757917e-06,
+      "loss": 0.4482,
+      "step": 3643
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.52951174014224e-06,
+      "loss": 0.4598,
+      "step": 3644
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.524543705575839e-06,
+      "loss": 0.4523,
+      "step": 3645
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.519576646453075e-06,
+      "loss": 0.4233,
+      "step": 3646
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.514610564168034e-06,
+      "loss": 0.4775,
+      "step": 3647
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.509645460114532e-06,
+      "loss": 0.418,
+      "step": 3648
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.504681335686098e-06,
+      "loss": 0.4211,
+      "step": 3649
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.499718192275999e-06,
+      "loss": 0.42,
+      "step": 3650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.494756031277224e-06,
+      "loss": 0.485,
+      "step": 3651
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.489794854082483e-06,
+      "loss": 0.4125,
+      "step": 3652
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.484834662084217e-06,
+      "loss": 0.4401,
+      "step": 3653
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.479875456674576e-06,
+      "loss": 0.4453,
+      "step": 3654
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.474917239245445e-06,
+      "loss": 0.4371,
+      "step": 3655
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.469960011188431e-06,
+      "loss": 0.4473,
+      "step": 3656
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.465003773894859e-06,
+      "loss": 0.4259,
+      "step": 3657
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.460048528755778e-06,
+      "loss": 0.4485,
+      "step": 3658
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.455094277161964e-06,
+      "loss": 0.4598,
+      "step": 3659
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.450141020503902e-06,
+      "loss": 0.4259,
+      "step": 3660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.445188760171806e-06,
+      "loss": 0.4785,
+      "step": 3661
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.4402374975556085e-06,
+      "loss": 0.4534,
+      "step": 3662
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.435287234044965e-06,
+      "loss": 0.45,
+      "step": 3663
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.4303379710292445e-06,
+      "loss": 0.4328,
+      "step": 3664
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.425389709897543e-06,
+      "loss": 0.4413,
+      "step": 3665
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.420442452038668e-06,
+      "loss": 0.3978,
+      "step": 3666
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.415496198841147e-06,
+      "loss": 0.424,
+      "step": 3667
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.410550951693228e-06,
+      "loss": 0.4175,
+      "step": 3668
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.405606711982877e-06,
+      "loss": 0.4138,
+      "step": 3669
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.400663481097774e-06,
+      "loss": 0.4631,
+      "step": 3670
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.395721260425321e-06,
+      "loss": 0.4291,
+      "step": 3671
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.390780051352627e-06,
+      "loss": 0.4527,
+      "step": 3672
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.385839855266527e-06,
+      "loss": 0.4537,
+      "step": 3673
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.380900673553567e-06,
+      "loss": 0.421,
+      "step": 3674
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.375962507600009e-06,
+      "loss": 0.4492,
+      "step": 3675
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.371025358791834e-06,
+      "loss": 0.5156,
+      "step": 3676
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.366089228514727e-06,
+      "loss": 0.4256,
+      "step": 3677
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.361154118154096e-06,
+      "loss": 0.4096,
+      "step": 3678
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.356220029095061e-06,
+      "loss": 0.4158,
+      "step": 3679
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.3512869627224535e-06,
+      "loss": 0.4658,
+      "step": 3680
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.346354920420822e-06,
+      "loss": 0.4171,
+      "step": 3681
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.341423903574426e-06,
+      "loss": 0.4592,
+      "step": 3682
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.33649391356723e-06,
+      "loss": 0.4661,
+      "step": 3683
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.331564951782918e-06,
+      "loss": 0.4568,
+      "step": 3684
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.326637019604888e-06,
+      "loss": 0.399,
+      "step": 3685
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.32171011841624e-06,
+      "loss": 0.4534,
+      "step": 3686
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.316784249599792e-06,
+      "loss": 0.4481,
+      "step": 3687
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.311859414538075e-06,
+      "loss": 0.4416,
+      "step": 3688
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.306935614613312e-06,
+      "loss": 0.4382,
+      "step": 3689
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.302012851207455e-06,
+      "loss": 0.4625,
+      "step": 3690
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.297091125702157e-06,
+      "loss": 0.4316,
+      "step": 3691
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.292170439478782e-06,
+      "loss": 0.4556,
+      "step": 3692
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.2872507939184e-06,
+      "loss": 0.4212,
+      "step": 3693
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.282332190401794e-06,
+      "loss": 0.4267,
+      "step": 3694
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.277414630309444e-06,
+      "loss": 0.4399,
+      "step": 3695
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.272498115021546e-06,
+      "loss": 0.4302,
+      "step": 3696
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.267582645918001e-06,
+      "loss": 0.4348,
+      "step": 3697
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.2626682243784155e-06,
+      "loss": 0.4467,
+      "step": 3698
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.257754851782104e-06,
+      "loss": 0.4496,
+      "step": 3699
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.252842529508081e-06,
+      "loss": 0.4466,
+      "step": 3700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.247931258935074e-06,
+      "loss": 0.4406,
+      "step": 3701
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.243021041441508e-06,
+      "loss": 0.456,
+      "step": 3702
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.238111878405518e-06,
+      "loss": 0.4616,
+      "step": 3703
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.23320377120494e-06,
+      "loss": 0.3853,
+      "step": 3704
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.228296721217317e-06,
+      "loss": 0.4379,
+      "step": 3705
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.223390729819889e-06,
+      "loss": 0.4125,
+      "step": 3706
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.218485798389604e-06,
+      "loss": 0.414,
+      "step": 3707
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.213581928303112e-06,
+      "loss": 0.4495,
+      "step": 3708
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.208679120936765e-06,
+      "loss": 0.4467,
+      "step": 3709
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.2037773776666134e-06,
+      "loss": 0.4587,
+      "step": 3710
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.198876699868415e-06,
+      "loss": 0.4626,
+      "step": 3711
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.193977088917622e-06,
+      "loss": 0.4559,
+      "step": 3712
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.189078546189393e-06,
+      "loss": 0.4426,
+      "step": 3713
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.184181073058582e-06,
+      "loss": 0.4434,
+      "step": 3714
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.179284670899745e-06,
+      "loss": 0.4431,
+      "step": 3715
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.174389341087138e-06,
+      "loss": 0.4622,
+      "step": 3716
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.1694950849947186e-06,
+      "loss": 0.4141,
+      "step": 3717
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.164601903996134e-06,
+      "loss": 0.4155,
+      "step": 3718
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.159709799464737e-06,
+      "loss": 0.434,
+      "step": 3719
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.154818772773579e-06,
+      "loss": 0.4769,
+      "step": 3720
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.149928825295406e-06,
+      "loss": 0.4362,
+      "step": 3721
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.14503995840266e-06,
+      "loss": 0.4318,
+      "step": 3722
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.140152173467486e-06,
+      "loss": 0.4583,
+      "step": 3723
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.1352654718617156e-06,
+      "loss": 0.4347,
+      "step": 3724
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.130379854956879e-06,
+      "loss": 0.4383,
+      "step": 3725
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.125495324124212e-06,
+      "loss": 0.4309,
+      "step": 3726
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.120611880734635e-06,
+      "loss": 0.4379,
+      "step": 3727
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.115729526158769e-06,
+      "loss": 0.4644,
+      "step": 3728
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.110848261766919e-06,
+      "loss": 0.418,
+      "step": 3729
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.105968088929098e-06,
+      "loss": 0.4124,
+      "step": 3730
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.1010890090150045e-06,
+      "loss": 0.4599,
+      "step": 3731
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0962110233940304e-06,
+      "loss": 0.4161,
+      "step": 3732
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.091334133435263e-06,
+      "loss": 0.4202,
+      "step": 3733
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.086458340507488e-06,
+      "loss": 0.4018,
+      "step": 3734
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.081583645979168e-06,
+      "loss": 0.4603,
+      "step": 3735
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.076710051218467e-06,
+      "loss": 0.4286,
+      "step": 3736
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.071837557593239e-06,
+      "loss": 0.4514,
+      "step": 3737
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.066966166471031e-06,
+      "loss": 0.4489,
+      "step": 3738
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.062095879219079e-06,
+      "loss": 0.4607,
+      "step": 3739
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.057226697204308e-06,
+      "loss": 0.4171,
+      "step": 3740
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.052358621793333e-06,
+      "loss": 0.4348,
+      "step": 3741
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.047491654352458e-06,
+      "loss": 0.4199,
+      "step": 3742
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.042625796247678e-06,
+      "loss": 0.4575,
+      "step": 3743
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.037761048844675e-06,
+      "loss": 0.4436,
+      "step": 3744
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.032897413508822e-06,
+      "loss": 0.4454,
+      "step": 3745
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.028034891605179e-06,
+      "loss": 0.4889,
+      "step": 3746
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.023173484498486e-06,
+      "loss": 0.4478,
+      "step": 3747
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.018313193553181e-06,
+      "loss": 0.413,
+      "step": 3748
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.013454020133382e-06,
+      "loss": 0.4513,
+      "step": 3749
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0085959656028994e-06,
+      "loss": 0.4412,
+      "step": 3750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.003739031325223e-06,
+      "loss": 0.4786,
+      "step": 3751
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.998883218663529e-06,
+      "loss": 0.4556,
+      "step": 3752
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.994028528980682e-06,
+      "loss": 0.4328,
+      "step": 3753
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.989174963639231e-06,
+      "loss": 0.4722,
+      "step": 3754
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.984322524001409e-06,
+      "loss": 0.4108,
+      "step": 3755
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.97947121142913e-06,
+      "loss": 0.4435,
+      "step": 3756
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.974621027284e-06,
+      "loss": 0.4526,
+      "step": 3757
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.969771972927294e-06,
+      "loss": 0.451,
+      "step": 3758
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.96492404971998e-06,
+      "loss": 0.4491,
+      "step": 3759
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.960077259022713e-06,
+      "loss": 0.4062,
+      "step": 3760
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.955231602195819e-06,
+      "loss": 0.4243,
+      "step": 3761
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9503870805993135e-06,
+      "loss": 0.4772,
+      "step": 3762
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9455436955928924e-06,
+      "loss": 0.4362,
+      "step": 3763
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9407014485359236e-06,
+      "loss": 0.4642,
+      "step": 3764
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9358603407874695e-06,
+      "loss": 0.4353,
+      "step": 3765
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.931020373706263e-06,
+      "loss": 0.4557,
+      "step": 3766
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.926181548650718e-06,
+      "loss": 0.4088,
+      "step": 3767
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.921343866978935e-06,
+      "loss": 0.4143,
+      "step": 3768
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.916507330048691e-06,
+      "loss": 0.4344,
+      "step": 3769
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9116719392174304e-06,
+      "loss": 0.4211,
+      "step": 3770
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.906837695842289e-06,
+      "loss": 0.4212,
+      "step": 3771
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.902004601280076e-06,
+      "loss": 0.4682,
+      "step": 3772
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.897172656887278e-06,
+      "loss": 0.4228,
+      "step": 3773
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.892341864020062e-06,
+      "loss": 0.4392,
+      "step": 3774
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.887512224034263e-06,
+      "loss": 0.452,
+      "step": 3775
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.882683738285404e-06,
+      "loss": 0.4441,
+      "step": 3776
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.877856408128675e-06,
+      "loss": 0.442,
+      "step": 3777
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.873030234918948e-06,
+      "loss": 0.4472,
+      "step": 3778
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.868205220010766e-06,
+      "loss": 0.4814,
+      "step": 3779
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8633813647583505e-06,
+      "loss": 0.4582,
+      "step": 3780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.858558670515591e-06,
+      "loss": 0.447,
+      "step": 3781
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.853737138636058e-06,
+      "loss": 0.4183,
+      "step": 3782
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.848916770472992e-06,
+      "loss": 0.4319,
+      "step": 3783
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.84409756737931e-06,
+      "loss": 0.4229,
+      "step": 3784
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8392795307076e-06,
+      "loss": 0.4262,
+      "step": 3785
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.834462661810123e-06,
+      "loss": 0.4874,
+      "step": 3786
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.829646962038812e-06,
+      "loss": 0.4452,
+      "step": 3787
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.824832432745267e-06,
+      "loss": 0.4797,
+      "step": 3788
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.820019075280772e-06,
+      "loss": 0.4496,
+      "step": 3789
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.815206890996267e-06,
+      "loss": 0.4279,
+      "step": 3790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.810395881242378e-06,
+      "loss": 0.4347,
+      "step": 3791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.805586047369389e-06,
+      "loss": 0.4187,
+      "step": 3792
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8007773907272565e-06,
+      "loss": 0.4372,
+      "step": 3793
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.795969912665615e-06,
+      "loss": 0.3969,
+      "step": 3794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.791163614533753e-06,
+      "loss": 0.4203,
+      "step": 3795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.786358497680647e-06,
+      "loss": 0.4339,
+      "step": 3796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7815545634549205e-06,
+      "loss": 0.4204,
+      "step": 3797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.776751813204887e-06,
+      "loss": 0.4482,
+      "step": 3798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.771950248278513e-06,
+      "loss": 0.4435,
+      "step": 3799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.76714987002343e-06,
+      "loss": 0.4183,
+      "step": 3800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7623506797869525e-06,
+      "loss": 0.4705,
+      "step": 3801
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.757552678916042e-06,
+      "loss": 0.4252,
+      "step": 3802
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.752755868757345e-06,
+      "loss": 0.4253,
+      "step": 3803
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7479602506571615e-06,
+      "loss": 0.4645,
+      "step": 3804
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.743165825961454e-06,
+      "loss": 0.4407,
+      "step": 3805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.738372596015867e-06,
+      "loss": 0.4389,
+      "step": 3806
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.733580562165688e-06,
+      "loss": 0.4458,
+      "step": 3807
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.72878972575589e-06,
+      "loss": 0.4327,
+      "step": 3808
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7240000881310945e-06,
+      "loss": 0.4501,
+      "step": 3809
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.719211650635586e-06,
+      "loss": 0.4477,
+      "step": 3810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.714424414613329e-06,
+      "loss": 0.4432,
+      "step": 3811
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.709638381407929e-06,
+      "loss": 0.4247,
+      "step": 3812
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.704853552362674e-06,
+      "loss": 0.4022,
+      "step": 3813
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.700069928820496e-06,
+      "loss": 0.4089,
+      "step": 3814
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.695287512124011e-06,
+      "loss": 0.464,
+      "step": 3815
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.690506303615461e-06,
+      "loss": 0.4569,
+      "step": 3816
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.685726304636787e-06,
+      "loss": 0.4443,
+      "step": 3817
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.680947516529566e-06,
+      "loss": 0.4363,
+      "step": 3818
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.676169940635045e-06,
+      "loss": 0.41,
+      "step": 3819
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.671393578294133e-06,
+      "loss": 0.449,
+      "step": 3820
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6666184308473915e-06,
+      "loss": 0.455,
+      "step": 3821
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.661844499635043e-06,
+      "loss": 0.4161,
+      "step": 3822
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.657071785996966e-06,
+      "loss": 0.4271,
+      "step": 3823
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.652300291272707e-06,
+      "loss": 0.4935,
+      "step": 3824
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.647530016801457e-06,
+      "loss": 0.4375,
+      "step": 3825
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6427609639220804e-06,
+      "loss": 0.4461,
+      "step": 3826
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.637993133973083e-06,
+      "loss": 0.4122,
+      "step": 3827
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.633226528292635e-06,
+      "loss": 0.4563,
+      "step": 3828
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.628461148218565e-06,
+      "loss": 0.4445,
+      "step": 3829
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.62369699508835e-06,
+      "loss": 0.4412,
+      "step": 3830
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.618934070239135e-06,
+      "loss": 0.4524,
+      "step": 3831
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6141723750077095e-06,
+      "loss": 0.4851,
+      "step": 3832
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6094119107305145e-06,
+      "loss": 0.4015,
+      "step": 3833
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.604652678743663e-06,
+      "loss": 0.4392,
+      "step": 3834
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5998946803829e-06,
+      "loss": 0.444,
+      "step": 3835
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.595137916983647e-06,
+      "loss": 0.4508,
+      "step": 3836
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5903823898809586e-06,
+      "loss": 0.4389,
+      "step": 3837
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.585628100409557e-06,
+      "loss": 0.3999,
+      "step": 3838
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.580875049903811e-06,
+      "loss": 0.4794,
+      "step": 3839
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.576123239697735e-06,
+      "loss": 0.4045,
+      "step": 3840
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.57137267112501e-06,
+      "loss": 0.4505,
+      "step": 3841
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5666233455189556e-06,
+      "loss": 0.4002,
+      "step": 3842
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.561875264212553e-06,
+      "loss": 0.4334,
+      "step": 3843
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.557128428538425e-06,
+      "loss": 0.4159,
+      "step": 3844
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.552382839828847e-06,
+      "loss": 0.4338,
+      "step": 3845
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.547638499415751e-06,
+      "loss": 0.459,
+      "step": 3846
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5428954086307085e-06,
+      "loss": 0.4299,
+      "step": 3847
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.538153568804951e-06,
+      "loss": 0.4158,
+      "step": 3848
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.53341298126935e-06,
+      "loss": 0.4511,
+      "step": 3849
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.528673647354432e-06,
+      "loss": 0.4259,
+      "step": 3850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.52393556839036e-06,
+      "loss": 0.4301,
+      "step": 3851
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.51919874570696e-06,
+      "loss": 0.4522,
+      "step": 3852
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.514463180633702e-06,
+      "loss": 0.4308,
+      "step": 3853
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.509728874499692e-06,
+      "loss": 0.4333,
+      "step": 3854
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.504995828633704e-06,
+      "loss": 0.4707,
+      "step": 3855
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.500264044364124e-06,
+      "loss": 0.4298,
+      "step": 3856
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.495533523019021e-06,
+      "loss": 0.4242,
+      "step": 3857
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.490804265926084e-06,
+      "loss": 0.4345,
+      "step": 3858
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.486076274412664e-06,
+      "loss": 0.4507,
+      "step": 3859
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.481349549805741e-06,
+      "loss": 0.4404,
+      "step": 3860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.476624093431956e-06,
+      "loss": 0.4434,
+      "step": 3861
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.471899906617581e-06,
+      "loss": 0.4121,
+      "step": 3862
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.467176990688534e-06,
+      "loss": 0.4289,
+      "step": 3863
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.462455346970383e-06,
+      "loss": 0.433,
+      "step": 3864
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.457734976788331e-06,
+      "loss": 0.4319,
+      "step": 3865
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4530158814672315e-06,
+      "loss": 0.448,
+      "step": 3866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.448298062331574e-06,
+      "loss": 0.4342,
+      "step": 3867
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.443581520705485e-06,
+      "loss": 0.4472,
+      "step": 3868
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.438866257912751e-06,
+      "loss": 0.4287,
+      "step": 3869
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.434152275276776e-06,
+      "loss": 0.4306,
+      "step": 3870
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.429439574120627e-06,
+      "loss": 0.4117,
+      "step": 3871
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.424728155766994e-06,
+      "loss": 0.4475,
+      "step": 3872
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.420018021538211e-06,
+      "loss": 0.4138,
+      "step": 3873
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.415309172756263e-06,
+      "loss": 0.4391,
+      "step": 3874
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.410601610742754e-06,
+      "loss": 0.4368,
+      "step": 3875
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4058953368189515e-06,
+      "loss": 0.4226,
+      "step": 3876
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.401190352305736e-06,
+      "loss": 0.4685,
+      "step": 3877
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.396486658523647e-06,
+      "loss": 0.4473,
+      "step": 3878
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.391784256792851e-06,
+      "loss": 0.437,
+      "step": 3879
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.38708314843315e-06,
+      "loss": 0.415,
+      "step": 3880
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.382383334763992e-06,
+      "loss": 0.4128,
+      "step": 3881
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.377684817104451e-06,
+      "loss": 0.4946,
+      "step": 3882
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.372987596773249e-06,
+      "loss": 0.4345,
+      "step": 3883
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.368291675088736e-06,
+      "loss": 0.457,
+      "step": 3884
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.363597053368897e-06,
+      "loss": 0.4408,
+      "step": 3885
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3589037329313534e-06,
+      "loss": 0.453,
+      "step": 3886
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.354211715093361e-06,
+      "loss": 0.4598,
+      "step": 3887
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.34952100117182e-06,
+      "loss": 0.4323,
+      "step": 3888
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.344831592483249e-06,
+      "loss": 0.4242,
+      "step": 3889
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.340143490343813e-06,
+      "loss": 0.4336,
+      "step": 3890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3354566960692945e-06,
+      "loss": 0.4291,
+      "step": 3891
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3307712109751274e-06,
+      "loss": 0.4126,
+      "step": 3892
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3260870363763635e-06,
+      "loss": 0.4418,
+      "step": 3893
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.321404173587696e-06,
+      "loss": 0.4345,
+      "step": 3894
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.316722623923454e-06,
+      "loss": 0.4443,
+      "step": 3895
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.312042388697582e-06,
+      "loss": 0.4251,
+      "step": 3896
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.307363469223667e-06,
+      "loss": 0.4331,
+      "step": 3897
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3026858668149205e-06,
+      "loss": 0.4175,
+      "step": 3898
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.298009582784196e-06,
+      "loss": 0.4254,
+      "step": 3899
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.293334618443962e-06,
+      "loss": 0.4211,
+      "step": 3900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.2886609751063275e-06,
+      "loss": 0.4537,
+      "step": 3901
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.283988654083029e-06,
+      "loss": 0.4242,
+      "step": 3902
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.279317656685421e-06,
+      "loss": 0.3906,
+      "step": 3903
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.274647984224506e-06,
+      "loss": 0.4283,
+      "step": 3904
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.269979638010893e-06,
+      "loss": 0.4255,
+      "step": 3905
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.265312619354843e-06,
+      "loss": 0.4389,
+      "step": 3906
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2606469295662224e-06,
+      "loss": 0.4016,
+      "step": 3907
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.25598256995453e-06,
+      "loss": 0.4627,
+      "step": 3908
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2513195418289035e-06,
+      "loss": 0.416,
+      "step": 3909
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.24665784649809e-06,
+      "loss": 0.4259,
+      "step": 3910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.241997485270478e-06,
+      "loss": 0.4351,
+      "step": 3911
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.237338459454067e-06,
+      "loss": 0.4337,
+      "step": 3912
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.232680770356495e-06,
+      "loss": 0.4478,
+      "step": 3913
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2280244192850185e-06,
+      "loss": 0.4425,
+      "step": 3914
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.223369407546509e-06,
+      "loss": 0.4473,
+      "step": 3915
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.218715736447484e-06,
+      "loss": 0.4829,
+      "step": 3916
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2140634072940624e-06,
+      "loss": 0.4156,
+      "step": 3917
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.209412421392005e-06,
+      "loss": 0.423,
+      "step": 3918
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.204762780046682e-06,
+      "loss": 0.4634,
+      "step": 3919
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2001144845630906e-06,
+      "loss": 0.4425,
+      "step": 3920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.195467536245855e-06,
+      "loss": 0.4434,
+      "step": 3921
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.19082193639921e-06,
+      "loss": 0.4177,
+      "step": 3922
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.186177686327029e-06,
+      "loss": 0.4291,
+      "step": 3923
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.181534787332791e-06,
+      "loss": 0.4606,
+      "step": 3924
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.176893240719602e-06,
+      "loss": 0.3969,
+      "step": 3925
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.172253047790187e-06,
+      "loss": 0.4138,
+      "step": 3926
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.16761420984689e-06,
+      "loss": 0.4162,
+      "step": 3927
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.162976728191685e-06,
+      "loss": 0.4464,
+      "step": 3928
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.158340604126148e-06,
+      "loss": 0.4454,
+      "step": 3929
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.153705838951495e-06,
+      "loss": 0.4015,
+      "step": 3930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.149072433968533e-06,
+      "loss": 0.4529,
+      "step": 3931
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.144440390477714e-06,
+      "loss": 0.4596,
+      "step": 3932
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.139809709779089e-06,
+      "loss": 0.4108,
+      "step": 3933
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.135180393172343e-06,
+      "loss": 0.4344,
+      "step": 3934
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1305524419567595e-06,
+      "loss": 0.4245,
+      "step": 3935
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.12592585743126e-06,
+      "loss": 0.41,
+      "step": 3936
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1213006408943645e-06,
+      "loss": 0.4678,
+      "step": 3937
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.116676793644212e-06,
+      "loss": 0.4478,
+      "step": 3938
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.11205431697857e-06,
+      "loss": 0.4218,
+      "step": 3939
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.107433212194801e-06,
+      "loss": 0.4355,
+      "step": 3940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.102813480589905e-06,
+      "loss": 0.4411,
+      "step": 3941
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.098195123460481e-06,
+      "loss": 0.4242,
+      "step": 3942
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.093578142102742e-06,
+      "loss": 0.4505,
+      "step": 3943
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.0889625378125255e-06,
+      "loss": 0.4334,
+      "step": 3944
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.08434831188527e-06,
+      "loss": 0.4675,
+      "step": 3945
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.079735465616041e-06,
+      "loss": 0.4508,
+      "step": 3946
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.075124000299506e-06,
+      "loss": 0.4567,
+      "step": 3947
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.070513917229944e-06,
+      "loss": 0.4502,
+      "step": 3948
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.065905217701257e-06,
+      "loss": 0.3799,
+      "step": 3949
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.061297903006943e-06,
+      "loss": 0.4242,
+      "step": 3950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.056691974440132e-06,
+      "loss": 0.4438,
+      "step": 3951
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.052087433293542e-06,
+      "loss": 0.4284,
+      "step": 3952
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0474842808595226e-06,
+      "loss": 0.4535,
+      "step": 3953
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.042882518430018e-06,
+      "loss": 0.4418,
+      "step": 3954
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.038282147296585e-06,
+      "loss": 0.4433,
+      "step": 3955
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.033683168750401e-06,
+      "loss": 0.4453,
+      "step": 3956
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.029085584082238e-06,
+      "loss": 0.4176,
+      "step": 3957
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.024489394582488e-06,
+      "loss": 0.4295,
+      "step": 3958
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.019894601541144e-06,
+      "loss": 0.4894,
+      "step": 3959
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.015301206247813e-06,
+      "loss": 0.4646,
+      "step": 3960
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.010709209991698e-06,
+      "loss": 0.4669,
+      "step": 3961
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0061186140616235e-06,
+      "loss": 0.4535,
+      "step": 3962
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0015294197460204e-06,
+      "loss": 0.4321,
+      "step": 3963
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.996941628332912e-06,
+      "loss": 0.4394,
+      "step": 3964
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.992355241109949e-06,
+      "loss": 0.4583,
+      "step": 3965
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.987770259364359e-06,
+      "loss": 0.4509,
+      "step": 3966
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.983186684383007e-06,
+      "loss": 0.3877,
+      "step": 3967
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.978604517452338e-06,
+      "loss": 0.3827,
+      "step": 3968
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.974023759858416e-06,
+      "loss": 0.4354,
+      "step": 3969
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.96944441288691e-06,
+      "loss": 0.427,
+      "step": 3970
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9648664778230856e-06,
+      "loss": 0.4127,
+      "step": 3971
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.960289955951813e-06,
+      "loss": 0.4374,
+      "step": 3972
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9557148485575665e-06,
+      "loss": 0.4438,
+      "step": 3973
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.951141156924432e-06,
+      "loss": 0.4572,
+      "step": 3974
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.94656888233608e-06,
+      "loss": 0.4247,
+      "step": 3975
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.941998026075806e-06,
+      "loss": 0.416,
+      "step": 3976
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.937428589426489e-06,
+      "loss": 0.4632,
+      "step": 3977
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.932860573670614e-06,
+      "loss": 0.4342,
+      "step": 3978
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9282939800902764e-06,
+      "loss": 0.4363,
+      "step": 3979
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.923728809967156e-06,
+      "loss": 0.4296,
+      "step": 3980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.919165064582553e-06,
+      "loss": 0.4325,
+      "step": 3981
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.914602745217352e-06,
+      "loss": 0.4414,
+      "step": 3982
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.910041853152038e-06,
+      "loss": 0.4372,
+      "step": 3983
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.905482389666708e-06,
+      "loss": 0.4409,
+      "step": 3984
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.900924356041044e-06,
+      "loss": 0.4433,
+      "step": 3985
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.896367753554336e-06,
+      "loss": 0.4482,
+      "step": 3986
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.891812583485467e-06,
+      "loss": 0.4383,
+      "step": 3987
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.887258847112923e-06,
+      "loss": 0.4165,
+      "step": 3988
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.882706545714783e-06,
+      "loss": 0.4243,
+      "step": 3989
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.878155680568721e-06,
+      "loss": 0.4543,
+      "step": 3990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.87360625295202e-06,
+      "loss": 0.4425,
+      "step": 3991
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.869058264141541e-06,
+      "loss": 0.4241,
+      "step": 3992
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.864511715413761e-06,
+      "loss": 0.4387,
+      "step": 3993
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8599666080447395e-06,
+      "loss": 0.4028,
+      "step": 3994
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.855422943310129e-06,
+      "loss": 0.4479,
+      "step": 3995
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.850880722485195e-06,
+      "loss": 0.4545,
+      "step": 3996
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.846339946844776e-06,
+      "loss": 0.4222,
+      "step": 3997
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.841800617663322e-06,
+      "loss": 0.4789,
+      "step": 3998
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8372627362148675e-06,
+      "loss": 0.4321,
+      "step": 3999
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.832726303773042e-06,
+      "loss": 0.4266,
+      "step": 4000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8281913216110665e-06,
+      "loss": 0.4097,
+      "step": 4001
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.823657791001766e-06,
+      "loss": 0.4569,
+      "step": 4002
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.819125713217541e-06,
+      "loss": 0.4252,
+      "step": 4003
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.814595089530396e-06,
+      "loss": 0.4612,
+      "step": 4004
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.810065921211936e-06,
+      "loss": 0.4303,
+      "step": 4005
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.805538209533328e-06,
+      "loss": 0.4585,
+      "step": 4006
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.801011955765361e-06,
+      "loss": 0.4431,
+      "step": 4007
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.796487161178394e-06,
+      "loss": 0.4408,
+      "step": 4008
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.791963827042392e-06,
+      "loss": 0.4571,
+      "step": 4009
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.787441954626895e-06,
+      "loss": 0.435,
+      "step": 4010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.782921545201049e-06,
+      "loss": 0.4716,
+      "step": 4011
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7784026000335755e-06,
+      "loss": 0.4543,
+      "step": 4012
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.773885120392788e-06,
+      "loss": 0.4447,
+      "step": 4013
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.769369107546598e-06,
+      "loss": 0.4332,
+      "step": 4014
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.764854562762491e-06,
+      "loss": 0.4549,
+      "step": 4015
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7603414873075545e-06,
+      "loss": 0.4357,
+      "step": 4016
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.755829882448455e-06,
+      "loss": 0.423,
+      "step": 4017
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.751319749451443e-06,
+      "loss": 0.4224,
+      "step": 4018
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.746811089582368e-06,
+      "loss": 0.3962,
+      "step": 4019
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.742303904106653e-06,
+      "loss": 0.4106,
+      "step": 4020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.73779819428932e-06,
+      "loss": 0.4429,
+      "step": 4021
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.733293961394967e-06,
+      "loss": 0.4902,
+      "step": 4022
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.728791206687776e-06,
+      "loss": 0.467,
+      "step": 4023
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7242899314315275e-06,
+      "loss": 0.4164,
+      "step": 4024
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.719790136889569e-06,
+      "loss": 0.4542,
+      "step": 4025
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.71529182432485e-06,
+      "loss": 0.403,
+      "step": 4026
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.710794994999888e-06,
+      "loss": 0.4358,
+      "step": 4027
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.706299650176802e-06,
+      "loss": 0.4522,
+      "step": 4028
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.701805791117269e-06,
+      "loss": 0.4088,
+      "step": 4029
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.697313419082573e-06,
+      "loss": 0.451,
+      "step": 4030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.692822535333575e-06,
+      "loss": 0.4132,
+      "step": 4031
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.688333141130705e-06,
+      "loss": 0.4412,
+      "step": 4032
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.683845237733996e-06,
+      "loss": 0.457,
+      "step": 4033
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.679358826403045e-06,
+      "loss": 0.431,
+      "step": 4034
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.674873908397039e-06,
+      "loss": 0.4441,
+      "step": 4035
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6703904849747376e-06,
+      "loss": 0.4424,
+      "step": 4036
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.665908557394492e-06,
+      "loss": 0.4337,
+      "step": 4037
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.661428126914233e-06,
+      "loss": 0.4703,
+      "step": 4038
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6569491947914555e-06,
+      "loss": 0.4479,
+      "step": 4039
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.65247176228326e-06,
+      "loss": 0.4456,
+      "step": 4040
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.647995830646296e-06,
+      "loss": 0.5017,
+      "step": 4041
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.643521401136816e-06,
+      "loss": 0.4032,
+      "step": 4042
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.639048475010636e-06,
+      "loss": 0.4391,
+      "step": 4043
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.634577053523162e-06,
+      "loss": 0.4689,
+      "step": 4044
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.630107137929365e-06,
+      "loss": 0.4273,
+      "step": 4045
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.625638729483808e-06,
+      "loss": 0.4457,
+      "step": 4046
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.621171829440617e-06,
+      "loss": 0.4647,
+      "step": 4047
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6167064390535e-06,
+      "loss": 0.4494,
+      "step": 4048
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.612242559575746e-06,
+      "loss": 0.4335,
+      "step": 4049
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6077801922602105e-06,
+      "loss": 0.4495,
+      "step": 4050
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.603319338359335e-06,
+      "loss": 0.459,
+      "step": 4051
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.598859999125132e-06,
+      "loss": 0.449,
+      "step": 4052
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.594402175809179e-06,
+      "loss": 0.489,
+      "step": 4053
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.589945869662647e-06,
+      "loss": 0.4142,
+      "step": 4054
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.585491081936263e-06,
+      "loss": 0.4366,
+      "step": 4055
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.581037813880344e-06,
+      "loss": 0.4595,
+      "step": 4056
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5765860667447685e-06,
+      "loss": 0.4137,
+      "step": 4057
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.572135841778989e-06,
+      "loss": 0.4234,
+      "step": 4058
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.56768714023204e-06,
+      "loss": 0.4251,
+      "step": 4059
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.563239963352517e-06,
+      "loss": 0.441,
+      "step": 4060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.558794312388598e-06,
+      "loss": 0.4508,
+      "step": 4061
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.554350188588021e-06,
+      "loss": 0.4518,
+      "step": 4062
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.549907593198111e-06,
+      "loss": 0.445,
+      "step": 4063
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.54546652746575e-06,
+      "loss": 0.4511,
+      "step": 4064
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5410269926373905e-06,
+      "loss": 0.3977,
+      "step": 4065
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.536588989959071e-06,
+      "loss": 0.4401,
+      "step": 4066
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5321525206763805e-06,
+      "loss": 0.4592,
+      "step": 4067
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.527717586034494e-06,
+      "loss": 0.4071,
+      "step": 4068
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.523284187278144e-06,
+      "loss": 0.4383,
+      "step": 4069
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.518852325651638e-06,
+      "loss": 0.5033,
+      "step": 4070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.514422002398846e-06,
+      "loss": 0.4573,
+      "step": 4071
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5099932187632146e-06,
+      "loss": 0.4646,
+      "step": 4072
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.505565975987757e-06,
+      "loss": 0.437,
+      "step": 4073
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.501140275315049e-06,
+      "loss": 0.4392,
+      "step": 4074
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.496716117987234e-06,
+      "loss": 0.4408,
+      "step": 4075
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.492293505246024e-06,
+      "loss": 0.4576,
+      "step": 4076
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4878724383327e-06,
+      "loss": 0.4386,
+      "step": 4077
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.483452918488104e-06,
+      "loss": 0.4628,
+      "step": 4078
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.479034946952646e-06,
+      "loss": 0.4258,
+      "step": 4079
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.474618524966313e-06,
+      "loss": 0.439,
+      "step": 4080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.47020365376863e-06,
+      "loss": 0.4353,
+      "step": 4081
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.465790334598712e-06,
+      "loss": 0.4503,
+      "step": 4082
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.461378568695225e-06,
+      "loss": 0.435,
+      "step": 4083
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.456968357296408e-06,
+      "loss": 0.4563,
+      "step": 4084
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.452559701640053e-06,
+      "loss": 0.4357,
+      "step": 4085
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.448152602963528e-06,
+      "loss": 0.4659,
+      "step": 4086
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4437470625037535e-06,
+      "loss": 0.4486,
+      "step": 4087
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.439343081497214e-06,
+      "loss": 0.4612,
+      "step": 4088
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.434940661179965e-06,
+      "loss": 0.4041,
+      "step": 4089
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.43053980278761e-06,
+      "loss": 0.4262,
+      "step": 4090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.426140507555331e-06,
+      "loss": 0.4269,
+      "step": 4091
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.421742776717857e-06,
+      "loss": 0.4871,
+      "step": 4092
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.41734661150948e-06,
+      "loss": 0.4581,
+      "step": 4093
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.412952013164062e-06,
+      "loss": 0.4301,
+      "step": 4094
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4085589829150125e-06,
+      "loss": 0.4401,
+      "step": 4095
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.404167521995315e-06,
+      "loss": 0.4588,
+      "step": 4096
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3997776316374995e-06,
+      "loss": 0.4712,
+      "step": 4097
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.395389313073659e-06,
+      "loss": 0.4159,
+      "step": 4098
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3910025675354515e-06,
+      "loss": 0.4163,
+      "step": 4099
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.386617396254085e-06,
+      "loss": 0.4337,
+      "step": 4100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3822338004603336e-06,
+      "loss": 0.4797,
+      "step": 4101
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.37785178138452e-06,
+      "loss": 0.4334,
+      "step": 4102
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.373471340256539e-06,
+      "loss": 0.4503,
+      "step": 4103
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.36909247830582e-06,
+      "loss": 0.4302,
+      "step": 4104
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.364715196761368e-06,
+      "loss": 0.4162,
+      "step": 4105
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.360339496851742e-06,
+      "loss": 0.4677,
+      "step": 4106
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.355965379805048e-06,
+      "loss": 0.446,
+      "step": 4107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.351592846848961e-06,
+      "loss": 0.4509,
+      "step": 4108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.347221899210698e-06,
+      "loss": 0.4808,
+      "step": 4109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.342852538117039e-06,
+      "loss": 0.4909,
+      "step": 4110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.338484764794312e-06,
+      "loss": 0.4291,
+      "step": 4111
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.334118580468411e-06,
+      "loss": 0.4433,
+      "step": 4112
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.32975398636477e-06,
+      "loss": 0.3994,
+      "step": 4113
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.325390983708388e-06,
+      "loss": 0.4079,
+      "step": 4114
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.32102957372382e-06,
+      "loss": 0.4606,
+      "step": 4115
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.316669757635153e-06,
+      "loss": 0.4422,
+      "step": 4116
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.31231153666605e-06,
+      "loss": 0.4542,
+      "step": 4117
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.30795491203971e-06,
+      "loss": 0.4131,
+      "step": 4118
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3035998849789e-06,
+      "loss": 0.417,
+      "step": 4119
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.299246456705921e-06,
+      "loss": 0.4232,
+      "step": 4120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2948946284426405e-06,
+      "loss": 0.447,
+      "step": 4121
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.290544401410468e-06,
+      "loss": 0.4466,
+      "step": 4122
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.286195776830362e-06,
+      "loss": 0.4679,
+      "step": 4123
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.281848755922842e-06,
+      "loss": 0.4319,
+      "step": 4124
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.277503339907961e-06,
+      "loss": 0.4105,
+      "step": 4125
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.273159530005343e-06,
+      "loss": 0.4393,
+      "step": 4126
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.268817327434141e-06,
+      "loss": 0.419,
+      "step": 4127
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2644767334130656e-06,
+      "loss": 0.4209,
+      "step": 4128
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.26013774916038e-06,
+      "loss": 0.4325,
+      "step": 4129
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.255800375893885e-06,
+      "loss": 0.4335,
+      "step": 4130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2514646148309415e-06,
+      "loss": 0.4273,
+      "step": 4131
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.247130467188448e-06,
+      "loss": 0.4356,
+      "step": 4132
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.242797934182853e-06,
+      "loss": 0.4615,
+      "step": 4133
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.238467017030156e-06,
+      "loss": 0.4425,
+      "step": 4134
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.234137716945897e-06,
+      "loss": 0.4223,
+      "step": 4135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.229810035145168e-06,
+      "loss": 0.4459,
+      "step": 4136
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.225483972842598e-06,
+      "loss": 0.4132,
+      "step": 4137
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.221159531252375e-06,
+      "loss": 0.423,
+      "step": 4138
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.21683671158822e-06,
+      "loss": 0.4428,
+      "step": 4139
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.212515515063399e-06,
+      "loss": 0.4683,
+      "step": 4140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2081959428907344e-06,
+      "loss": 0.4144,
+      "step": 4141
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.203877996282577e-06,
+      "loss": 0.4614,
+      "step": 4142
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.199561676450837e-06,
+      "loss": 0.4072,
+      "step": 4143
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.195246984606957e-06,
+      "loss": 0.4281,
+      "step": 4144
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1909339219619225e-06,
+      "loss": 0.3942,
+      "step": 4145
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.186622489726265e-06,
+      "loss": 0.4136,
+      "step": 4146
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.182312689110062e-06,
+      "loss": 0.4671,
+      "step": 4147
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1780045213229316e-06,
+      "loss": 0.4197,
+      "step": 4148
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.173697987574028e-06,
+      "loss": 0.4237,
+      "step": 4149
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.16939308907205e-06,
+      "loss": 0.4691,
+      "step": 4150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.165089827025236e-06,
+      "loss": 0.4331,
+      "step": 4151
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.160788202641373e-06,
+      "loss": 0.4336,
+      "step": 4152
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.156488217127774e-06,
+      "loss": 0.4559,
+      "step": 4153
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.152189871691306e-06,
+      "loss": 0.4417,
+      "step": 4154
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.147893167538375e-06,
+      "loss": 0.4707,
+      "step": 4155
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.143598105874908e-06,
+      "loss": 0.4617,
+      "step": 4156
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.139304687906395e-06,
+      "loss": 0.4391,
+      "step": 4157
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.135012914837846e-06,
+      "loss": 0.4303,
+      "step": 4158
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.130722787873825e-06,
+      "loss": 0.4488,
+      "step": 4159
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.126434308218421e-06,
+      "loss": 0.4475,
+      "step": 4160
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.4476,
+      "step": 4161
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.117862295647539e-06,
+      "loss": 0.4312,
+      "step": 4162
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.113578765137931e-06,
+      "loss": 0.4398,
+      "step": 4163
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.109296886748695e-06,
+      "loss": 0.4868,
+      "step": 4164
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.105016661681605e-06,
+      "loss": 0.4264,
+      "step": 4165
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.10073809113798e-06,
+      "loss": 0.4519,
+      "step": 4166
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.096461176318671e-06,
+      "loss": 0.4329,
+      "step": 4167
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.092185918424057e-06,
+      "loss": 0.4167,
+      "step": 4168
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.087912318654071e-06,
+      "loss": 0.4253,
+      "step": 4169
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.083640378208156e-06,
+      "loss": 0.436,
+      "step": 4170
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.079370098285311e-06,
+      "loss": 0.4489,
+      "step": 4171
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.075101480084058e-06,
+      "loss": 0.4391,
+      "step": 4172
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.07083452480245e-06,
+      "loss": 0.4768,
+      "step": 4173
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.066569233638083e-06,
+      "loss": 0.4319,
+      "step": 4174
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0623056077880775e-06,
+      "loss": 0.4375,
+      "step": 4175
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0580436484490925e-06,
+      "loss": 0.451,
+      "step": 4176
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.053783356817313e-06,
+      "loss": 0.4474,
+      "step": 4177
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0495247340884684e-06,
+      "loss": 0.4197,
+      "step": 4178
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.045267781457797e-06,
+      "loss": 0.4391,
+      "step": 4179
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.04101250012009e-06,
+      "loss": 0.4508,
+      "step": 4180
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.036758891269663e-06,
+      "loss": 0.4634,
+      "step": 4181
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.032506956100356e-06,
+      "loss": 0.4542,
+      "step": 4182
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0282566958055505e-06,
+      "loss": 0.4723,
+      "step": 4183
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.024008111578147e-06,
+      "loss": 0.4217,
+      "step": 4184
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0197612046105815e-06,
+      "loss": 0.4617,
+      "step": 4185
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.015515976094815e-06,
+      "loss": 0.4487,
+      "step": 4186
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.011272427222345e-06,
+      "loss": 0.4489,
+      "step": 4187
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0070305591841885e-06,
+      "loss": 0.4625,
+      "step": 4188
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.002790373170896e-06,
+      "loss": 0.4363,
+      "step": 4189
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.998551870372554e-06,
+      "loss": 0.4572,
+      "step": 4190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.994315051978753e-06,
+      "loss": 0.4542,
+      "step": 4191
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.990079919178636e-06,
+      "loss": 0.4342,
+      "step": 4192
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9858464731608545e-06,
+      "loss": 0.4327,
+      "step": 4193
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.981614715113602e-06,
+      "loss": 0.4045,
+      "step": 4194
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.977384646224584e-06,
+      "loss": 0.4705,
+      "step": 4195
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.973156267681044e-06,
+      "loss": 0.4499,
+      "step": 4196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.968929580669743e-06,
+      "loss": 0.4813,
+      "step": 4197
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9647045863769685e-06,
+      "loss": 0.4303,
+      "step": 4198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.960481285988538e-06,
+      "loss": 0.4538,
+      "step": 4199
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.956259680689784e-06,
+      "loss": 0.4587,
+      "step": 4200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.952039771665575e-06,
+      "loss": 0.4618,
+      "step": 4201
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9478215601002955e-06,
+      "loss": 0.4581,
+      "step": 4202
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.943605047177852e-06,
+      "loss": 0.4384,
+      "step": 4203
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9393902340816835e-06,
+      "loss": 0.4719,
+      "step": 4204
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.935177121994741e-06,
+      "loss": 0.4446,
+      "step": 4205
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.930965712099508e-06,
+      "loss": 0.3973,
+      "step": 4206
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.926756005577984e-06,
+      "loss": 0.4523,
+      "step": 4207
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.922548003611688e-06,
+      "loss": 0.4173,
+      "step": 4208
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.918341707381672e-06,
+      "loss": 0.4462,
+      "step": 4209
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9141371180684925e-06,
+      "loss": 0.4445,
+      "step": 4210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.909934236852246e-06,
+      "loss": 0.4312,
+      "step": 4211
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9057330649125325e-06,
+      "loss": 0.4166,
+      "step": 4212
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.901533603428489e-06,
+      "loss": 0.4373,
+      "step": 4213
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.897335853578748e-06,
+      "loss": 0.4164,
+      "step": 4214
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.893139816541487e-06,
+      "loss": 0.4386,
+      "step": 4215
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.888945493494393e-06,
+      "loss": 0.4224,
+      "step": 4216
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.884752885614665e-06,
+      "loss": 0.454,
+      "step": 4217
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.8805619940790355e-06,
+      "loss": 0.4281,
+      "step": 4218
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.876372820063741e-06,
+      "loss": 0.4241,
+      "step": 4219
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.872185364744543e-06,
+      "loss": 0.464,
+      "step": 4220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.867999629296715e-06,
+      "loss": 0.4272,
+      "step": 4221
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.863815614895056e-06,
+      "loss": 0.418,
+      "step": 4222
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.859633322713882e-06,
+      "loss": 0.4392,
+      "step": 4223
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.855452753927018e-06,
+      "loss": 0.4264,
+      "step": 4224
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.851273909707809e-06,
+      "loss": 0.4696,
+      "step": 4225
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.847096791229111e-06,
+      "loss": 0.3982,
+      "step": 4226
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.842921399663309e-06,
+      "loss": 0.4344,
+      "step": 4227
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.838747736182289e-06,
+      "loss": 0.4585,
+      "step": 4228
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.834575801957463e-06,
+      "loss": 0.408,
+      "step": 4229
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8304055981597495e-06,
+      "loss": 0.3865,
+      "step": 4230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.826237125959582e-06,
+      "loss": 0.4493,
+      "step": 4231
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.822070386526916e-06,
+      "loss": 0.4511,
+      "step": 4232
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.817905381031208e-06,
+      "loss": 0.4405,
+      "step": 4233
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.813742110641443e-06,
+      "loss": 0.4229,
+      "step": 4234
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.809580576526104e-06,
+      "loss": 0.4621,
+      "step": 4235
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8054207798531995e-06,
+      "loss": 0.407,
+      "step": 4236
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8012627217902408e-06,
+      "loss": 0.4144,
+      "step": 4237
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7971064035042515e-06,
+      "loss": 0.4219,
+      "step": 4238
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7929518261617794e-06,
+      "loss": 0.4361,
+      "step": 4239
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7887989909288648e-06,
+      "loss": 0.451,
+      "step": 4240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7846478989710776e-06,
+      "loss": 0.4412,
+      "step": 4241
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7804985514534853e-06,
+      "loss": 0.4464,
+      "step": 4242
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.776350949540666e-06,
+      "loss": 0.457,
+      "step": 4243
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7722050943967203e-06,
+      "loss": 0.4425,
+      "step": 4244
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7680609871852436e-06,
+      "loss": 0.4208,
+      "step": 4245
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.763918629069352e-06,
+      "loss": 0.4589,
+      "step": 4246
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7597780212116653e-06,
+      "loss": 0.4631,
+      "step": 4247
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7556391647743074e-06,
+      "loss": 0.478,
+      "step": 4248
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7515020609189234e-06,
+      "loss": 0.4755,
+      "step": 4249
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7473667108066524e-06,
+      "loss": 0.4718,
+      "step": 4250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.743233115598156e-06,
+      "loss": 0.4399,
+      "step": 4251
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.739101276453586e-06,
+      "loss": 0.4437,
+      "step": 4252
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7349711945326238e-06,
+      "loss": 0.4421,
+      "step": 4253
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.730842870994428e-06,
+      "loss": 0.4646,
+      "step": 4254
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.726716306997692e-06,
+      "loss": 0.4549,
+      "step": 4255
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7225915037005966e-06,
+      "loss": 0.4529,
+      "step": 4256
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.718468462260838e-06,
+      "loss": 0.4405,
+      "step": 4257
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7143471838356182e-06,
+      "loss": 0.4525,
+      "step": 4258
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7102276695816397e-06,
+      "loss": 0.4476,
+      "step": 4259
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.70610992065511e-06,
+      "loss": 0.4759,
+      "step": 4260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7019939382117397e-06,
+      "loss": 0.4314,
+      "step": 4261
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6978797234067533e-06,
+      "loss": 0.4395,
+      "step": 4262
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6937672773948663e-06,
+      "loss": 0.4323,
+      "step": 4263
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.689656601330307e-06,
+      "loss": 0.4383,
+      "step": 4264
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.68554769636681e-06,
+      "loss": 0.4169,
+      "step": 4265
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6814405636575935e-06,
+      "loss": 0.4528,
+      "step": 4266
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.677335204355401e-06,
+      "loss": 0.4464,
+      "step": 4267
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.673231619612464e-06,
+      "loss": 0.4339,
+      "step": 4268
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.669129810580525e-06,
+      "loss": 0.4291,
+      "step": 4269
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.665029778410819e-06,
+      "loss": 0.459,
+      "step": 4270
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.660931524254092e-06,
+      "loss": 0.4409,
+      "step": 4271
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.656835049260584e-06,
+      "loss": 0.4234,
+      "step": 4272
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6527403545800344e-06,
+      "loss": 0.4002,
+      "step": 4273
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6486474413616913e-06,
+      "loss": 0.4492,
+      "step": 4274
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6445563107542925e-06,
+      "loss": 0.4307,
+      "step": 4275
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6404669639060875e-06,
+      "loss": 0.417,
+      "step": 4276
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6363794019648147e-06,
+      "loss": 0.4121,
+      "step": 4277
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.632293626077711e-06,
+      "loss": 0.4349,
+      "step": 4278
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.628209637391523e-06,
+      "loss": 0.4208,
+      "step": 4279
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.624127437052484e-06,
+      "loss": 0.4198,
+      "step": 4280
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.620047026206335e-06,
+      "loss": 0.4568,
+      "step": 4281
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.615968405998308e-06,
+      "loss": 0.4398,
+      "step": 4282
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.61189157757313e-06,
+      "loss": 0.4163,
+      "step": 4283
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6078165420750366e-06,
+      "loss": 0.4107,
+      "step": 4284
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6037433006477475e-06,
+      "loss": 0.4602,
+      "step": 4285
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.59967185443449e-06,
+      "loss": 0.4634,
+      "step": 4286
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.595602204577975e-06,
+      "loss": 0.4582,
+      "step": 4287
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5915343522204284e-06,
+      "loss": 0.4493,
+      "step": 4288
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5874682985035437e-06,
+      "loss": 0.4612,
+      "step": 4289
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5834040445685325e-06,
+      "loss": 0.4329,
+      "step": 4290
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.579341591556099e-06,
+      "loss": 0.4494,
+      "step": 4291
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.575280940606429e-06,
+      "loss": 0.46,
+      "step": 4292
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.571222092859218e-06,
+      "loss": 0.4479,
+      "step": 4293
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.567165049453644e-06,
+      "loss": 0.431,
+      "step": 4294
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5631098115283833e-06,
+      "loss": 0.4712,
+      "step": 4295
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5590563802216004e-06,
+      "loss": 0.4498,
+      "step": 4296
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5550047566709646e-06,
+      "loss": 0.4396,
+      "step": 4297
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5509549420136235e-06,
+      "loss": 0.4477,
+      "step": 4298
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5469069373862296e-06,
+      "loss": 0.4312,
+      "step": 4299
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5428607439249197e-06,
+      "loss": 0.437,
+      "step": 4300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5388163627653203e-06,
+      "loss": 0.4517,
+      "step": 4301
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5347737950425587e-06,
+      "loss": 0.44,
+      "step": 4302
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5307330418912424e-06,
+      "loss": 0.4242,
+      "step": 4303
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5266941044454816e-06,
+      "loss": 0.4391,
+      "step": 4304
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5226569838388647e-06,
+      "loss": 0.4229,
+      "step": 4305
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.518621681204475e-06,
+      "loss": 0.4624,
+      "step": 4306
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.514588197674891e-06,
+      "loss": 0.4474,
+      "step": 4307
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.510556534382169e-06,
+      "loss": 0.4697,
+      "step": 4308
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.50652669245787e-06,
+      "loss": 0.4441,
+      "step": 4309
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.502498673033026e-06,
+      "loss": 0.3996,
+      "step": 4310
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.498472477238174e-06,
+      "loss": 0.4416,
+      "step": 4311
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4944481062033287e-06,
+      "loss": 0.4109,
+      "step": 4312
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.490425561057993e-06,
+      "loss": 0.4199,
+      "step": 4313
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4864048429311647e-06,
+      "loss": 0.4525,
+      "step": 4314
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.482385952951318e-06,
+      "loss": 0.4347,
+      "step": 4315
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4783688922464275e-06,
+      "loss": 0.429,
+      "step": 4316
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4743536619439446e-06,
+      "loss": 0.4719,
+      "step": 4317
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.470340263170804e-06,
+      "loss": 0.419,
+      "step": 4318
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.466328697053438e-06,
+      "loss": 0.431,
+      "step": 4319
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4623189647177533e-06,
+      "loss": 0.4558,
+      "step": 4320
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.458311067289153e-06,
+      "loss": 0.457,
+      "step": 4321
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.454305005892515e-06,
+      "loss": 0.4181,
+      "step": 4322
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.450300781652208e-06,
+      "loss": 0.4683,
+      "step": 4323
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.446298395692077e-06,
+      "loss": 0.4823,
+      "step": 4324
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.442297849135462e-06,
+      "loss": 0.4603,
+      "step": 4325
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4382991431051847e-06,
+      "loss": 0.4328,
+      "step": 4326
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.43430227872354e-06,
+      "loss": 0.4391,
+      "step": 4327
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.430307257112324e-06,
+      "loss": 0.4487,
+      "step": 4328
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4263140793927917e-06,
+      "loss": 0.454,
+      "step": 4329
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4223227466857045e-06,
+      "loss": 0.4178,
+      "step": 4330
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.418333260111286e-06,
+      "loss": 0.4356,
+      "step": 4331
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4143456207892555e-06,
+      "loss": 0.4359,
+      "step": 4332
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.410359829838814e-06,
+      "loss": 0.4786,
+      "step": 4333
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4063758883786334e-06,
+      "loss": 0.4715,
+      "step": 4334
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4023937975268728e-06,
+      "loss": 0.4947,
+      "step": 4335
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.398413558401168e-06,
+      "loss": 0.4878,
+      "step": 4336
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.394435172118644e-06,
+      "loss": 0.4288,
+      "step": 4337
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.390458639795895e-06,
+      "loss": 0.4409,
+      "step": 4338
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.386483962549004e-06,
+      "loss": 0.4346,
+      "step": 4339
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3825111414935287e-06,
+      "loss": 0.4387,
+      "step": 4340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3785401777445003e-06,
+      "loss": 0.4444,
+      "step": 4341
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3745710724164437e-06,
+      "loss": 0.4477,
+      "step": 4342
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3706038266233456e-06,
+      "loss": 0.4481,
+      "step": 4343
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3666384414786834e-06,
+      "loss": 0.464,
+      "step": 4344
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3626749180954033e-06,
+      "loss": 0.467,
+      "step": 4345
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3587132575859383e-06,
+      "loss": 0.4718,
+      "step": 4346
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.354753461062189e-06,
+      "loss": 0.4453,
+      "step": 4347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3507955296355364e-06,
+      "loss": 0.4207,
+      "step": 4348
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.346839464416842e-06,
+      "loss": 0.425,
+      "step": 4349
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.342885266516436e-06,
+      "loss": 0.4642,
+      "step": 4350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.338932937044135e-06,
+      "loss": 0.4285,
+      "step": 4351
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.33498247710922e-06,
+      "loss": 0.4545,
+      "step": 4352
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3310338878204507e-06,
+      "loss": 0.4381,
+      "step": 4353
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3270871702860686e-06,
+      "loss": 0.4506,
+      "step": 4354
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3231423256137784e-06,
+      "loss": 0.4668,
+      "step": 4355
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3191993549107725e-06,
+      "loss": 0.4344,
+      "step": 4356
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3152582592837058e-06,
+      "loss": 0.4408,
+      "step": 4357
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3113190398387076e-06,
+      "loss": 0.432,
+      "step": 4358
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.307381697681392e-06,
+      "loss": 0.426,
+      "step": 4359
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3034462339168317e-06,
+      "loss": 0.439,
+      "step": 4360
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2995126496495857e-06,
+      "loss": 0.4136,
+      "step": 4361
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.295580945983671e-06,
+      "loss": 0.4472,
+      "step": 4362
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2916511240225958e-06,
+      "loss": 0.4354,
+      "step": 4363
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2877231848693134e-06,
+      "loss": 0.47,
+      "step": 4364
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.283797129626274e-06,
+      "loss": 0.4369,
+      "step": 4365
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2798729593953903e-06,
+      "loss": 0.4427,
+      "step": 4366
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.275950675278039e-06,
+      "loss": 0.4411,
+      "step": 4367
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.272030278375079e-06,
+      "loss": 0.4469,
+      "step": 4368
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2681117697868325e-06,
+      "loss": 0.4536,
+      "step": 4369
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.264195150613091e-06,
+      "loss": 0.4256,
+      "step": 4370
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.260280421953115e-06,
+      "loss": 0.4578,
+      "step": 4371
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.256367584905643e-06,
+      "loss": 0.4293,
+      "step": 4372
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2524566405688716e-06,
+      "loss": 0.4374,
+      "step": 4373
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.248547590040477e-06,
+      "loss": 0.4327,
+      "step": 4374
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.244640434417595e-06,
+      "loss": 0.4799,
+      "step": 4375
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.24073517479683e-06,
+      "loss": 0.4319,
+      "step": 4376
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2368318122742628e-06,
+      "loss": 0.4631,
+      "step": 4377
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.232930347945429e-06,
+      "loss": 0.4231,
+      "step": 4378
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2290307829053456e-06,
+      "loss": 0.4184,
+      "step": 4379
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2251331182484868e-06,
+      "loss": 0.4542,
+      "step": 4380
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2212373550687903e-06,
+      "loss": 0.4319,
+      "step": 4381
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2173434944596747e-06,
+      "loss": 0.4312,
+      "step": 4382
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.213451537514007e-06,
+      "loss": 0.4563,
+      "step": 4383
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2095614853241376e-06,
+      "loss": 0.4362,
+      "step": 4384
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.205673338981865e-06,
+      "loss": 0.449,
+      "step": 4385
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.201787099578467e-06,
+      "loss": 0.4091,
+      "step": 4386
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.197902768204678e-06,
+      "loss": 0.4509,
+      "step": 4387
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.194020345950697e-06,
+      "loss": 0.4645,
+      "step": 4388
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.190139833906193e-06,
+      "loss": 0.4458,
+      "step": 4389
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1862612331602906e-06,
+      "loss": 0.4272,
+      "step": 4390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.182384544801589e-06,
+      "loss": 0.4513,
+      "step": 4391
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.17850976991814e-06,
+      "loss": 0.4219,
+      "step": 4392
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1746369095974594e-06,
+      "loss": 0.4353,
+      "step": 4393
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1707659649265367e-06,
+      "loss": 0.4773,
+      "step": 4394
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.166896936991808e-06,
+      "loss": 0.4513,
+      "step": 4395
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.163029826879186e-06,
+      "loss": 0.4285,
+      "step": 4396
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.159164635674035e-06,
+      "loss": 0.4718,
+      "step": 4397
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.155301364461184e-06,
+      "loss": 0.4337,
+      "step": 4398
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1514400143249203e-06,
+      "loss": 0.4778,
+      "step": 4399
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.147580586348998e-06,
+      "loss": 0.4533,
+      "step": 4400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.143723081616633e-06,
+      "loss": 0.467,
+      "step": 4401
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1398675012104885e-06,
+      "loss": 0.4248,
+      "step": 4402
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1360138462127075e-06,
+      "loss": 0.4543,
+      "step": 4403
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.132162117704869e-06,
+      "loss": 0.4144,
+      "step": 4404
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1283123167680306e-06,
+      "loss": 0.4495,
+      "step": 4405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1244644444826975e-06,
+      "loss": 0.454,
+      "step": 4406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1206185019288414e-06,
+      "loss": 0.4439,
+      "step": 4407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.116774490185891e-06,
+      "loss": 0.4164,
+      "step": 4408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1129324103327284e-06,
+      "loss": 0.4221,
+      "step": 4409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1090922634476963e-06,
+      "loss": 0.4739,
+      "step": 4410
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1052540506085903e-06,
+      "loss": 0.44,
+      "step": 4411
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1014177728926755e-06,
+      "loss": 0.4639,
+      "step": 4412
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0975834313766594e-06,
+      "loss": 0.4563,
+      "step": 4413
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0937510271367177e-06,
+      "loss": 0.4436,
+      "step": 4414
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.089920561248476e-06,
+      "loss": 0.4189,
+      "step": 4415
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0860920347870127e-06,
+      "loss": 0.4294,
+      "step": 4416
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0822654488268733e-06,
+      "loss": 0.4462,
+      "step": 4417
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0784408044420456e-06,
+      "loss": 0.4727,
+      "step": 4418
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.074618102705984e-06,
+      "loss": 0.4233,
+      "step": 4419
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0707973446915863e-06,
+      "loss": 0.4338,
+      "step": 4420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0669785314712173e-06,
+      "loss": 0.4456,
+      "step": 4421
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.063161664116686e-06,
+      "loss": 0.4378,
+      "step": 4422
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0593467436992565e-06,
+      "loss": 0.4272,
+      "step": 4423
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0555337712896527e-06,
+      "loss": 0.466,
+      "step": 4424
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0517227479580425e-06,
+      "loss": 0.4391,
+      "step": 4425
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.047913674774059e-06,
+      "loss": 0.426,
+      "step": 4426
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.044106552806777e-06,
+      "loss": 0.4233,
+      "step": 4427
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0403013831247243e-06,
+      "loss": 0.4555,
+      "step": 4428
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0364981667958903e-06,
+      "loss": 0.4368,
+      "step": 4429
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0326969048877032e-06,
+      "loss": 0.4209,
+      "step": 4430
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0288975984670564e-06,
+      "loss": 0.4147,
+      "step": 4431
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0251002486002843e-06,
+      "loss": 0.4626,
+      "step": 4432
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0213048563531713e-06,
+      "loss": 0.4124,
+      "step": 4433
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.017511422790963e-06,
+      "loss": 0.4729,
+      "step": 4434
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.013719948978342e-06,
+      "loss": 0.4275,
+      "step": 4435
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0099304359794536e-06,
+      "loss": 0.4516,
+      "step": 4436
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.006142884857881e-06,
+      "loss": 0.4363,
+      "step": 4437
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.002357296676672e-06,
+      "loss": 0.4255,
+      "step": 4438
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9985736724982995e-06,
+      "loss": 0.4323,
+      "step": 4439
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9947920133847108e-06,
+      "loss": 0.4297,
+      "step": 4440
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.991012320397283e-06,
+      "loss": 0.457,
+      "step": 4441
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9872345945968528e-06,
+      "loss": 0.4242,
+      "step": 4442
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9834588370437035e-06,
+      "loss": 0.4239,
+      "step": 4443
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9796850487975595e-06,
+      "loss": 0.4534,
+      "step": 4444
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.975913230917595e-06,
+      "loss": 0.4305,
+      "step": 4445
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9721433844624327e-06,
+      "loss": 0.4283,
+      "step": 4446
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.968375510490146e-06,
+      "loss": 0.4348,
+      "step": 4447
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9646096100582423e-06,
+      "loss": 0.4003,
+      "step": 4448
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9608456842236923e-06,
+      "loss": 0.4318,
+      "step": 4449
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9570837340428994e-06,
+      "loss": 0.473,
+      "step": 4450
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9533237605717125e-06,
+      "loss": 0.4205,
+      "step": 4451
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9495657648654364e-06,
+      "loss": 0.4145,
+      "step": 4452
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9458097479788073e-06,
+      "loss": 0.4234,
+      "step": 4453
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9420557109660197e-06,
+      "loss": 0.4268,
+      "step": 4454
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.938303654880702e-06,
+      "loss": 0.4549,
+      "step": 4455
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 5940,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1485,
+  "total_flos": 7210953804349440.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}