diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,4741 +1,10525 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.0,
-  "global_step": 786,
+  "epoch": 10.0,
+  "global_step": 1750,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0,
-      "learning_rate": 4.993638676844784e-05,
-      "loss": 3.5902,
+      "epoch": 0.01,
+      "learning_rate": 4.9971428571428576e-05,
+      "loss": 3.6087,
       "step": 1
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.9872773536895677e-05,
-      "loss": 3.2854,
+      "learning_rate": 4.994285714285715e-05,
+      "loss": 3.2923,
       "step": 2
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.9809160305343514e-05,
-      "loss": 3.2426,
+      "epoch": 0.02,
+      "learning_rate": 4.9914285714285717e-05,
+      "loss": 3.2591,
       "step": 3
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.974554707379135e-05,
-      "loss": 2.9834,
+      "learning_rate": 4.9885714285714283e-05,
+      "loss": 3.039,
       "step": 4
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.968193384223919e-05,
-      "loss": 2.9,
+      "epoch": 0.03,
+      "learning_rate": 4.985714285714286e-05,
+      "loss": 2.8404,
       "step": 5
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.9618320610687025e-05,
-      "loss": 2.8669,
+      "epoch": 0.03,
+      "learning_rate": 4.982857142857143e-05,
+      "loss": 2.7964,
       "step": 6
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.955470737913486e-05,
-      "loss": 2.6986,
+      "epoch": 0.04,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 2.6194,
       "step": 7
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.94910941475827e-05,
-      "loss": 2.6329,
+      "epoch": 0.05,
+      "learning_rate": 4.977142857142857e-05,
+      "loss": 2.5721,
       "step": 8
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.9427480916030536e-05,
-      "loss": 2.5975,
+      "epoch": 0.05,
+      "learning_rate": 4.9742857142857145e-05,
+      "loss": 2.5876,
       "step": 9
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.936386768447838e-05,
-      "loss": 2.5687,
+      "epoch": 0.06,
+      "learning_rate": 4.971428571428572e-05,
+      "loss": 2.4753,
       "step": 10
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.930025445292621e-05,
-      "loss": 2.3833,
+      "epoch": 0.06,
+      "learning_rate": 4.9685714285714285e-05,
+      "loss": 2.5941,
       "step": 11
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.923664122137405e-05,
-      "loss": 2.4913,
+      "epoch": 0.07,
+      "learning_rate": 4.965714285714286e-05,
+      "loss": 2.5473,
       "step": 12
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.9173027989821884e-05,
-      "loss": 2.5164,
+      "epoch": 0.07,
+      "learning_rate": 4.962857142857143e-05,
+      "loss": 2.5097,
       "step": 13
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.910941475826972e-05,
-      "loss": 2.3806,
+      "epoch": 0.08,
+      "learning_rate": 4.96e-05,
+      "loss": 2.3575,
       "step": 14
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.904580152671756e-05,
-      "loss": 2.3433,
+      "epoch": 0.09,
+      "learning_rate": 4.957142857142857e-05,
+      "loss": 2.3955,
       "step": 15
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.8982188295165396e-05,
-      "loss": 2.512,
+      "epoch": 0.09,
+      "learning_rate": 4.954285714285715e-05,
+      "loss": 2.2897,
       "step": 16
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.891857506361323e-05,
-      "loss": 2.41,
+      "epoch": 0.1,
+      "learning_rate": 4.951428571428572e-05,
+      "loss": 2.3772,
       "step": 17
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.885496183206107e-05,
-      "loss": 2.4081,
+      "epoch": 0.1,
+      "learning_rate": 4.9485714285714294e-05,
+      "loss": 2.4377,
       "step": 18
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.879134860050891e-05,
-      "loss": 2.371,
+      "epoch": 0.11,
+      "learning_rate": 4.9457142857142854e-05,
+      "loss": 2.2162,
       "step": 19
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.8727735368956744e-05,
-      "loss": 2.3742,
+      "epoch": 0.11,
+      "learning_rate": 4.942857142857143e-05,
+      "loss": 2.1841,
       "step": 20
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.866412213740458e-05,
-      "loss": 2.2301,
+      "epoch": 0.12,
+      "learning_rate": 4.94e-05,
+      "loss": 2.2659,
       "step": 21
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.860050890585242e-05,
-      "loss": 2.3423,
+      "epoch": 0.13,
+      "learning_rate": 4.9371428571428575e-05,
+      "loss": 2.2422,
       "step": 22
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.853689567430026e-05,
-      "loss": 2.1729,
+      "epoch": 0.13,
+      "learning_rate": 4.934285714285715e-05,
+      "loss": 2.2387,
       "step": 23
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.847328244274809e-05,
-      "loss": 2.2131,
+      "epoch": 0.14,
+      "learning_rate": 4.9314285714285716e-05,
+      "loss": 2.2341,
       "step": 24
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.840966921119593e-05,
-      "loss": 2.2021,
+      "epoch": 0.14,
+      "learning_rate": 4.928571428571429e-05,
+      "loss": 2.1916,
       "step": 25
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.8346055979643766e-05,
-      "loss": 2.3912,
+      "epoch": 0.15,
+      "learning_rate": 4.9257142857142856e-05,
+      "loss": 2.1618,
       "step": 26
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.82824427480916e-05,
-      "loss": 2.319,
+      "epoch": 0.15,
+      "learning_rate": 4.922857142857143e-05,
+      "loss": 2.2922,
       "step": 27
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.821882951653944e-05,
-      "loss": 2.201,
+      "epoch": 0.16,
+      "learning_rate": 4.92e-05,
+      "loss": 2.2532,
       "step": 28
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.8155216284987284e-05,
-      "loss": 2.1006,
+      "epoch": 0.17,
+      "learning_rate": 4.917142857142858e-05,
+      "loss": 2.1959,
       "step": 29
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.809160305343512e-05,
-      "loss": 2.1128,
+      "epoch": 0.17,
+      "learning_rate": 4.9142857142857144e-05,
+      "loss": 2.2305,
       "step": 30
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.802798982188295e-05,
-      "loss": 2.1793,
+      "epoch": 0.18,
+      "learning_rate": 4.911428571428572e-05,
+      "loss": 2.2708,
       "step": 31
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.796437659033079e-05,
-      "loss": 2.1679,
+      "epoch": 0.18,
+      "learning_rate": 4.908571428571429e-05,
+      "loss": 2.1294,
       "step": 32
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.7900763358778626e-05,
-      "loss": 2.1725,
+      "epoch": 0.19,
+      "learning_rate": 4.905714285714286e-05,
+      "loss": 2.135,
       "step": 33
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.783715012722646e-05,
-      "loss": 2.171,
+      "epoch": 0.19,
+      "learning_rate": 4.902857142857143e-05,
+      "loss": 2.1147,
       "step": 34
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.7773536895674307e-05,
-      "loss": 2.1676,
+      "epoch": 0.2,
+      "learning_rate": 4.9e-05,
+      "loss": 2.164,
       "step": 35
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.7709923664122144e-05,
-      "loss": 2.187,
+      "epoch": 0.21,
+      "learning_rate": 4.897142857142857e-05,
+      "loss": 2.1336,
       "step": 36
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.764631043256998e-05,
-      "loss": 2.1405,
+      "epoch": 0.21,
+      "learning_rate": 4.8942857142857146e-05,
+      "loss": 2.1323,
       "step": 37
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.758269720101781e-05,
-      "loss": 2.057,
+      "epoch": 0.22,
+      "learning_rate": 4.891428571428572e-05,
+      "loss": 1.9827,
       "step": 38
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.751908396946565e-05,
-      "loss": 2.0775,
+      "epoch": 0.22,
+      "learning_rate": 4.888571428571429e-05,
+      "loss": 2.1043,
       "step": 39
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.7455470737913485e-05,
-      "loss": 2.2024,
+      "epoch": 0.23,
+      "learning_rate": 4.885714285714286e-05,
+      "loss": 2.1156,
       "step": 40
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.739185750636133e-05,
-      "loss": 2.2435,
+      "epoch": 0.23,
+      "learning_rate": 4.882857142857143e-05,
+      "loss": 2.1938,
       "step": 41
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.7328244274809166e-05,
-      "loss": 2.1347,
+      "epoch": 0.24,
+      "learning_rate": 4.88e-05,
+      "loss": 2.0351,
       "step": 42
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.7264631043257e-05,
-      "loss": 2.1104,
+      "epoch": 0.25,
+      "learning_rate": 4.8771428571428574e-05,
+      "loss": 1.9892,
       "step": 43
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.720101781170484e-05,
-      "loss": 2.0484,
+      "epoch": 0.25,
+      "learning_rate": 4.874285714285715e-05,
+      "loss": 1.9887,
       "step": 44
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.713740458015267e-05,
-      "loss": 2.1963,
+      "epoch": 0.26,
+      "learning_rate": 4.8714285714285714e-05,
+      "loss": 2.0619,
       "step": 45
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.707379134860051e-05,
-      "loss": 2.1868,
+      "epoch": 0.26,
+      "learning_rate": 4.868571428571429e-05,
+      "loss": 1.9874,
       "step": 46
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.7010178117048345e-05,
-      "loss": 2.0827,
+      "epoch": 0.27,
+      "learning_rate": 4.865714285714286e-05,
+      "loss": 2.0538,
       "step": 47
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.694656488549619e-05,
-      "loss": 2.0513,
+      "epoch": 0.27,
+      "learning_rate": 4.862857142857143e-05,
+      "loss": 1.9582,
       "step": 48
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.6882951653944026e-05,
-      "loss": 2.031,
+      "epoch": 0.28,
+      "learning_rate": 4.86e-05,
+      "loss": 1.9682,
       "step": 49
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.681933842239186e-05,
-      "loss": 2.1287,
+      "epoch": 0.29,
+      "learning_rate": 4.8571428571428576e-05,
+      "loss": 2.0755,
       "step": 50
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.675572519083969e-05,
-      "loss": 2.0014,
+      "epoch": 0.29,
+      "learning_rate": 4.854285714285714e-05,
+      "loss": 1.9472,
       "step": 51
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.669211195928753e-05,
-      "loss": 2.1081,
+      "epoch": 0.3,
+      "learning_rate": 4.8514285714285716e-05,
+      "loss": 2.0844,
       "step": 52
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.662849872773537e-05,
-      "loss": 2.1631,
+      "epoch": 0.3,
+      "learning_rate": 4.848571428571429e-05,
+      "loss": 2.0636,
       "step": 53
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.656488549618321e-05,
-      "loss": 1.9476,
+      "epoch": 0.31,
+      "learning_rate": 4.8457142857142864e-05,
+      "loss": 2.0661,
       "step": 54
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.650127226463105e-05,
-      "loss": 2.1122,
+      "epoch": 0.31,
+      "learning_rate": 4.842857142857143e-05,
+      "loss": 2.1533,
       "step": 55
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.6437659033078885e-05,
-      "loss": 1.9889,
+      "epoch": 0.32,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 1.9542,
       "step": 56
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 4.637404580152672e-05,
-      "loss": 1.9505,
+      "epoch": 0.33,
+      "learning_rate": 4.837142857142857e-05,
+      "loss": 1.9903,
       "step": 57
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 4.631043256997455e-05,
-      "loss": 1.9858,
+      "epoch": 0.33,
+      "learning_rate": 4.8342857142857145e-05,
+      "loss": 2.1032,
       "step": 58
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.624681933842239e-05,
-      "loss": 2.0182,
+      "epoch": 0.34,
+      "learning_rate": 4.831428571428572e-05,
+      "loss": 1.9825,
       "step": 59
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.618320610687023e-05,
-      "loss": 2.0939,
+      "epoch": 0.34,
+      "learning_rate": 4.828571428571429e-05,
+      "loss": 2.0152,
       "step": 60
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.611959287531807e-05,
-      "loss": 2.1672,
+      "epoch": 0.35,
+      "learning_rate": 4.825714285714286e-05,
+      "loss": 1.9837,
       "step": 61
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.605597964376591e-05,
-      "loss": 2.0219,
+      "epoch": 0.35,
+      "learning_rate": 4.8228571428571426e-05,
+      "loss": 1.9918,
       "step": 62
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.5992366412213745e-05,
-      "loss": 2.0243,
+      "epoch": 0.36,
+      "learning_rate": 4.82e-05,
+      "loss": 1.9813,
       "step": 63
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.592875318066158e-05,
-      "loss": 2.0021,
+      "epoch": 0.37,
+      "learning_rate": 4.817142857142857e-05,
+      "loss": 1.8869,
       "step": 64
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.586513994910941e-05,
-      "loss": 1.8785,
+      "epoch": 0.37,
+      "learning_rate": 4.8142857142857147e-05,
+      "loss": 1.9366,
       "step": 65
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.5801526717557256e-05,
-      "loss": 1.9561,
+      "epoch": 0.38,
+      "learning_rate": 4.811428571428572e-05,
+      "loss": 1.9793,
       "step": 66
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.573791348600509e-05,
-      "loss": 2.0123,
+      "epoch": 0.38,
+      "learning_rate": 4.808571428571429e-05,
+      "loss": 1.9499,
       "step": 67
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.567430025445293e-05,
-      "loss": 1.9843,
+      "epoch": 0.39,
+      "learning_rate": 4.805714285714286e-05,
+      "loss": 1.9999,
       "step": 68
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.561068702290077e-05,
-      "loss": 1.9598,
+      "epoch": 0.39,
+      "learning_rate": 4.802857142857143e-05,
+      "loss": 1.9717,
       "step": 69
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.5547073791348604e-05,
-      "loss": 2.021,
+      "epoch": 0.4,
+      "learning_rate": 4.8e-05,
+      "loss": 2.0035,
       "step": 70
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.548346055979644e-05,
-      "loss": 2.0623,
+      "epoch": 0.41,
+      "learning_rate": 4.7971428571428575e-05,
+      "loss": 1.9975,
       "step": 71
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.541984732824428e-05,
-      "loss": 1.8997,
+      "epoch": 0.41,
+      "learning_rate": 4.794285714285714e-05,
+      "loss": 1.9979,
       "step": 72
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.5356234096692115e-05,
-      "loss": 1.9407,
+      "epoch": 0.42,
+      "learning_rate": 4.7914285714285715e-05,
+      "loss": 2.0238,
       "step": 73
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.529262086513995e-05,
-      "loss": 1.8768,
+      "epoch": 0.42,
+      "learning_rate": 4.788571428571429e-05,
+      "loss": 2.0658,
       "step": 74
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 4.522900763358779e-05,
-      "loss": 2.123,
+      "epoch": 0.43,
+      "learning_rate": 4.785714285714286e-05,
+      "loss": 1.9228,
       "step": 75
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 4.5165394402035626e-05,
-      "loss": 1.9216,
+      "epoch": 0.43,
+      "learning_rate": 4.782857142857143e-05,
+      "loss": 2.0645,
       "step": 76
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 4.5101781170483464e-05,
-      "loss": 1.9373,
+      "epoch": 0.44,
+      "learning_rate": 4.78e-05,
+      "loss": 1.9114,
       "step": 77
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 4.5038167938931294e-05,
-      "loss": 2.0328,
+      "epoch": 0.45,
+      "learning_rate": 4.777142857142857e-05,
+      "loss": 2.0026,
       "step": 78
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 4.497455470737914e-05,
-      "loss": 1.9575,
+      "epoch": 0.45,
+      "learning_rate": 4.7742857142857144e-05,
+      "loss": 1.8819,
       "step": 79
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 4.4910941475826975e-05,
-      "loss": 2.0379,
+      "epoch": 0.46,
+      "learning_rate": 4.771428571428572e-05,
+      "loss": 1.8794,
       "step": 80
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 4.484732824427481e-05,
-      "loss": 2.0398,
+      "epoch": 0.46,
+      "learning_rate": 4.768571428571429e-05,
+      "loss": 1.8938,
       "step": 81
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 4.478371501272265e-05,
-      "loss": 2.1423,
+      "epoch": 0.47,
+      "learning_rate": 4.7657142857142865e-05,
+      "loss": 2.0258,
       "step": 82
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 4.4720101781170486e-05,
-      "loss": 2.0026,
+      "epoch": 0.47,
+      "learning_rate": 4.762857142857143e-05,
+      "loss": 1.8302,
       "step": 83
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 4.465648854961832e-05,
-      "loss": 1.8533,
+      "epoch": 0.48,
+      "learning_rate": 4.76e-05,
+      "loss": 1.9461,
       "step": 84
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 4.459287531806616e-05,
-      "loss": 1.8805,
+      "epoch": 0.49,
+      "learning_rate": 4.757142857142857e-05,
+      "loss": 1.9959,
       "step": 85
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 4.4529262086514e-05,
-      "loss": 2.1085,
+      "epoch": 0.49,
+      "learning_rate": 4.7542857142857146e-05,
+      "loss": 1.932,
       "step": 86
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 4.4465648854961834e-05,
-      "loss": 2.0008,
+      "epoch": 0.5,
+      "learning_rate": 4.751428571428572e-05,
+      "loss": 1.8174,
       "step": 87
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 4.440203562340967e-05,
-      "loss": 1.8727,
+      "epoch": 0.5,
+      "learning_rate": 4.7485714285714286e-05,
+      "loss": 1.97,
       "step": 88
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 4.433842239185751e-05,
-      "loss": 1.9518,
+      "epoch": 0.51,
+      "learning_rate": 4.745714285714286e-05,
+      "loss": 1.9027,
       "step": 89
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 4.4274809160305345e-05,
-      "loss": 2.0203,
+      "epoch": 0.51,
+      "learning_rate": 4.742857142857143e-05,
+      "loss": 1.9142,
       "step": 90
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.421119592875318e-05,
-      "loss": 1.9664,
+      "epoch": 0.52,
+      "learning_rate": 4.74e-05,
+      "loss": 1.995,
       "step": 91
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.414758269720102e-05,
-      "loss": 1.9706,
+      "epoch": 0.53,
+      "learning_rate": 4.7371428571428574e-05,
+      "loss": 1.9443,
       "step": 92
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.408396946564886e-05,
-      "loss": 1.8899,
+      "epoch": 0.53,
+      "learning_rate": 4.734285714285715e-05,
+      "loss": 1.8668,
       "step": 93
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 4.4020356234096694e-05,
-      "loss": 1.9836,
+      "epoch": 0.54,
+      "learning_rate": 4.7314285714285714e-05,
+      "loss": 1.8348,
       "step": 94
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 4.395674300254453e-05,
-      "loss": 1.9058,
+      "epoch": 0.54,
+      "learning_rate": 4.728571428571429e-05,
+      "loss": 1.8173,
       "step": 95
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 4.389312977099237e-05,
-      "loss": 1.743,
+      "epoch": 0.55,
+      "learning_rate": 4.725714285714286e-05,
+      "loss": 1.8596,
       "step": 96
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 4.3829516539440205e-05,
-      "loss": 1.8766,
+      "epoch": 0.55,
+      "learning_rate": 4.7228571428571435e-05,
+      "loss": 1.7175,
       "step": 97
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 4.376590330788805e-05,
-      "loss": 1.9296,
+      "epoch": 0.56,
+      "learning_rate": 4.72e-05,
+      "loss": 1.8843,
       "step": 98
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 4.370229007633588e-05,
-      "loss": 1.9061,
+      "epoch": 0.57,
+      "learning_rate": 4.717142857142857e-05,
+      "loss": 1.8863,
       "step": 99
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 4.3638676844783716e-05,
-      "loss": 1.8887,
+      "epoch": 0.57,
+      "learning_rate": 4.714285714285714e-05,
+      "loss": 1.7893,
       "step": 100
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 4.357506361323155e-05,
-      "loss": 2.0148,
+      "epoch": 0.58,
+      "learning_rate": 4.7114285714285716e-05,
+      "loss": 1.852,
       "step": 101
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 4.351145038167939e-05,
-      "loss": 1.8899,
+      "epoch": 0.58,
+      "learning_rate": 4.708571428571429e-05,
+      "loss": 1.8383,
       "step": 102
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 4.344783715012723e-05,
-      "loss": 1.9485,
+      "epoch": 0.59,
+      "learning_rate": 4.7057142857142864e-05,
+      "loss": 1.9066,
       "step": 103
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 4.3384223918575064e-05,
-      "loss": 1.9325,
+      "epoch": 0.59,
+      "learning_rate": 4.702857142857143e-05,
+      "loss": 1.9866,
       "step": 104
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 4.332061068702291e-05,
-      "loss": 1.9845,
+      "epoch": 0.6,
+      "learning_rate": 4.7e-05,
+      "loss": 1.8563,
       "step": 105
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 4.325699745547074e-05,
-      "loss": 2.0388,
+      "epoch": 0.61,
+      "learning_rate": 4.697142857142857e-05,
+      "loss": 1.7936,
       "step": 106
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 4.3193384223918576e-05,
-      "loss": 1.8658,
+      "epoch": 0.61,
+      "learning_rate": 4.6942857142857145e-05,
+      "loss": 1.8427,
       "step": 107
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 4.312977099236641e-05,
-      "loss": 1.9221,
+      "epoch": 0.62,
+      "learning_rate": 4.691428571428572e-05,
+      "loss": 1.8277,
       "step": 108
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 4.306615776081425e-05,
-      "loss": 1.9323,
+      "epoch": 0.62,
+      "learning_rate": 4.6885714285714285e-05,
+      "loss": 1.8244,
       "step": 109
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 4.300254452926209e-05,
-      "loss": 2.0252,
+      "epoch": 0.63,
+      "learning_rate": 4.685714285714286e-05,
+      "loss": 1.9399,
       "step": 110
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 4.293893129770993e-05,
-      "loss": 2.0264,
+      "epoch": 0.63,
+      "learning_rate": 4.682857142857143e-05,
+      "loss": 1.9104,
       "step": 111
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 4.287531806615776e-05,
-      "loss": 1.8449,
+      "epoch": 0.64,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 1.7908,
       "step": 112
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 4.28117048346056e-05,
-      "loss": 2.0663,
+      "epoch": 0.65,
+      "learning_rate": 4.677142857142857e-05,
+      "loss": 1.8952,
       "step": 113
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 4.2748091603053435e-05,
-      "loss": 1.9847,
+      "epoch": 0.65,
+      "learning_rate": 4.6742857142857146e-05,
+      "loss": 1.8587,
       "step": 114
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 4.268447837150127e-05,
-      "loss": 1.8867,
+      "epoch": 0.66,
+      "learning_rate": 4.671428571428571e-05,
+      "loss": 1.8822,
       "step": 115
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 4.262086513994911e-05,
-      "loss": 1.8354,
+      "epoch": 0.66,
+      "learning_rate": 4.668571428571429e-05,
+      "loss": 1.9028,
       "step": 116
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 4.255725190839695e-05,
-      "loss": 1.9582,
+      "epoch": 0.67,
+      "learning_rate": 4.665714285714286e-05,
+      "loss": 1.872,
       "step": 117
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 4.249363867684479e-05,
-      "loss": 1.8275,
+      "epoch": 0.67,
+      "learning_rate": 4.6628571428571434e-05,
+      "loss": 1.8945,
       "step": 118
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 4.243002544529262e-05,
-      "loss": 1.8622,
+      "epoch": 0.68,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 1.8815,
       "step": 119
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.236641221374046e-05,
-      "loss": 1.8426,
+      "epoch": 0.69,
+      "learning_rate": 4.6571428571428575e-05,
+      "loss": 1.7866,
       "step": 120
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.2302798982188295e-05,
-      "loss": 1.8311,
+      "epoch": 0.69,
+      "learning_rate": 4.654285714285714e-05,
+      "loss": 1.8106,
       "step": 121
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.223918575063613e-05,
-      "loss": 1.911,
+      "epoch": 0.7,
+      "learning_rate": 4.6514285714285715e-05,
+      "loss": 1.9063,
       "step": 122
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.2175572519083975e-05,
-      "loss": 1.9679,
+      "epoch": 0.7,
+      "learning_rate": 4.648571428571429e-05,
+      "loss": 1.8769,
       "step": 123
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.211195928753181e-05,
-      "loss": 1.7676,
+      "epoch": 0.71,
+      "learning_rate": 4.645714285714286e-05,
+      "loss": 1.8105,
       "step": 124
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.204834605597965e-05,
-      "loss": 1.8708,
+      "epoch": 0.71,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 1.7734,
       "step": 125
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.198473282442748e-05,
-      "loss": 1.8456,
+      "epoch": 0.72,
+      "learning_rate": 4.64e-05,
+      "loss": 1.7694,
       "step": 126
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.192111959287532e-05,
-      "loss": 1.9192,
+      "epoch": 0.73,
+      "learning_rate": 4.637142857142857e-05,
+      "loss": 1.8728,
       "step": 127
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.1857506361323154e-05,
-      "loss": 1.9453,
+      "epoch": 0.73,
+      "learning_rate": 4.6342857142857143e-05,
+      "loss": 1.7541,
       "step": 128
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.1793893129771e-05,
-      "loss": 1.8886,
+      "epoch": 0.74,
+      "learning_rate": 4.631428571428572e-05,
+      "loss": 1.8828,
       "step": 129
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.1730279898218835e-05,
-      "loss": 1.8233,
+      "epoch": 0.74,
+      "learning_rate": 4.628571428571429e-05,
+      "loss": 1.8825,
       "step": 130
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.166666666666667e-05,
-      "loss": 1.8474,
+      "epoch": 0.75,
+      "learning_rate": 4.625714285714286e-05,
+      "loss": 1.869,
       "step": 131
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.160305343511451e-05,
-      "loss": 1.9059,
+      "epoch": 0.75,
+      "learning_rate": 4.622857142857143e-05,
+      "loss": 1.8532,
       "step": 132
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.153944020356234e-05,
-      "loss": 1.7812,
+      "epoch": 0.76,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 1.7864,
       "step": 133
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.1475826972010176e-05,
-      "loss": 1.9022,
+      "epoch": 0.77,
+      "learning_rate": 4.617142857142857e-05,
+      "loss": 1.9118,
       "step": 134
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.1412213740458014e-05,
-      "loss": 1.9114,
+      "epoch": 0.77,
+      "learning_rate": 4.6142857142857145e-05,
+      "loss": 1.8429,
       "step": 135
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.134860050890586e-05,
-      "loss": 1.9314,
+      "epoch": 0.78,
+      "learning_rate": 4.611428571428571e-05,
+      "loss": 1.8928,
       "step": 136
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 4.1284987277353694e-05,
-      "loss": 1.9968,
+      "epoch": 0.78,
+      "learning_rate": 4.6085714285714286e-05,
+      "loss": 1.756,
       "step": 137
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.122137404580153e-05,
-      "loss": 1.8303,
+      "epoch": 0.79,
+      "learning_rate": 4.605714285714286e-05,
+      "loss": 1.7718,
       "step": 138
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.115776081424936e-05,
-      "loss": 1.8287,
+      "epoch": 0.79,
+      "learning_rate": 4.602857142857143e-05,
+      "loss": 1.709,
       "step": 139
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.10941475826972e-05,
-      "loss": 1.7839,
+      "epoch": 0.8,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.7784,
       "step": 140
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.1030534351145036e-05,
-      "loss": 1.8441,
+      "epoch": 0.81,
+      "learning_rate": 4.5971428571428574e-05,
+      "loss": 1.7954,
       "step": 141
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.096692111959288e-05,
-      "loss": 1.8067,
+      "epoch": 0.81,
+      "learning_rate": 4.594285714285714e-05,
+      "loss": 1.694,
       "step": 142
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.090330788804072e-05,
-      "loss": 1.7333,
+      "epoch": 0.82,
+      "learning_rate": 4.5914285714285714e-05,
+      "loss": 1.7413,
       "step": 143
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.0839694656488554e-05,
-      "loss": 1.8521,
+      "epoch": 0.82,
+      "learning_rate": 4.588571428571429e-05,
+      "loss": 1.7266,
       "step": 144
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.077608142493639e-05,
-      "loss": 1.7242,
+      "epoch": 0.83,
+      "learning_rate": 4.585714285714286e-05,
+      "loss": 1.862,
       "step": 145
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.071246819338422e-05,
-      "loss": 1.7233,
+      "epoch": 0.83,
+      "learning_rate": 4.5828571428571435e-05,
+      "loss": 1.767,
       "step": 146
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.064885496183206e-05,
-      "loss": 1.844,
+      "epoch": 0.84,
+      "learning_rate": 4.58e-05,
+      "loss": 1.7439,
       "step": 147
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.05852417302799e-05,
-      "loss": 1.8458,
+      "epoch": 0.85,
+      "learning_rate": 4.5771428571428576e-05,
+      "loss": 1.685,
       "step": 148
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 4.052162849872774e-05,
-      "loss": 1.7438,
+      "epoch": 0.85,
+      "learning_rate": 4.574285714285714e-05,
+      "loss": 1.8017,
       "step": 149
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 4.0458015267175576e-05,
-      "loss": 1.8411,
+      "epoch": 0.86,
+      "learning_rate": 4.5714285714285716e-05,
+      "loss": 1.7417,
       "step": 150
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 4.0394402035623413e-05,
-      "loss": 1.8313,
+      "epoch": 0.86,
+      "learning_rate": 4.568571428571429e-05,
+      "loss": 1.7689,
       "step": 151
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 4.033078880407125e-05,
-      "loss": 1.7893,
+      "epoch": 0.87,
+      "learning_rate": 4.5657142857142857e-05,
+      "loss": 1.7246,
       "step": 152
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 4.026717557251908e-05,
-      "loss": 1.7745,
+      "epoch": 0.87,
+      "learning_rate": 4.562857142857143e-05,
+      "loss": 1.7656,
       "step": 153
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 4.0203562340966925e-05,
-      "loss": 1.8041,
+      "epoch": 0.88,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 1.8091,
       "step": 154
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 4.013994910941476e-05,
-      "loss": 2.0707,
+      "epoch": 0.89,
+      "learning_rate": 4.557142857142858e-05,
+      "loss": 1.8071,
       "step": 155
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 4.00763358778626e-05,
-      "loss": 1.8498,
+      "epoch": 0.89,
+      "learning_rate": 4.5542857142857144e-05,
+      "loss": 1.8548,
       "step": 156
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 4.0012722646310436e-05,
-      "loss": 1.835,
+      "epoch": 0.9,
+      "learning_rate": 4.551428571428572e-05,
+      "loss": 1.7864,
       "step": 157
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 3.994910941475827e-05,
-      "loss": 1.7922,
+      "epoch": 0.9,
+      "learning_rate": 4.5485714285714285e-05,
+      "loss": 1.7857,
       "step": 158
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.988549618320611e-05,
-      "loss": 1.7463,
+      "epoch": 0.91,
+      "learning_rate": 4.545714285714286e-05,
+      "loss": 1.8296,
       "step": 159
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.982188295165395e-05,
-      "loss": 1.7811,
+      "epoch": 0.91,
+      "learning_rate": 4.542857142857143e-05,
+      "loss": 1.7885,
       "step": 160
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 3.9758269720101784e-05,
-      "loss": 1.7887,
+      "epoch": 0.92,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 1.8123,
       "step": 161
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.969465648854962e-05,
-      "loss": 1.8297,
+      "epoch": 0.93,
+      "learning_rate": 4.537142857142857e-05,
+      "loss": 1.782,
       "step": 162
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 3.963104325699746e-05,
-      "loss": 1.7985,
+      "epoch": 0.93,
+      "learning_rate": 4.534285714285714e-05,
+      "loss": 1.753,
       "step": 163
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.9567430025445295e-05,
-      "loss": 1.8592,
+      "epoch": 0.94,
+      "learning_rate": 4.531428571428571e-05,
+      "loss": 1.7099,
       "step": 164
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.950381679389313e-05,
-      "loss": 1.8001,
+      "epoch": 0.94,
+      "learning_rate": 4.528571428571429e-05,
+      "loss": 1.7098,
       "step": 165
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 3.944020356234097e-05,
-      "loss": 1.9156,
+      "epoch": 0.95,
+      "learning_rate": 4.525714285714286e-05,
+      "loss": 1.6973,
       "step": 166
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.9376590330788807e-05,
-      "loss": 1.7389,
+      "epoch": 0.95,
+      "learning_rate": 4.5228571428571434e-05,
+      "loss": 1.7178,
       "step": 167
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 3.9312977099236644e-05,
-      "loss": 1.8288,
+      "epoch": 0.96,
+      "learning_rate": 4.52e-05,
+      "loss": 1.7575,
       "step": 168
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.924936386768448e-05,
-      "loss": 1.9269,
+      "epoch": 0.97,
+      "learning_rate": 4.5171428571428575e-05,
+      "loss": 1.7206,
       "step": 169
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.918575063613232e-05,
-      "loss": 1.8447,
+      "epoch": 0.97,
+      "learning_rate": 4.514285714285714e-05,
+      "loss": 1.7937,
       "step": 170
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 3.9122137404580155e-05,
-      "loss": 1.7465,
+      "epoch": 0.98,
+      "learning_rate": 4.5114285714285715e-05,
+      "loss": 1.7444,
       "step": 171
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.905852417302799e-05,
-      "loss": 1.8731,
+      "epoch": 0.98,
+      "learning_rate": 4.508571428571429e-05,
+      "loss": 1.824,
       "step": 172
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.899491094147583e-05,
-      "loss": 1.8109,
+      "epoch": 0.99,
+      "learning_rate": 4.5057142857142856e-05,
+      "loss": 1.7378,
       "step": 173
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 3.8931297709923666e-05,
-      "loss": 1.8166,
+      "epoch": 0.99,
+      "learning_rate": 4.502857142857143e-05,
+      "loss": 1.7011,
       "step": 174
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.88676844783715e-05,
-      "loss": 1.7502,
+      "epoch": 1.0,
+      "learning_rate": 4.5e-05,
+      "loss": 1.8994,
       "step": 175
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 3.880407124681934e-05,
-      "loss": 1.9094,
+      "epoch": 1.01,
+      "learning_rate": 4.4971428571428576e-05,
+      "loss": 1.7652,
       "step": 176
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 3.874045801526718e-05,
-      "loss": 1.8702,
+      "epoch": 1.01,
+      "learning_rate": 4.494285714285715e-05,
+      "loss": 1.7149,
       "step": 177
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 3.8676844783715014e-05,
-      "loss": 1.8245,
+      "epoch": 1.02,
+      "learning_rate": 4.491428571428572e-05,
+      "loss": 1.8049,
       "step": 178
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 3.861323155216285e-05,
-      "loss": 1.838,
+      "epoch": 1.02,
+      "learning_rate": 4.4885714285714284e-05,
+      "loss": 1.7698,
       "step": 179
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 3.854961832061069e-05,
-      "loss": 1.7528,
+      "epoch": 1.03,
+      "learning_rate": 4.485714285714286e-05,
+      "loss": 1.6461,
       "step": 180
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 3.8486005089058526e-05,
-      "loss": 1.7561,
+      "epoch": 1.03,
+      "learning_rate": 4.482857142857143e-05,
+      "loss": 1.7399,
       "step": 181
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 3.842239185750636e-05,
-      "loss": 1.8,
+      "epoch": 1.04,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 1.7278,
       "step": 182
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 3.83587786259542e-05,
-      "loss": 1.9431,
+      "epoch": 1.05,
+      "learning_rate": 4.477142857142858e-05,
+      "loss": 1.7141,
       "step": 183
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 3.829516539440204e-05,
-      "loss": 1.9274,
+      "epoch": 1.05,
+      "learning_rate": 4.4742857142857145e-05,
+      "loss": 1.7329,
       "step": 184
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 3.8231552162849874e-05,
-      "loss": 1.786,
+      "epoch": 1.06,
+      "learning_rate": 4.471428571428571e-05,
+      "loss": 1.6483,
       "step": 185
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 3.816793893129771e-05,
-      "loss": 1.7577,
+      "epoch": 1.06,
+      "learning_rate": 4.4685714285714286e-05,
+      "loss": 1.6479,
       "step": 186
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 3.810432569974555e-05,
-      "loss": 1.744,
+      "epoch": 1.07,
+      "learning_rate": 4.465714285714286e-05,
+      "loss": 1.6389,
       "step": 187
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.8040712468193385e-05,
-      "loss": 1.7136,
+      "epoch": 1.07,
+      "learning_rate": 4.462857142857143e-05,
+      "loss": 1.7568,
       "step": 188
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.797709923664122e-05,
-      "loss": 1.7748,
+      "epoch": 1.08,
+      "learning_rate": 4.46e-05,
+      "loss": 1.7051,
       "step": 189
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.791348600508906e-05,
-      "loss": 1.8062,
+      "epoch": 1.09,
+      "learning_rate": 4.4571428571428574e-05,
+      "loss": 1.6474,
       "step": 190
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.7849872773536896e-05,
-      "loss": 1.9949,
+      "epoch": 1.09,
+      "learning_rate": 4.454285714285715e-05,
+      "loss": 1.6429,
       "step": 191
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 3.778625954198473e-05,
-      "loss": 1.566,
+      "epoch": 1.1,
+      "learning_rate": 4.4514285714285714e-05,
+      "loss": 1.7265,
       "step": 192
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.772264631043258e-05,
-      "loss": 1.9316,
+      "epoch": 1.1,
+      "learning_rate": 4.448571428571429e-05,
+      "loss": 1.75,
       "step": 193
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.765903307888041e-05,
-      "loss": 1.7205,
+      "epoch": 1.11,
+      "learning_rate": 4.445714285714286e-05,
+      "loss": 1.6394,
       "step": 194
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 3.7595419847328244e-05,
-      "loss": 1.9365,
+      "epoch": 1.11,
+      "learning_rate": 4.442857142857143e-05,
+      "loss": 1.7156,
       "step": 195
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.753180661577608e-05,
-      "loss": 1.8079,
+      "epoch": 1.12,
+      "learning_rate": 4.44e-05,
+      "loss": 1.7182,
       "step": 196
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 3.746819338422392e-05,
-      "loss": 1.9424,
+      "epoch": 1.13,
+      "learning_rate": 4.4371428571428575e-05,
+      "loss": 1.6993,
       "step": 197
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.7404580152671756e-05,
-      "loss": 1.725,
+      "epoch": 1.13,
+      "learning_rate": 4.434285714285715e-05,
+      "loss": 1.6851,
       "step": 198
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.73409669211196e-05,
-      "loss": 1.6289,
+      "epoch": 1.14,
+      "learning_rate": 4.4314285714285716e-05,
+      "loss": 1.6644,
       "step": 199
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 3.727735368956743e-05,
-      "loss": 1.894,
+      "epoch": 1.14,
+      "learning_rate": 4.428571428571428e-05,
+      "loss": 1.6259,
       "step": 200
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.721374045801527e-05,
-      "loss": 1.8859,
+      "epoch": 1.15,
+      "learning_rate": 4.4257142857142856e-05,
+      "loss": 1.7085,
       "step": 201
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.7150127226463104e-05,
-      "loss": 1.7943,
+      "epoch": 1.15,
+      "learning_rate": 4.422857142857143e-05,
+      "loss": 1.634,
       "step": 202
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 3.708651399491094e-05,
-      "loss": 1.7328,
+      "epoch": 1.16,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 1.6908,
       "step": 203
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 3.702290076335878e-05,
-      "loss": 1.9442,
+      "epoch": 1.17,
+      "learning_rate": 4.417142857142858e-05,
+      "loss": 1.7158,
       "step": 204
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 3.695928753180662e-05,
-      "loss": 1.722,
+      "epoch": 1.17,
+      "learning_rate": 4.4142857142857144e-05,
+      "loss": 1.7171,
       "step": 205
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 3.689567430025446e-05,
-      "loss": 1.7147,
+      "epoch": 1.18,
+      "learning_rate": 4.411428571428572e-05,
+      "loss": 1.6457,
       "step": 206
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 3.683206106870229e-05,
-      "loss": 1.7762,
+      "epoch": 1.18,
+      "learning_rate": 4.4085714285714285e-05,
+      "loss": 1.6225,
       "step": 207
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 3.6768447837150126e-05,
-      "loss": 1.712,
+      "epoch": 1.19,
+      "learning_rate": 4.405714285714286e-05,
+      "loss": 1.7145,
       "step": 208
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 3.6704834605597963e-05,
-      "loss": 1.5977,
+      "epoch": 1.19,
+      "learning_rate": 4.402857142857143e-05,
+      "loss": 1.6146,
       "step": 209
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 3.66412213740458e-05,
-      "loss": 1.8007,
+      "epoch": 1.2,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.6774,
       "step": 210
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 3.6577608142493644e-05,
-      "loss": 1.8174,
+      "epoch": 1.21,
+      "learning_rate": 4.397142857142857e-05,
+      "loss": 1.6598,
       "step": 211
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 3.651399491094148e-05,
-      "loss": 1.6079,
+      "epoch": 1.21,
+      "learning_rate": 4.3942857142857146e-05,
+      "loss": 1.6714,
       "step": 212
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 3.645038167938932e-05,
-      "loss": 1.7263,
+      "epoch": 1.22,
+      "learning_rate": 4.391428571428572e-05,
+      "loss": 1.7704,
       "step": 213
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 3.638676844783715e-05,
-      "loss": 1.6333,
+      "epoch": 1.22,
+      "learning_rate": 4.388571428571429e-05,
+      "loss": 1.6439,
       "step": 214
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 3.6323155216284986e-05,
-      "loss": 1.7735,
+      "epoch": 1.23,
+      "learning_rate": 4.385714285714286e-05,
+      "loss": 1.6367,
       "step": 215
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 3.625954198473282e-05,
-      "loss": 1.6724,
+      "epoch": 1.23,
+      "learning_rate": 4.382857142857143e-05,
+      "loss": 1.6505,
       "step": 216
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 3.619592875318067e-05,
-      "loss": 1.8875,
+      "epoch": 1.24,
+      "learning_rate": 4.38e-05,
+      "loss": 1.6664,
       "step": 217
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 3.6132315521628504e-05,
-      "loss": 1.7282,
+      "epoch": 1.25,
+      "learning_rate": 4.3771428571428574e-05,
+      "loss": 1.5998,
       "step": 218
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 3.606870229007634e-05,
-      "loss": 1.713,
+      "epoch": 1.25,
+      "learning_rate": 4.374285714285715e-05,
+      "loss": 1.698,
       "step": 219
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 3.600508905852418e-05,
-      "loss": 1.7317,
+      "epoch": 1.26,
+      "learning_rate": 4.371428571428572e-05,
+      "loss": 1.664,
       "step": 220
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 3.594147582697201e-05,
-      "loss": 1.7028,
+      "epoch": 1.26,
+      "learning_rate": 4.368571428571429e-05,
+      "loss": 1.6258,
       "step": 221
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 3.5877862595419845e-05,
-      "loss": 1.637,
+      "epoch": 1.27,
+      "learning_rate": 4.3657142857142855e-05,
+      "loss": 1.6731,
       "step": 222
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 3.581424936386768e-05,
-      "loss": 1.7293,
+      "epoch": 1.27,
+      "learning_rate": 4.362857142857143e-05,
+      "loss": 1.7655,
       "step": 223
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 3.5750636132315526e-05,
-      "loss": 1.7743,
+      "epoch": 1.28,
+      "learning_rate": 4.36e-05,
+      "loss": 1.579,
       "step": 224
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 3.568702290076336e-05,
-      "loss": 1.6725,
+      "epoch": 1.29,
+      "learning_rate": 4.3571428571428576e-05,
+      "loss": 1.6825,
       "step": 225
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 3.56234096692112e-05,
-      "loss": 1.7859,
+      "epoch": 1.29,
+      "learning_rate": 4.354285714285714e-05,
+      "loss": 1.6713,
       "step": 226
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 3.555979643765903e-05,
-      "loss": 1.662,
+      "epoch": 1.3,
+      "learning_rate": 4.351428571428572e-05,
+      "loss": 1.6229,
       "step": 227
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 3.549618320610687e-05,
-      "loss": 1.7467,
+      "epoch": 1.3,
+      "learning_rate": 4.3485714285714284e-05,
+      "loss": 1.6599,
       "step": 228
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 3.5432569974554705e-05,
-      "loss": 1.6985,
+      "epoch": 1.31,
+      "learning_rate": 4.345714285714286e-05,
+      "loss": 1.6295,
       "step": 229
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 3.536895674300255e-05,
-      "loss": 1.7305,
+      "epoch": 1.31,
+      "learning_rate": 4.342857142857143e-05,
+      "loss": 1.6991,
       "step": 230
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 3.5305343511450386e-05,
-      "loss": 1.8776,
+      "epoch": 1.32,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 1.7486,
       "step": 231
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 3.524173027989822e-05,
-      "loss": 1.7783,
+      "epoch": 1.33,
+      "learning_rate": 4.337142857142857e-05,
+      "loss": 1.6171,
       "step": 232
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 3.517811704834606e-05,
-      "loss": 1.7381,
+      "epoch": 1.33,
+      "learning_rate": 4.3342857142857145e-05,
+      "loss": 1.6926,
       "step": 233
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 3.511450381679389e-05,
-      "loss": 1.8489,
+      "epoch": 1.34,
+      "learning_rate": 4.331428571428572e-05,
+      "loss": 1.7428,
       "step": 234
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 3.505089058524173e-05,
-      "loss": 1.7912,
+      "epoch": 1.34,
+      "learning_rate": 4.328571428571429e-05,
+      "loss": 1.6501,
       "step": 235
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 3.498727735368957e-05,
-      "loss": 1.7976,
+      "epoch": 1.35,
+      "learning_rate": 4.325714285714286e-05,
+      "loss": 1.6126,
       "step": 236
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 3.492366412213741e-05,
-      "loss": 1.7394,
+      "epoch": 1.35,
+      "learning_rate": 4.3228571428571426e-05,
+      "loss": 1.5454,
       "step": 237
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 3.4860050890585245e-05,
-      "loss": 1.6584,
+      "epoch": 1.36,
+      "learning_rate": 4.32e-05,
+      "loss": 1.6699,
       "step": 238
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 3.479643765903308e-05,
-      "loss": 1.8044,
+      "epoch": 1.37,
+      "learning_rate": 4.317142857142857e-05,
+      "loss": 1.6015,
       "step": 239
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.473282442748092e-05,
-      "loss": 1.7773,
+      "epoch": 1.37,
+      "learning_rate": 4.314285714285715e-05,
+      "loss": 1.6008,
       "step": 240
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.466921119592875e-05,
-      "loss": 1.76,
+      "epoch": 1.38,
+      "learning_rate": 4.311428571428572e-05,
+      "loss": 1.7071,
       "step": 241
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 3.4605597964376594e-05,
-      "loss": 1.7517,
+      "epoch": 1.38,
+      "learning_rate": 4.308571428571429e-05,
+      "loss": 1.6925,
       "step": 242
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 3.454198473282443e-05,
-      "loss": 1.7994,
+      "epoch": 1.39,
+      "learning_rate": 4.3057142857142854e-05,
+      "loss": 1.6395,
       "step": 243
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 3.447837150127227e-05,
-      "loss": 1.7396,
+      "epoch": 1.39,
+      "learning_rate": 4.302857142857143e-05,
+      "loss": 1.7645,
       "step": 244
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 3.4414758269720105e-05,
-      "loss": 1.7299,
+      "epoch": 1.4,
+      "learning_rate": 4.3e-05,
+      "loss": 1.6794,
       "step": 245
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 3.435114503816794e-05,
-      "loss": 1.6403,
+      "epoch": 1.41,
+      "learning_rate": 4.2971428571428575e-05,
+      "loss": 1.6595,
       "step": 246
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 3.428753180661578e-05,
-      "loss": 1.6074,
+      "epoch": 1.41,
+      "learning_rate": 4.294285714285715e-05,
+      "loss": 1.5587,
       "step": 247
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 3.4223918575063616e-05,
-      "loss": 1.7362,
+      "epoch": 1.42,
+      "learning_rate": 4.2914285714285716e-05,
+      "loss": 1.6271,
       "step": 248
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 3.416030534351145e-05,
-      "loss": 1.6809,
+      "epoch": 1.42,
+      "learning_rate": 4.288571428571429e-05,
+      "loss": 1.5774,
       "step": 249
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 3.409669211195929e-05,
-      "loss": 1.7009,
+      "epoch": 1.43,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 1.7164,
       "step": 250
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 3.403307888040713e-05,
-      "loss": 1.7218,
+      "epoch": 1.43,
+      "learning_rate": 4.282857142857143e-05,
+      "loss": 1.6044,
       "step": 251
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 3.3969465648854964e-05,
-      "loss": 1.7234,
+      "epoch": 1.44,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 1.607,
       "step": 252
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 3.39058524173028e-05,
-      "loss": 1.6854,
+      "epoch": 1.45,
+      "learning_rate": 4.277142857142857e-05,
+      "loss": 1.4833,
       "step": 253
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 3.384223918575064e-05,
-      "loss": 1.7727,
+      "epoch": 1.45,
+      "learning_rate": 4.2742857142857144e-05,
+      "loss": 1.6581,
       "step": 254
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 3.3778625954198475e-05,
-      "loss": 1.7518,
+      "epoch": 1.46,
+      "learning_rate": 4.271428571428572e-05,
+      "loss": 1.6506,
       "step": 255
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 3.371501272264631e-05,
-      "loss": 1.6936,
+      "epoch": 1.46,
+      "learning_rate": 4.268571428571429e-05,
+      "loss": 1.6264,
       "step": 256
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 3.365139949109415e-05,
-      "loss": 1.8428,
+      "epoch": 1.47,
+      "learning_rate": 4.265714285714286e-05,
+      "loss": 1.5609,
       "step": 257
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 3.358778625954199e-05,
-      "loss": 1.725,
+      "epoch": 1.47,
+      "learning_rate": 4.262857142857143e-05,
+      "loss": 1.6017,
       "step": 258
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 3.3524173027989824e-05,
-      "loss": 1.6898,
+      "epoch": 1.48,
+      "learning_rate": 4.26e-05,
+      "loss": 1.6361,
       "step": 259
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 3.346055979643766e-05,
-      "loss": 1.7259,
+      "epoch": 1.49,
+      "learning_rate": 4.257142857142857e-05,
+      "loss": 1.6868,
       "step": 260
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 3.33969465648855e-05,
-      "loss": 1.665,
+      "epoch": 1.49,
+      "learning_rate": 4.2542857142857146e-05,
+      "loss": 1.6199,
       "step": 261
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 1.8255,
+      "epoch": 1.5,
+      "learning_rate": 4.251428571428572e-05,
+      "loss": 1.6722,
       "step": 262
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 3.326972010178117e-05,
-      "loss": 1.7793,
+      "epoch": 1.5,
+      "learning_rate": 4.2485714285714286e-05,
+      "loss": 1.576,
       "step": 263
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 3.320610687022901e-05,
-      "loss": 1.7146,
+      "epoch": 1.51,
+      "learning_rate": 4.245714285714285e-05,
+      "loss": 1.6202,
       "step": 264
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 3.3142493638676846e-05,
-      "loss": 1.6895,
+      "epoch": 1.51,
+      "learning_rate": 4.242857142857143e-05,
+      "loss": 1.5588,
       "step": 265
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 3.307888040712468e-05,
-      "loss": 1.7582,
+      "epoch": 1.52,
+      "learning_rate": 4.24e-05,
+      "loss": 1.6601,
       "step": 266
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 3.301526717557252e-05,
-      "loss": 1.7078,
+      "epoch": 1.53,
+      "learning_rate": 4.2371428571428574e-05,
+      "loss": 1.7059,
       "step": 267
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 3.295165394402036e-05,
-      "loss": 1.7425,
+      "epoch": 1.53,
+      "learning_rate": 4.234285714285715e-05,
+      "loss": 1.621,
       "step": 268
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 3.2888040712468194e-05,
-      "loss": 1.6471,
+      "epoch": 1.54,
+      "learning_rate": 4.2314285714285715e-05,
+      "loss": 1.5852,
       "step": 269
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 3.282442748091603e-05,
-      "loss": 1.6362,
+      "epoch": 1.54,
+      "learning_rate": 4.228571428571429e-05,
+      "loss": 1.5851,
       "step": 270
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 3.276081424936387e-05,
-      "loss": 1.7659,
+      "epoch": 1.55,
+      "learning_rate": 4.225714285714286e-05,
+      "loss": 1.6144,
       "step": 271
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 3.2697201017811706e-05,
-      "loss": 1.6603,
+      "epoch": 1.55,
+      "learning_rate": 4.222857142857143e-05,
+      "loss": 1.5543,
       "step": 272
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 3.263358778625954e-05,
-      "loss": 1.7568,
+      "epoch": 1.56,
+      "learning_rate": 4.22e-05,
+      "loss": 1.5907,
       "step": 273
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 3.256997455470738e-05,
-      "loss": 1.6339,
+      "epoch": 1.57,
+      "learning_rate": 4.2171428571428576e-05,
+      "loss": 1.6405,
       "step": 274
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 3.250636132315522e-05,
-      "loss": 1.7437,
+      "epoch": 1.57,
+      "learning_rate": 4.214285714285714e-05,
+      "loss": 1.5462,
       "step": 275
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 3.2442748091603054e-05,
-      "loss": 1.6258,
+      "epoch": 1.58,
+      "learning_rate": 4.211428571428572e-05,
+      "loss": 1.6405,
       "step": 276
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 3.237913486005089e-05,
-      "loss": 1.6025,
+      "epoch": 1.58,
+      "learning_rate": 4.208571428571429e-05,
+      "loss": 1.7377,
       "step": 277
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 3.231552162849873e-05,
-      "loss": 1.5755,
+      "epoch": 1.59,
+      "learning_rate": 4.2057142857142864e-05,
+      "loss": 1.6308,
       "step": 278
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 3.2251908396946565e-05,
-      "loss": 1.6287,
+      "epoch": 1.59,
+      "learning_rate": 4.202857142857143e-05,
+      "loss": 1.564,
       "step": 279
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 3.21882951653944e-05,
-      "loss": 1.6126,
+      "epoch": 1.6,
+      "learning_rate": 4.2e-05,
+      "loss": 1.7135,
       "step": 280
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 3.2124681933842246e-05,
-      "loss": 1.7431,
+      "epoch": 1.61,
+      "learning_rate": 4.197142857142857e-05,
+      "loss": 1.5188,
       "step": 281
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 3.2061068702290076e-05,
-      "loss": 1.7279,
+      "epoch": 1.61,
+      "learning_rate": 4.1942857142857145e-05,
+      "loss": 1.6114,
       "step": 282
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 3.1997455470737913e-05,
-      "loss": 1.5722,
+      "epoch": 1.62,
+      "learning_rate": 4.191428571428572e-05,
+      "loss": 1.7466,
       "step": 283
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 3.193384223918575e-05,
-      "loss": 1.5911,
+      "epoch": 1.62,
+      "learning_rate": 4.188571428571429e-05,
+      "loss": 1.5667,
       "step": 284
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 3.187022900763359e-05,
-      "loss": 1.5626,
+      "epoch": 1.63,
+      "learning_rate": 4.185714285714286e-05,
+      "loss": 1.6834,
       "step": 285
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 3.1806615776081425e-05,
-      "loss": 1.6939,
+      "epoch": 1.63,
+      "learning_rate": 4.1828571428571426e-05,
+      "loss": 1.5684,
       "step": 286
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 3.174300254452927e-05,
-      "loss": 1.6078,
+      "epoch": 1.64,
+      "learning_rate": 4.18e-05,
+      "loss": 1.6175,
       "step": 287
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 3.16793893129771e-05,
-      "loss": 1.8064,
+      "epoch": 1.65,
+      "learning_rate": 4.177142857142857e-05,
+      "loss": 1.6588,
       "step": 288
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 3.1615776081424936e-05,
-      "loss": 1.7212,
+      "epoch": 1.65,
+      "learning_rate": 4.174285714285715e-05,
+      "loss": 1.6149,
       "step": 289
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 3.155216284987277e-05,
-      "loss": 1.5587,
+      "epoch": 1.66,
+      "learning_rate": 4.1714285714285714e-05,
+      "loss": 1.6206,
       "step": 290
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 3.148854961832061e-05,
-      "loss": 1.7182,
+      "epoch": 1.66,
+      "learning_rate": 4.168571428571429e-05,
+      "loss": 1.6661,
       "step": 291
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 3.142493638676845e-05,
-      "loss": 1.6683,
+      "epoch": 1.67,
+      "learning_rate": 4.165714285714286e-05,
+      "loss": 1.6249,
       "step": 292
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 3.136132315521629e-05,
-      "loss": 1.7011,
+      "epoch": 1.67,
+      "learning_rate": 4.162857142857143e-05,
+      "loss": 1.65,
       "step": 293
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 3.129770992366413e-05,
-      "loss": 1.6802,
+      "epoch": 1.68,
+      "learning_rate": 4.16e-05,
+      "loss": 1.5942,
       "step": 294
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 3.123409669211196e-05,
-      "loss": 1.6369,
+      "epoch": 1.69,
+      "learning_rate": 4.1571428571428575e-05,
+      "loss": 1.5924,
       "step": 295
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 3.1170483460559795e-05,
-      "loss": 1.6679,
+      "epoch": 1.69,
+      "learning_rate": 4.154285714285714e-05,
+      "loss": 1.5961,
       "step": 296
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 3.110687022900763e-05,
-      "loss": 1.645,
+      "epoch": 1.7,
+      "learning_rate": 4.1514285714285716e-05,
+      "loss": 1.6469,
       "step": 297
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 3.104325699745547e-05,
-      "loss": 1.6605,
+      "epoch": 1.7,
+      "learning_rate": 4.148571428571429e-05,
+      "loss": 1.6676,
       "step": 298
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 3.097964376590331e-05,
-      "loss": 1.5894,
+      "epoch": 1.71,
+      "learning_rate": 4.145714285714286e-05,
+      "loss": 1.5858,
       "step": 299
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 3.091603053435115e-05,
-      "loss": 1.6941,
+      "epoch": 1.71,
+      "learning_rate": 4.1428571428571437e-05,
+      "loss": 1.5751,
       "step": 300
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 3.085241730279899e-05,
-      "loss": 1.7037,
+      "epoch": 1.72,
+      "learning_rate": 4.14e-05,
+      "loss": 1.5768,
       "step": 301
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 3.078880407124682e-05,
-      "loss": 1.595,
+      "epoch": 1.73,
+      "learning_rate": 4.137142857142857e-05,
+      "loss": 1.5679,
       "step": 302
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 3.0725190839694655e-05,
-      "loss": 1.6429,
+      "epoch": 1.73,
+      "learning_rate": 4.1342857142857144e-05,
+      "loss": 1.4867,
       "step": 303
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 3.066157760814249e-05,
-      "loss": 1.6009,
+      "epoch": 1.74,
+      "learning_rate": 4.131428571428572e-05,
+      "loss": 1.6305,
       "step": 304
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 3.0597964376590336e-05,
-      "loss": 1.7152,
+      "epoch": 1.74,
+      "learning_rate": 4.128571428571429e-05,
+      "loss": 1.5274,
       "step": 305
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 3.053435114503817e-05,
-      "loss": 1.7318,
+      "epoch": 1.75,
+      "learning_rate": 4.125714285714286e-05,
+      "loss": 1.7055,
       "step": 306
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 3.047073791348601e-05,
-      "loss": 1.6328,
+      "epoch": 1.75,
+      "learning_rate": 4.122857142857143e-05,
+      "loss": 1.6375,
       "step": 307
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 3.0407124681933847e-05,
-      "loss": 1.6322,
+      "epoch": 1.76,
+      "learning_rate": 4.12e-05,
+      "loss": 1.6141,
       "step": 308
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 3.0343511450381677e-05,
-      "loss": 1.5534,
+      "epoch": 1.77,
+      "learning_rate": 4.117142857142857e-05,
+      "loss": 1.6588,
       "step": 309
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 3.0279898218829518e-05,
-      "loss": 1.6483,
+      "epoch": 1.77,
+      "learning_rate": 4.1142857142857146e-05,
+      "loss": 1.5327,
       "step": 310
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 3.0216284987277355e-05,
-      "loss": 1.5785,
+      "epoch": 1.78,
+      "learning_rate": 4.111428571428572e-05,
+      "loss": 1.522,
       "step": 311
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 3.0152671755725192e-05,
-      "loss": 1.7712,
+      "epoch": 1.78,
+      "learning_rate": 4.1085714285714286e-05,
+      "loss": 1.525,
       "step": 312
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 3.008905852417303e-05,
-      "loss": 1.5643,
+      "epoch": 1.79,
+      "learning_rate": 4.105714285714286e-05,
+      "loss": 1.6038,
       "step": 313
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 3.002544529262087e-05,
-      "loss": 1.6814,
+      "epoch": 1.79,
+      "learning_rate": 4.1028571428571434e-05,
+      "loss": 1.6005,
       "step": 314
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 2.9961832061068706e-05,
-      "loss": 1.5747,
+      "epoch": 1.8,
+      "learning_rate": 4.1e-05,
+      "loss": 1.685,
       "step": 315
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 2.989821882951654e-05,
-      "loss": 1.6542,
+      "epoch": 1.81,
+      "learning_rate": 4.0971428571428574e-05,
+      "loss": 1.5959,
       "step": 316
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 2.9834605597964377e-05,
-      "loss": 1.5891,
+      "epoch": 1.81,
+      "learning_rate": 4.094285714285714e-05,
+      "loss": 1.6811,
       "step": 317
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 2.9770992366412214e-05,
-      "loss": 1.66,
+      "epoch": 1.82,
+      "learning_rate": 4.0914285714285715e-05,
+      "loss": 1.6708,
       "step": 318
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 2.970737913486005e-05,
-      "loss": 1.7994,
+      "epoch": 1.82,
+      "learning_rate": 4.088571428571429e-05,
+      "loss": 1.5821,
       "step": 319
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 2.9643765903307892e-05,
-      "loss": 1.6422,
+      "epoch": 1.83,
+      "learning_rate": 4.085714285714286e-05,
+      "loss": 1.6497,
       "step": 320
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 2.958015267175573e-05,
-      "loss": 1.6212,
+      "epoch": 1.83,
+      "learning_rate": 4.0828571428571436e-05,
+      "loss": 1.5811,
       "step": 321
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 2.9516539440203562e-05,
-      "loss": 1.5684,
+      "epoch": 1.84,
+      "learning_rate": 4.08e-05,
+      "loss": 1.5892,
       "step": 322
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 2.94529262086514e-05,
-      "loss": 1.5803,
+      "epoch": 1.85,
+      "learning_rate": 4.077142857142857e-05,
+      "loss": 1.5138,
       "step": 323
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 2.9389312977099237e-05,
-      "loss": 1.677,
+      "epoch": 1.85,
+      "learning_rate": 4.074285714285714e-05,
+      "loss": 1.5428,
       "step": 324
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 2.9325699745547074e-05,
-      "loss": 1.6106,
+      "epoch": 1.86,
+      "learning_rate": 4.0714285714285717e-05,
+      "loss": 1.6193,
       "step": 325
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 2.9262086513994914e-05,
-      "loss": 1.6198,
+      "epoch": 1.86,
+      "learning_rate": 4.068571428571429e-05,
+      "loss": 1.5663,
       "step": 326
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 2.919847328244275e-05,
-      "loss": 1.575,
+      "epoch": 1.87,
+      "learning_rate": 4.065714285714286e-05,
+      "loss": 1.6173,
       "step": 327
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 2.9134860050890588e-05,
-      "loss": 1.6581,
+      "epoch": 1.87,
+      "learning_rate": 4.062857142857143e-05,
+      "loss": 1.5654,
       "step": 328
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 2.9071246819338422e-05,
-      "loss": 1.655,
+      "epoch": 1.88,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 1.5059,
       "step": 329
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 2.900763358778626e-05,
-      "loss": 1.58,
+      "epoch": 1.89,
+      "learning_rate": 4.057142857142857e-05,
+      "loss": 1.5655,
       "step": 330
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 2.8944020356234096e-05,
-      "loss": 1.6716,
+      "epoch": 1.89,
+      "learning_rate": 4.0542857142857145e-05,
+      "loss": 1.6775,
       "step": 331
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 2.8880407124681937e-05,
-      "loss": 1.6456,
+      "epoch": 1.9,
+      "learning_rate": 4.051428571428572e-05,
+      "loss": 1.6367,
       "step": 332
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 2.8816793893129774e-05,
-      "loss": 1.6167,
+      "epoch": 1.9,
+      "learning_rate": 4.0485714285714285e-05,
+      "loss": 1.6234,
       "step": 333
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 2.875318066157761e-05,
-      "loss": 1.7806,
+      "epoch": 1.91,
+      "learning_rate": 4.045714285714286e-05,
+      "loss": 1.6336,
       "step": 334
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 2.8689567430025448e-05,
-      "loss": 1.5511,
+      "epoch": 1.91,
+      "learning_rate": 4.042857142857143e-05,
+      "loss": 1.5717,
       "step": 335
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 2.862595419847328e-05,
-      "loss": 1.5907,
+      "epoch": 1.92,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 1.6782,
       "step": 336
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 2.856234096692112e-05,
-      "loss": 1.6821,
+      "epoch": 1.93,
+      "learning_rate": 4.037142857142857e-05,
+      "loss": 1.4754,
       "step": 337
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 2.849872773536896e-05,
-      "loss": 1.6402,
+      "epoch": 1.93,
+      "learning_rate": 4.034285714285715e-05,
+      "loss": 1.626,
       "step": 338
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 2.8435114503816796e-05,
-      "loss": 1.5662,
+      "epoch": 1.94,
+      "learning_rate": 4.0314285714285714e-05,
+      "loss": 1.5332,
       "step": 339
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 2.8371501272264633e-05,
-      "loss": 1.6689,
+      "epoch": 1.94,
+      "learning_rate": 4.028571428571429e-05,
+      "loss": 1.4984,
       "step": 340
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 2.830788804071247e-05,
-      "loss": 1.5855,
+      "epoch": 1.95,
+      "learning_rate": 4.025714285714286e-05,
+      "loss": 1.5877,
       "step": 341
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 2.824427480916031e-05,
-      "loss": 1.6353,
+      "epoch": 1.95,
+      "learning_rate": 4.0228571428571434e-05,
+      "loss": 1.6498,
       "step": 342
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 2.818066157760814e-05,
-      "loss": 1.6703,
+      "epoch": 1.96,
+      "learning_rate": 4.02e-05,
+      "loss": 1.6192,
       "step": 343
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 2.8117048346055978e-05,
-      "loss": 1.6515,
+      "epoch": 1.97,
+      "learning_rate": 4.017142857142857e-05,
+      "loss": 1.5261,
       "step": 344
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 2.805343511450382e-05,
-      "loss": 1.6773,
+      "epoch": 1.97,
+      "learning_rate": 4.014285714285714e-05,
+      "loss": 1.6036,
       "step": 345
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 2.7989821882951656e-05,
-      "loss": 1.7637,
+      "epoch": 1.98,
+      "learning_rate": 4.0114285714285715e-05,
+      "loss": 1.6491,
       "step": 346
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 2.7926208651399493e-05,
-      "loss": 1.6192,
+      "epoch": 1.98,
+      "learning_rate": 4.008571428571429e-05,
+      "loss": 1.5306,
       "step": 347
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 2.7862595419847333e-05,
-      "loss": 1.6012,
+      "epoch": 1.99,
+      "learning_rate": 4.005714285714286e-05,
+      "loss": 1.5968,
       "step": 348
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 2.7798982188295163e-05,
-      "loss": 1.6582,
+      "epoch": 1.99,
+      "learning_rate": 4.002857142857143e-05,
+      "loss": 1.6654,
       "step": 349
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 2.7735368956743e-05,
-      "loss": 1.7287,
+      "epoch": 2.0,
+      "learning_rate": 4e-05,
+      "loss": 1.6264,
       "step": 350
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 2.767175572519084e-05,
-      "loss": 1.7468,
+      "epoch": 2.01,
+      "learning_rate": 3.997142857142857e-05,
+      "loss": 1.5485,
       "step": 351
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 2.7608142493638678e-05,
-      "loss": 1.5716,
+      "epoch": 2.01,
+      "learning_rate": 3.9942857142857144e-05,
+      "loss": 1.5662,
       "step": 352
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 2.7544529262086515e-05,
-      "loss": 1.6535,
+      "epoch": 2.02,
+      "learning_rate": 3.991428571428572e-05,
+      "loss": 1.5641,
       "step": 353
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 2.7480916030534355e-05,
-      "loss": 1.479,
+      "epoch": 2.02,
+      "learning_rate": 3.9885714285714284e-05,
+      "loss": 1.6021,
       "step": 354
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 2.7417302798982193e-05,
-      "loss": 1.5744,
+      "epoch": 2.03,
+      "learning_rate": 3.985714285714286e-05,
+      "loss": 1.5712,
       "step": 355
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 2.7353689567430023e-05,
-      "loss": 1.6393,
+      "epoch": 2.03,
+      "learning_rate": 3.982857142857143e-05,
+      "loss": 1.4838,
       "step": 356
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 2.7290076335877863e-05,
-      "loss": 1.5814,
+      "epoch": 2.04,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.5519,
       "step": 357
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 2.72264631043257e-05,
-      "loss": 1.6316,
+      "epoch": 2.05,
+      "learning_rate": 3.977142857142857e-05,
+      "loss": 1.4563,
       "step": 358
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 2.7162849872773537e-05,
-      "loss": 1.6321,
+      "epoch": 2.05,
+      "learning_rate": 3.9742857142857146e-05,
+      "loss": 1.4443,
       "step": 359
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 2.7099236641221375e-05,
-      "loss": 1.6176,
+      "epoch": 2.06,
+      "learning_rate": 3.971428571428571e-05,
+      "loss": 1.5643,
       "step": 360
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 2.7035623409669215e-05,
-      "loss": 1.6889,
+      "epoch": 2.06,
+      "learning_rate": 3.9685714285714286e-05,
+      "loss": 1.5135,
       "step": 361
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 2.6972010178117052e-05,
-      "loss": 1.6337,
+      "epoch": 2.07,
+      "learning_rate": 3.965714285714286e-05,
+      "loss": 1.4882,
       "step": 362
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 2.6908396946564886e-05,
-      "loss": 1.6375,
+      "epoch": 2.07,
+      "learning_rate": 3.9628571428571433e-05,
+      "loss": 1.4846,
       "step": 363
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 2.6844783715012723e-05,
-      "loss": 1.6314,
+      "epoch": 2.08,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 1.4128,
       "step": 364
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 2.678117048346056e-05,
-      "loss": 1.73,
+      "epoch": 2.09,
+      "learning_rate": 3.9571428571428574e-05,
+      "loss": 1.5847,
       "step": 365
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 2.6717557251908397e-05,
-      "loss": 1.676,
+      "epoch": 2.09,
+      "learning_rate": 3.954285714285714e-05,
+      "loss": 1.5023,
       "step": 366
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 2.6653944020356237e-05,
-      "loss": 1.6735,
+      "epoch": 2.1,
+      "learning_rate": 3.9514285714285714e-05,
+      "loss": 1.4424,
       "step": 367
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 2.6590330788804074e-05,
-      "loss": 1.6559,
+      "epoch": 2.1,
+      "learning_rate": 3.948571428571429e-05,
+      "loss": 1.5342,
       "step": 368
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 2.652671755725191e-05,
-      "loss": 1.5847,
+      "epoch": 2.11,
+      "learning_rate": 3.945714285714286e-05,
+      "loss": 1.5231,
       "step": 369
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 2.6463104325699745e-05,
-      "loss": 1.4857,
+      "epoch": 2.11,
+      "learning_rate": 3.942857142857143e-05,
+      "loss": 1.4948,
       "step": 370
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 2.6399491094147582e-05,
-      "loss": 1.5846,
+      "epoch": 2.12,
+      "learning_rate": 3.94e-05,
+      "loss": 1.5297,
       "step": 371
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 2.633587786259542e-05,
-      "loss": 1.6077,
+      "epoch": 2.13,
+      "learning_rate": 3.9371428571428576e-05,
+      "loss": 1.5705,
       "step": 372
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 2.627226463104326e-05,
-      "loss": 1.5612,
+      "epoch": 2.13,
+      "learning_rate": 3.934285714285714e-05,
+      "loss": 1.5808,
       "step": 373
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 2.6208651399491097e-05,
-      "loss": 1.7214,
+      "epoch": 2.14,
+      "learning_rate": 3.9314285714285716e-05,
+      "loss": 1.6704,
       "step": 374
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 2.6145038167938934e-05,
-      "loss": 1.6248,
+      "epoch": 2.14,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 1.5372,
       "step": 375
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 2.608142493638677e-05,
-      "loss": 1.5545,
+      "epoch": 2.15,
+      "learning_rate": 3.925714285714286e-05,
+      "loss": 1.548,
       "step": 376
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 2.6017811704834605e-05,
-      "loss": 1.5943,
+      "epoch": 2.15,
+      "learning_rate": 3.922857142857143e-05,
+      "loss": 1.4958,
       "step": 377
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 2.5954198473282442e-05,
-      "loss": 1.5038,
+      "epoch": 2.16,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 1.4967,
       "step": 378
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 2.5890585241730282e-05,
-      "loss": 1.4453,
+      "epoch": 2.17,
+      "learning_rate": 3.917142857142858e-05,
+      "loss": 1.4774,
       "step": 379
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 2.582697201017812e-05,
-      "loss": 1.6681,
+      "epoch": 2.17,
+      "learning_rate": 3.9142857142857145e-05,
+      "loss": 1.5052,
       "step": 380
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 2.5763358778625956e-05,
-      "loss": 1.5682,
+      "epoch": 2.18,
+      "learning_rate": 3.911428571428571e-05,
+      "loss": 1.4988,
       "step": 381
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 2.5699745547073793e-05,
-      "loss": 1.6644,
+      "epoch": 2.18,
+      "learning_rate": 3.9085714285714285e-05,
+      "loss": 1.507,
       "step": 382
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 2.5636132315521627e-05,
-      "loss": 1.565,
+      "epoch": 2.19,
+      "learning_rate": 3.905714285714286e-05,
+      "loss": 1.439,
       "step": 383
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 2.5572519083969464e-05,
-      "loss": 1.656,
+      "epoch": 2.19,
+      "learning_rate": 3.902857142857143e-05,
+      "loss": 1.4346,
       "step": 384
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 2.5508905852417305e-05,
-      "loss": 1.496,
+      "epoch": 2.2,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.5832,
       "step": 385
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 2.5445292620865142e-05,
-      "loss": 1.6039,
+      "epoch": 2.21,
+      "learning_rate": 3.897142857142857e-05,
+      "loss": 1.5271,
       "step": 386
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 2.538167938931298e-05,
-      "loss": 1.5849,
+      "epoch": 2.21,
+      "learning_rate": 3.894285714285714e-05,
+      "loss": 1.553,
       "step": 387
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 2.5318066157760816e-05,
-      "loss": 1.6688,
+      "epoch": 2.22,
+      "learning_rate": 3.8914285714285713e-05,
+      "loss": 1.5303,
       "step": 388
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 2.5254452926208656e-05,
-      "loss": 1.6819,
+      "epoch": 2.22,
+      "learning_rate": 3.888571428571429e-05,
+      "loss": 1.4648,
       "step": 389
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 2.5190839694656487e-05,
-      "loss": 1.6531,
+      "epoch": 2.23,
+      "learning_rate": 3.885714285714286e-05,
+      "loss": 1.5702,
       "step": 390
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 2.5127226463104324e-05,
-      "loss": 1.6093,
+      "epoch": 2.23,
+      "learning_rate": 3.882857142857143e-05,
+      "loss": 1.527,
       "step": 391
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 2.5063613231552164e-05,
-      "loss": 1.594,
+      "epoch": 2.24,
+      "learning_rate": 3.88e-05,
+      "loss": 1.5519,
       "step": 392
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 2.5e-05,
-      "loss": 1.6348,
+      "epoch": 2.25,
+      "learning_rate": 3.8771428571428575e-05,
+      "loss": 1.4864,
       "step": 393
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 2.4936386768447838e-05,
-      "loss": 1.5857,
+      "epoch": 2.25,
+      "learning_rate": 3.874285714285715e-05,
+      "loss": 1.4479,
       "step": 394
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.4872773536895675e-05,
-      "loss": 1.5962,
+      "epoch": 2.26,
+      "learning_rate": 3.8714285714285715e-05,
+      "loss": 1.4971,
       "step": 395
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.4809160305343512e-05,
-      "loss": 1.5256,
+      "epoch": 2.26,
+      "learning_rate": 3.868571428571429e-05,
+      "loss": 1.51,
       "step": 396
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.474554707379135e-05,
-      "loss": 1.5459,
+      "epoch": 2.27,
+      "learning_rate": 3.8657142857142856e-05,
+      "loss": 1.5306,
       "step": 397
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.468193384223919e-05,
-      "loss": 1.7054,
+      "epoch": 2.27,
+      "learning_rate": 3.862857142857143e-05,
+      "loss": 1.471,
       "step": 398
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.4618320610687024e-05,
-      "loss": 1.5398,
+      "epoch": 2.28,
+      "learning_rate": 3.86e-05,
+      "loss": 1.5155,
       "step": 399
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.455470737913486e-05,
-      "loss": 1.7594,
+      "epoch": 2.29,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 1.533,
       "step": 400
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.4491094147582698e-05,
-      "loss": 1.614,
+      "epoch": 2.29,
+      "learning_rate": 3.854285714285715e-05,
+      "loss": 1.5073,
       "step": 401
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.4427480916030535e-05,
-      "loss": 1.5603,
+      "epoch": 2.3,
+      "learning_rate": 3.851428571428571e-05,
+      "loss": 1.5348,
       "step": 402
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.4363867684478372e-05,
-      "loss": 1.6489,
+      "epoch": 2.3,
+      "learning_rate": 3.8485714285714284e-05,
+      "loss": 1.4625,
       "step": 403
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.430025445292621e-05,
-      "loss": 1.5972,
+      "epoch": 2.31,
+      "learning_rate": 3.845714285714286e-05,
+      "loss": 1.5146,
       "step": 404
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.4236641221374046e-05,
-      "loss": 1.523,
+      "epoch": 2.31,
+      "learning_rate": 3.842857142857143e-05,
+      "loss": 1.4724,
       "step": 405
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.4173027989821883e-05,
-      "loss": 1.6404,
+      "epoch": 2.32,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 1.4763,
       "step": 406
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.410941475826972e-05,
-      "loss": 1.514,
+      "epoch": 2.33,
+      "learning_rate": 3.837142857142857e-05,
+      "loss": 1.4617,
       "step": 407
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.404580152671756e-05,
-      "loss": 1.5987,
+      "epoch": 2.33,
+      "learning_rate": 3.8342857142857146e-05,
+      "loss": 1.5511,
       "step": 408
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.3982188295165394e-05,
-      "loss": 1.5628,
+      "epoch": 2.34,
+      "learning_rate": 3.831428571428571e-05,
+      "loss": 1.6107,
       "step": 409
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.391857506361323e-05,
-      "loss": 1.5771,
+      "epoch": 2.34,
+      "learning_rate": 3.8285714285714286e-05,
+      "loss": 1.5942,
       "step": 410
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3854961832061072e-05,
-      "loss": 1.6674,
+      "epoch": 2.35,
+      "learning_rate": 3.825714285714286e-05,
+      "loss": 1.4694,
       "step": 411
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3791348600508906e-05,
-      "loss": 1.4884,
+      "epoch": 2.35,
+      "learning_rate": 3.822857142857143e-05,
+      "loss": 1.5044,
       "step": 412
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.3727735368956743e-05,
-      "loss": 1.6313,
+      "epoch": 2.36,
+      "learning_rate": 3.82e-05,
+      "loss": 1.4824,
       "step": 413
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.3664122137404583e-05,
-      "loss": 1.6976,
+      "epoch": 2.37,
+      "learning_rate": 3.8171428571428574e-05,
+      "loss": 1.6118,
       "step": 414
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.360050890585242e-05,
-      "loss": 1.7927,
+      "epoch": 2.37,
+      "learning_rate": 3.814285714285715e-05,
+      "loss": 1.5563,
       "step": 415
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.3536895674300254e-05,
-      "loss": 1.6724,
+      "epoch": 2.38,
+      "learning_rate": 3.8114285714285714e-05,
+      "loss": 1.4661,
       "step": 416
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.3473282442748094e-05,
-      "loss": 1.5105,
+      "epoch": 2.38,
+      "learning_rate": 3.808571428571429e-05,
+      "loss": 1.5446,
       "step": 417
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.340966921119593e-05,
-      "loss": 1.5567,
+      "epoch": 2.39,
+      "learning_rate": 3.8057142857142855e-05,
+      "loss": 1.5212,
       "step": 418
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.3346055979643765e-05,
-      "loss": 1.7265,
+      "epoch": 2.39,
+      "learning_rate": 3.802857142857143e-05,
+      "loss": 1.4784,
       "step": 419
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.3282442748091605e-05,
-      "loss": 1.5087,
+      "epoch": 2.4,
+      "learning_rate": 3.8e-05,
+      "loss": 1.5487,
       "step": 420
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 2.3218829516539443e-05,
-      "loss": 1.5541,
+      "epoch": 2.41,
+      "learning_rate": 3.7971428571428576e-05,
+      "loss": 1.5192,
       "step": 421
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 2.3155216284987276e-05,
-      "loss": 1.6069,
+      "epoch": 2.41,
+      "learning_rate": 3.794285714285715e-05,
+      "loss": 1.5653,
       "step": 422
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 2.3091603053435117e-05,
-      "loss": 1.722,
+      "epoch": 2.42,
+      "learning_rate": 3.7914285714285716e-05,
+      "loss": 1.5503,
       "step": 423
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 2.3027989821882954e-05,
-      "loss": 1.6451,
+      "epoch": 2.42,
+      "learning_rate": 3.788571428571428e-05,
+      "loss": 1.4077,
       "step": 424
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 2.296437659033079e-05,
-      "loss": 1.6769,
+      "epoch": 2.43,
+      "learning_rate": 3.785714285714286e-05,
+      "loss": 1.5807,
       "step": 425
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 2.2900763358778628e-05,
-      "loss": 1.5947,
+      "epoch": 2.43,
+      "learning_rate": 3.782857142857143e-05,
+      "loss": 1.5496,
       "step": 426
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 2.2837150127226465e-05,
-      "loss": 1.6694,
+      "epoch": 2.44,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.4505,
       "step": 427
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 2.2773536895674302e-05,
-      "loss": 1.6272,
+      "epoch": 2.45,
+      "learning_rate": 3.777142857142858e-05,
+      "loss": 1.313,
       "step": 428
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 2.270992366412214e-05,
-      "loss": 1.4854,
+      "epoch": 2.45,
+      "learning_rate": 3.7742857142857145e-05,
+      "loss": 1.4256,
       "step": 429
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 2.2646310432569976e-05,
-      "loss": 1.6514,
+      "epoch": 2.46,
+      "learning_rate": 3.771428571428572e-05,
+      "loss": 1.5583,
       "step": 430
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 2.2582697201017813e-05,
-      "loss": 1.5813,
+      "epoch": 2.46,
+      "learning_rate": 3.7685714285714285e-05,
+      "loss": 1.5243,
       "step": 431
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 2.2519083969465647e-05,
-      "loss": 1.6264,
+      "epoch": 2.47,
+      "learning_rate": 3.765714285714286e-05,
+      "loss": 1.4804,
       "step": 432
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 2.2455470737913487e-05,
-      "loss": 1.6744,
+      "epoch": 2.47,
+      "learning_rate": 3.762857142857143e-05,
+      "loss": 1.5216,
       "step": 433
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 2.2391857506361324e-05,
-      "loss": 1.5913,
+      "epoch": 2.48,
+      "learning_rate": 3.76e-05,
+      "loss": 1.4271,
       "step": 434
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 2.232824427480916e-05,
-      "loss": 1.5928,
+      "epoch": 2.49,
+      "learning_rate": 3.757142857142857e-05,
+      "loss": 1.4489,
       "step": 435
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 2.2264631043257e-05,
-      "loss": 1.676,
+      "epoch": 2.49,
+      "learning_rate": 3.7542857142857146e-05,
+      "loss": 1.4317,
       "step": 436
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 2.2201017811704836e-05,
-      "loss": 1.5805,
+      "epoch": 2.5,
+      "learning_rate": 3.751428571428572e-05,
+      "loss": 1.439,
       "step": 437
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 2.2137404580152673e-05,
-      "loss": 1.7112,
+      "epoch": 2.5,
+      "learning_rate": 3.748571428571429e-05,
+      "loss": 1.4071,
       "step": 438
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 2.207379134860051e-05,
-      "loss": 1.6407,
+      "epoch": 2.51,
+      "learning_rate": 3.745714285714286e-05,
+      "loss": 1.6372,
       "step": 439
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 2.2010178117048347e-05,
-      "loss": 1.5911,
+      "epoch": 2.51,
+      "learning_rate": 3.742857142857143e-05,
+      "loss": 1.5466,
       "step": 440
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 2.1946564885496184e-05,
-      "loss": 1.6219,
+      "epoch": 2.52,
+      "learning_rate": 3.74e-05,
+      "loss": 1.4779,
       "step": 441
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 2.1882951653944024e-05,
-      "loss": 1.5206,
+      "epoch": 2.53,
+      "learning_rate": 3.7371428571428575e-05,
+      "loss": 1.5009,
       "step": 442
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 2.1819338422391858e-05,
-      "loss": 1.5171,
+      "epoch": 2.53,
+      "learning_rate": 3.734285714285715e-05,
+      "loss": 1.4252,
       "step": 443
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 2.1755725190839695e-05,
-      "loss": 1.6333,
+      "epoch": 2.54,
+      "learning_rate": 3.7314285714285715e-05,
+      "loss": 1.5151,
       "step": 444
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 2.1692111959287532e-05,
-      "loss": 1.6168,
+      "epoch": 2.54,
+      "learning_rate": 3.728571428571428e-05,
+      "loss": 1.5126,
       "step": 445
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 2.162849872773537e-05,
-      "loss": 1.6254,
+      "epoch": 2.55,
+      "learning_rate": 3.7257142857142856e-05,
+      "loss": 1.4436,
       "step": 446
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 2.1564885496183206e-05,
-      "loss": 1.7057,
+      "epoch": 2.55,
+      "learning_rate": 3.722857142857143e-05,
+      "loss": 1.4778,
       "step": 447
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 2.1501272264631043e-05,
-      "loss": 1.5022,
+      "epoch": 2.56,
+      "learning_rate": 3.72e-05,
+      "loss": 1.5047,
       "step": 448
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 2.143765903307888e-05,
-      "loss": 1.5533,
+      "epoch": 2.57,
+      "learning_rate": 3.717142857142858e-05,
+      "loss": 1.4456,
       "step": 449
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 2.1374045801526718e-05,
-      "loss": 1.5795,
+      "epoch": 2.57,
+      "learning_rate": 3.7142857142857143e-05,
+      "loss": 1.4651,
       "step": 450
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 2.1310432569974555e-05,
-      "loss": 1.5262,
+      "epoch": 2.58,
+      "learning_rate": 3.711428571428572e-05,
+      "loss": 1.5173,
       "step": 451
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 2.1246819338422395e-05,
-      "loss": 1.5867,
+      "epoch": 2.58,
+      "learning_rate": 3.7085714285714284e-05,
+      "loss": 1.4479,
       "step": 452
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 2.118320610687023e-05,
-      "loss": 1.4787,
+      "epoch": 2.59,
+      "learning_rate": 3.705714285714286e-05,
+      "loss": 1.5072,
       "step": 453
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 2.1119592875318066e-05,
-      "loss": 1.4586,
+      "epoch": 2.59,
+      "learning_rate": 3.702857142857143e-05,
+      "loss": 1.4199,
       "step": 454
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 2.1055979643765906e-05,
-      "loss": 1.6172,
+      "epoch": 2.6,
+      "learning_rate": 3.7e-05,
+      "loss": 1.528,
       "step": 455
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 2.099236641221374e-05,
-      "loss": 1.6178,
+      "epoch": 2.61,
+      "learning_rate": 3.697142857142857e-05,
+      "loss": 1.4877,
       "step": 456
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 2.0928753180661577e-05,
-      "loss": 1.4876,
+      "epoch": 2.61,
+      "learning_rate": 3.6942857142857145e-05,
+      "loss": 1.4451,
       "step": 457
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 2.0865139949109417e-05,
-      "loss": 1.5535,
+      "epoch": 2.62,
+      "learning_rate": 3.691428571428572e-05,
+      "loss": 1.5055,
       "step": 458
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 2.0801526717557255e-05,
-      "loss": 1.6825,
+      "epoch": 2.62,
+      "learning_rate": 3.688571428571429e-05,
+      "loss": 1.6091,
       "step": 459
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 2.0737913486005088e-05,
-      "loss": 1.6678,
+      "epoch": 2.63,
+      "learning_rate": 3.685714285714286e-05,
+      "loss": 1.4793,
       "step": 460
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 2.067430025445293e-05,
-      "loss": 1.581,
+      "epoch": 2.63,
+      "learning_rate": 3.6828571428571426e-05,
+      "loss": 1.5033,
       "step": 461
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 2.0610687022900766e-05,
-      "loss": 1.6045,
+      "epoch": 2.64,
+      "learning_rate": 3.68e-05,
+      "loss": 1.4965,
       "step": 462
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 2.05470737913486e-05,
-      "loss": 1.6811,
+      "epoch": 2.65,
+      "learning_rate": 3.6771428571428574e-05,
+      "loss": 1.4552,
       "step": 463
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 2.048346055979644e-05,
-      "loss": 1.5485,
+      "epoch": 2.65,
+      "learning_rate": 3.674285714285715e-05,
+      "loss": 1.5252,
       "step": 464
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 2.0419847328244277e-05,
-      "loss": 1.4733,
+      "epoch": 2.66,
+      "learning_rate": 3.671428571428572e-05,
+      "loss": 1.5216,
       "step": 465
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 2.035623409669211e-05,
-      "loss": 1.5043,
+      "epoch": 2.66,
+      "learning_rate": 3.668571428571429e-05,
+      "loss": 1.4452,
       "step": 466
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 2.029262086513995e-05,
-      "loss": 1.4733,
+      "epoch": 2.67,
+      "learning_rate": 3.6657142857142855e-05,
+      "loss": 1.4532,
       "step": 467
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 2.0229007633587788e-05,
-      "loss": 1.5753,
+      "epoch": 2.67,
+      "learning_rate": 3.662857142857143e-05,
+      "loss": 1.5612,
       "step": 468
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 2.0165394402035625e-05,
-      "loss": 1.5792,
+      "epoch": 2.68,
+      "learning_rate": 3.66e-05,
+      "loss": 1.5364,
       "step": 469
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 2.0101781170483462e-05,
-      "loss": 1.5859,
+      "epoch": 2.69,
+      "learning_rate": 3.6571428571428576e-05,
+      "loss": 1.4591,
       "step": 470
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 2.00381679389313e-05,
-      "loss": 1.6503,
+      "epoch": 2.69,
+      "learning_rate": 3.654285714285714e-05,
+      "loss": 1.5026,
       "step": 471
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 1.9974554707379136e-05,
-      "loss": 1.6675,
+      "epoch": 2.7,
+      "learning_rate": 3.6514285714285716e-05,
+      "loss": 1.5257,
       "step": 472
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 1.9910941475826974e-05,
-      "loss": 1.6658,
+      "epoch": 2.7,
+      "learning_rate": 3.648571428571429e-05,
+      "loss": 1.3648,
       "step": 473
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 1.984732824427481e-05,
-      "loss": 1.589,
+      "epoch": 2.71,
+      "learning_rate": 3.6457142857142857e-05,
+      "loss": 1.5092,
       "step": 474
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 1.9783715012722648e-05,
-      "loss": 1.6497,
+      "epoch": 2.71,
+      "learning_rate": 3.642857142857143e-05,
+      "loss": 1.4937,
       "step": 475
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 1.9720101781170485e-05,
-      "loss": 1.622,
+      "epoch": 2.72,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 1.4092,
       "step": 476
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 1.9656488549618322e-05,
-      "loss": 1.6244,
+      "epoch": 2.73,
+      "learning_rate": 3.637142857142857e-05,
+      "loss": 1.4851,
       "step": 477
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 1.959287531806616e-05,
-      "loss": 1.6003,
+      "epoch": 2.73,
+      "learning_rate": 3.6342857142857144e-05,
+      "loss": 1.4616,
       "step": 478
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 1.9529262086513996e-05,
-      "loss": 1.6027,
+      "epoch": 2.74,
+      "learning_rate": 3.631428571428572e-05,
+      "loss": 1.5117,
       "step": 479
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 1.9465648854961833e-05,
-      "loss": 1.6063,
+      "epoch": 2.74,
+      "learning_rate": 3.628571428571429e-05,
+      "loss": 1.5603,
       "step": 480
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 1.940203562340967e-05,
-      "loss": 1.5458,
+      "epoch": 2.75,
+      "learning_rate": 3.625714285714286e-05,
+      "loss": 1.4146,
       "step": 481
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 1.9338422391857507e-05,
-      "loss": 1.5434,
+      "epoch": 2.75,
+      "learning_rate": 3.6228571428571425e-05,
+      "loss": 1.4809,
       "step": 482
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 1.9274809160305344e-05,
-      "loss": 1.5534,
+      "epoch": 2.76,
+      "learning_rate": 3.62e-05,
+      "loss": 1.4996,
       "step": 483
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 1.921119592875318e-05,
-      "loss": 1.4813,
+      "epoch": 2.77,
+      "learning_rate": 3.617142857142857e-05,
+      "loss": 1.4392,
       "step": 484
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 1.914758269720102e-05,
-      "loss": 1.5318,
+      "epoch": 2.77,
+      "learning_rate": 3.6142857142857146e-05,
+      "loss": 1.4901,
       "step": 485
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 1.9083969465648855e-05,
-      "loss": 1.5503,
+      "epoch": 2.78,
+      "learning_rate": 3.611428571428572e-05,
+      "loss": 1.5351,
       "step": 486
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 1.9020356234096693e-05,
-      "loss": 1.6099,
+      "epoch": 2.78,
+      "learning_rate": 3.608571428571429e-05,
+      "loss": 1.4808,
       "step": 487
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 1.895674300254453e-05,
-      "loss": 1.5531,
+      "epoch": 2.79,
+      "learning_rate": 3.605714285714286e-05,
+      "loss": 1.5067,
       "step": 488
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 1.8893129770992367e-05,
-      "loss": 1.5838,
+      "epoch": 2.79,
+      "learning_rate": 3.602857142857143e-05,
+      "loss": 1.488,
       "step": 489
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 1.8829516539440204e-05,
-      "loss": 1.5685,
+      "epoch": 2.8,
+      "learning_rate": 3.6e-05,
+      "loss": 1.5572,
       "step": 490
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 1.876590330788804e-05,
-      "loss": 1.5997,
+      "epoch": 2.81,
+      "learning_rate": 3.5971428571428575e-05,
+      "loss": 1.467,
       "step": 491
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.8702290076335878e-05,
-      "loss": 1.5092,
+      "epoch": 2.81,
+      "learning_rate": 3.594285714285714e-05,
+      "loss": 1.4395,
       "step": 492
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.8638676844783715e-05,
-      "loss": 1.4721,
+      "epoch": 2.82,
+      "learning_rate": 3.5914285714285715e-05,
+      "loss": 1.4298,
       "step": 493
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.8575063613231552e-05,
-      "loss": 1.5468,
+      "epoch": 2.82,
+      "learning_rate": 3.588571428571429e-05,
+      "loss": 1.4397,
       "step": 494
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.851145038167939e-05,
-      "loss": 1.6105,
+      "epoch": 2.83,
+      "learning_rate": 3.585714285714286e-05,
+      "loss": 1.4718,
       "step": 495
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.844783715012723e-05,
-      "loss": 1.6517,
+      "epoch": 2.83,
+      "learning_rate": 3.582857142857143e-05,
+      "loss": 1.4222,
       "step": 496
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.8384223918575063e-05,
-      "loss": 1.5767,
+      "epoch": 2.84,
+      "learning_rate": 3.58e-05,
+      "loss": 1.419,
       "step": 497
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.83206106870229e-05,
-      "loss": 1.6751,
+      "epoch": 2.85,
+      "learning_rate": 3.577142857142857e-05,
+      "loss": 1.3971,
       "step": 498
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.825699745547074e-05,
-      "loss": 1.6003,
+      "epoch": 2.85,
+      "learning_rate": 3.574285714285714e-05,
+      "loss": 1.4151,
       "step": 499
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.8193384223918574e-05,
-      "loss": 1.5979,
+      "epoch": 2.86,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 1.4268,
       "step": 500
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.812977099236641e-05,
-      "loss": 1.6409,
+      "epoch": 2.86,
+      "learning_rate": 3.568571428571429e-05,
+      "loss": 1.494,
       "step": 501
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 1.8066157760814252e-05,
-      "loss": 1.5534,
+      "epoch": 2.87,
+      "learning_rate": 3.5657142857142864e-05,
+      "loss": 1.457,
       "step": 502
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 1.800254452926209e-05,
-      "loss": 1.717,
+      "epoch": 2.87,
+      "learning_rate": 3.562857142857143e-05,
+      "loss": 1.5671,
       "step": 503
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 1.7938931297709923e-05,
-      "loss": 1.458,
+      "epoch": 2.88,
+      "learning_rate": 3.56e-05,
+      "loss": 1.5733,
       "step": 504
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 1.7875318066157763e-05,
-      "loss": 1.5411,
+      "epoch": 2.89,
+      "learning_rate": 3.557142857142857e-05,
+      "loss": 1.4372,
       "step": 505
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 1.78117048346056e-05,
-      "loss": 1.626,
+      "epoch": 2.89,
+      "learning_rate": 3.5542857142857145e-05,
+      "loss": 1.5491,
       "step": 506
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 1.7748091603053434e-05,
-      "loss": 1.5568,
+      "epoch": 2.9,
+      "learning_rate": 3.551428571428572e-05,
+      "loss": 1.4364,
       "step": 507
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 1.7684478371501274e-05,
-      "loss": 1.5211,
+      "epoch": 2.9,
+      "learning_rate": 3.5485714285714286e-05,
+      "loss": 1.5118,
       "step": 508
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 1.762086513994911e-05,
-      "loss": 1.5756,
+      "epoch": 2.91,
+      "learning_rate": 3.545714285714286e-05,
+      "loss": 1.3832,
       "step": 509
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 1.7557251908396945e-05,
-      "loss": 1.4589,
+      "epoch": 2.91,
+      "learning_rate": 3.5428571428571426e-05,
+      "loss": 1.4804,
       "step": 510
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 1.7493638676844786e-05,
-      "loss": 1.5143,
+      "epoch": 2.92,
+      "learning_rate": 3.54e-05,
+      "loss": 1.4572,
       "step": 511
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 1.7430025445292623e-05,
-      "loss": 1.6454,
+      "epoch": 2.93,
+      "learning_rate": 3.5371428571428574e-05,
+      "loss": 1.4794,
       "step": 512
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.736641221374046e-05,
-      "loss": 1.6771,
+      "epoch": 2.93,
+      "learning_rate": 3.534285714285715e-05,
+      "loss": 1.4656,
       "step": 513
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.7302798982188297e-05,
-      "loss": 1.604,
+      "epoch": 2.94,
+      "learning_rate": 3.5314285714285714e-05,
+      "loss": 1.4926,
       "step": 514
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.7239185750636134e-05,
-      "loss": 1.4414,
+      "epoch": 2.94,
+      "learning_rate": 3.528571428571429e-05,
+      "loss": 1.528,
       "step": 515
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.717557251908397e-05,
-      "loss": 1.5744,
+      "epoch": 2.95,
+      "learning_rate": 3.525714285714286e-05,
+      "loss": 1.5246,
       "step": 516
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.7111959287531808e-05,
-      "loss": 1.6602,
+      "epoch": 2.95,
+      "learning_rate": 3.5228571428571435e-05,
+      "loss": 1.4466,
       "step": 517
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 1.7048346055979645e-05,
-      "loss": 1.6471,
+      "epoch": 2.96,
+      "learning_rate": 3.52e-05,
+      "loss": 1.4569,
       "step": 518
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 1.6984732824427482e-05,
-      "loss": 1.5913,
+      "epoch": 2.97,
+      "learning_rate": 3.517142857142857e-05,
+      "loss": 1.5635,
       "step": 519
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 1.692111959287532e-05,
-      "loss": 1.5145,
+      "epoch": 2.97,
+      "learning_rate": 3.514285714285714e-05,
+      "loss": 1.4915,
       "step": 520
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.6857506361323156e-05,
-      "loss": 1.5259,
+      "epoch": 2.98,
+      "learning_rate": 3.5114285714285716e-05,
+      "loss": 1.5432,
       "step": 521
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.6793893129770993e-05,
-      "loss": 1.7474,
+      "epoch": 2.98,
+      "learning_rate": 3.508571428571429e-05,
+      "loss": 1.5036,
       "step": 522
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 1.673027989821883e-05,
-      "loss": 1.5921,
+      "epoch": 2.99,
+      "learning_rate": 3.505714285714286e-05,
+      "loss": 1.4524,
       "step": 523
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 1.6666666666666667e-05,
-      "loss": 1.6021,
+      "epoch": 2.99,
+      "learning_rate": 3.502857142857143e-05,
+      "loss": 1.4675,
       "step": 524
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 1.6603053435114505e-05,
-      "loss": 1.5225,
+      "epoch": 3.0,
+      "learning_rate": 3.5e-05,
+      "loss": 1.3013,
       "step": 525
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 1.653944020356234e-05,
-      "loss": 1.5166,
+      "epoch": 3.01,
+      "learning_rate": 3.497142857142857e-05,
+      "loss": 1.4759,
       "step": 526
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 1.647582697201018e-05,
-      "loss": 1.5947,
+      "epoch": 3.01,
+      "learning_rate": 3.4942857142857144e-05,
+      "loss": 1.4916,
       "step": 527
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 1.6412213740458016e-05,
-      "loss": 1.4637,
+      "epoch": 3.02,
+      "learning_rate": 3.491428571428572e-05,
+      "loss": 1.3988,
       "step": 528
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 1.6348600508905853e-05,
-      "loss": 1.6541,
+      "epoch": 3.02,
+      "learning_rate": 3.488571428571429e-05,
+      "loss": 1.4334,
       "step": 529
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 1.628498727735369e-05,
-      "loss": 1.5789,
+      "epoch": 3.03,
+      "learning_rate": 3.485714285714286e-05,
+      "loss": 1.439,
       "step": 530
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 1.6221374045801527e-05,
-      "loss": 1.5832,
+      "epoch": 3.03,
+      "learning_rate": 3.482857142857143e-05,
+      "loss": 1.4937,
       "step": 531
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 1.6157760814249364e-05,
-      "loss": 1.5627,
+      "epoch": 3.04,
+      "learning_rate": 3.48e-05,
+      "loss": 1.3822,
       "step": 532
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 1.60941475826972e-05,
-      "loss": 1.4415,
+      "epoch": 3.05,
+      "learning_rate": 3.477142857142857e-05,
+      "loss": 1.4023,
       "step": 533
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 1.6030534351145038e-05,
-      "loss": 1.5264,
+      "epoch": 3.05,
+      "learning_rate": 3.4742857142857146e-05,
+      "loss": 1.3926,
       "step": 534
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 1.5966921119592875e-05,
-      "loss": 1.522,
+      "epoch": 3.06,
+      "learning_rate": 3.471428571428571e-05,
+      "loss": 1.3613,
       "step": 535
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 1.5903307888040712e-05,
-      "loss": 1.4428,
+      "epoch": 3.06,
+      "learning_rate": 3.468571428571429e-05,
+      "loss": 1.3861,
       "step": 536
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 1.583969465648855e-05,
-      "loss": 1.4254,
+      "epoch": 3.07,
+      "learning_rate": 3.465714285714286e-05,
+      "loss": 1.3687,
       "step": 537
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 1.5776081424936386e-05,
-      "loss": 1.4867,
+      "epoch": 3.07,
+      "learning_rate": 3.4628571428571434e-05,
+      "loss": 1.4331,
       "step": 538
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 1.5712468193384224e-05,
-      "loss": 1.5695,
+      "epoch": 3.08,
+      "learning_rate": 3.46e-05,
+      "loss": 1.3683,
       "step": 539
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 1.5648854961832064e-05,
-      "loss": 1.4549,
+      "epoch": 3.09,
+      "learning_rate": 3.4571428571428574e-05,
+      "loss": 1.4502,
       "step": 540
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 1.5585241730279898e-05,
-      "loss": 1.5384,
+      "epoch": 3.09,
+      "learning_rate": 3.454285714285714e-05,
+      "loss": 1.3738,
       "step": 541
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 1.5521628498727735e-05,
-      "loss": 1.5204,
+      "epoch": 3.1,
+      "learning_rate": 3.4514285714285715e-05,
+      "loss": 1.4079,
       "step": 542
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 1.5458015267175575e-05,
-      "loss": 1.4854,
+      "epoch": 3.1,
+      "learning_rate": 3.448571428571429e-05,
+      "loss": 1.399,
       "step": 543
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 1.539440203562341e-05,
-      "loss": 1.4278,
+      "epoch": 3.11,
+      "learning_rate": 3.445714285714286e-05,
+      "loss": 1.4222,
       "step": 544
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 1.5330788804071246e-05,
-      "loss": 1.4553,
+      "epoch": 3.11,
+      "learning_rate": 3.442857142857143e-05,
+      "loss": 1.3754,
       "step": 545
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 1.5267175572519086e-05,
-      "loss": 1.563,
+      "epoch": 3.12,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 1.4712,
       "step": 546
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 1.5203562340966923e-05,
-      "loss": 1.6154,
+      "epoch": 3.13,
+      "learning_rate": 3.437142857142857e-05,
+      "loss": 1.4293,
       "step": 547
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 1.5139949109414759e-05,
-      "loss": 1.4265,
+      "epoch": 3.13,
+      "learning_rate": 3.434285714285714e-05,
+      "loss": 1.4502,
       "step": 548
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 1.5076335877862596e-05,
-      "loss": 1.4601,
+      "epoch": 3.14,
+      "learning_rate": 3.431428571428572e-05,
+      "loss": 1.4555,
       "step": 549
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 1.5012722646310435e-05,
-      "loss": 1.5103,
+      "epoch": 3.14,
+      "learning_rate": 3.428571428571429e-05,
+      "loss": 1.442,
       "step": 550
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 1.494910941475827e-05,
-      "loss": 1.4919,
+      "epoch": 3.15,
+      "learning_rate": 3.425714285714286e-05,
+      "loss": 1.4041,
       "step": 551
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 1.4885496183206107e-05,
-      "loss": 1.5718,
+      "epoch": 3.15,
+      "learning_rate": 3.422857142857143e-05,
+      "loss": 1.4795,
       "step": 552
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 1.4821882951653946e-05,
-      "loss": 1.4859,
+      "epoch": 3.16,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 1.4393,
       "step": 553
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 1.4758269720101781e-05,
-      "loss": 1.4904,
+      "epoch": 3.17,
+      "learning_rate": 3.417142857142857e-05,
+      "loss": 1.5046,
       "step": 554
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 1.4694656488549618e-05,
-      "loss": 1.5625,
+      "epoch": 3.17,
+      "learning_rate": 3.4142857142857145e-05,
+      "loss": 1.3706,
       "step": 555
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 1.4631043256997457e-05,
-      "loss": 1.5098,
+      "epoch": 3.18,
+      "learning_rate": 3.411428571428571e-05,
+      "loss": 1.4684,
       "step": 556
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 1.4567430025445294e-05,
-      "loss": 1.6042,
+      "epoch": 3.18,
+      "learning_rate": 3.4085714285714286e-05,
+      "loss": 1.3998,
       "step": 557
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 1.450381679389313e-05,
-      "loss": 1.553,
+      "epoch": 3.19,
+      "learning_rate": 3.405714285714286e-05,
+      "loss": 1.4217,
       "step": 558
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 1.4440203562340968e-05,
-      "loss": 1.6064,
+      "epoch": 3.19,
+      "learning_rate": 3.402857142857143e-05,
+      "loss": 1.4942,
       "step": 559
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 1.4376590330788805e-05,
-      "loss": 1.7289,
+      "epoch": 3.2,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.3545,
       "step": 560
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 1.431297709923664e-05,
-      "loss": 1.5271,
+      "epoch": 3.21,
+      "learning_rate": 3.397142857142857e-05,
+      "loss": 1.4337,
       "step": 561
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 1.424936386768448e-05,
-      "loss": 1.5183,
+      "epoch": 3.21,
+      "learning_rate": 3.394285714285714e-05,
+      "loss": 1.375,
       "step": 562
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 1.4185750636132317e-05,
-      "loss": 1.6272,
+      "epoch": 3.22,
+      "learning_rate": 3.3914285714285714e-05,
+      "loss": 1.3336,
       "step": 563
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 1.4122137404580155e-05,
-      "loss": 1.4932,
+      "epoch": 3.22,
+      "learning_rate": 3.388571428571429e-05,
+      "loss": 1.5026,
       "step": 564
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 1.4058524173027989e-05,
-      "loss": 1.474,
+      "epoch": 3.23,
+      "learning_rate": 3.385714285714286e-05,
+      "loss": 1.4207,
       "step": 565
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 1.3994910941475828e-05,
-      "loss": 1.5057,
+      "epoch": 3.23,
+      "learning_rate": 3.3828571428571435e-05,
+      "loss": 1.3992,
       "step": 566
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 1.3931297709923667e-05,
-      "loss": 1.465,
+      "epoch": 3.24,
+      "learning_rate": 3.38e-05,
+      "loss": 1.4199,
       "step": 567
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 1.38676844783715e-05,
-      "loss": 1.5127,
+      "epoch": 3.25,
+      "learning_rate": 3.377142857142857e-05,
+      "loss": 1.4919,
       "step": 568
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 1.3804071246819339e-05,
-      "loss": 1.521,
+      "epoch": 3.25,
+      "learning_rate": 3.374285714285714e-05,
+      "loss": 1.4955,
       "step": 569
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 1.3740458015267178e-05,
-      "loss": 1.5181,
+      "epoch": 3.26,
+      "learning_rate": 3.3714285714285716e-05,
+      "loss": 1.3187,
       "step": 570
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 1.3676844783715011e-05,
-      "loss": 1.5222,
+      "epoch": 3.26,
+      "learning_rate": 3.368571428571429e-05,
+      "loss": 1.4405,
       "step": 571
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 1.361323155216285e-05,
-      "loss": 1.5319,
+      "epoch": 3.27,
+      "learning_rate": 3.3657142857142856e-05,
+      "loss": 1.4199,
       "step": 572
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 1.3549618320610687e-05,
-      "loss": 1.3958,
+      "epoch": 3.27,
+      "learning_rate": 3.362857142857143e-05,
+      "loss": 1.4144,
       "step": 573
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 1.3486005089058526e-05,
-      "loss": 1.4805,
+      "epoch": 3.28,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 1.3605,
       "step": 574
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 1.3422391857506361e-05,
-      "loss": 1.4634,
+      "epoch": 3.29,
+      "learning_rate": 3.357142857142857e-05,
+      "loss": 1.4023,
       "step": 575
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 1.3358778625954198e-05,
-      "loss": 1.6046,
+      "epoch": 3.29,
+      "learning_rate": 3.3542857142857144e-05,
+      "loss": 1.4218,
       "step": 576
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 1.3295165394402037e-05,
-      "loss": 1.5216,
+      "epoch": 3.3,
+      "learning_rate": 3.351428571428572e-05,
+      "loss": 1.3927,
       "step": 577
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 1.3231552162849873e-05,
-      "loss": 1.5602,
+      "epoch": 3.3,
+      "learning_rate": 3.3485714285714285e-05,
+      "loss": 1.4504,
       "step": 578
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 1.316793893129771e-05,
-      "loss": 1.6176,
+      "epoch": 3.31,
+      "learning_rate": 3.345714285714286e-05,
+      "loss": 1.4313,
       "step": 579
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 1.3104325699745548e-05,
-      "loss": 1.5447,
+      "epoch": 3.31,
+      "learning_rate": 3.342857142857143e-05,
+      "loss": 1.3784,
       "step": 580
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 1.3040712468193386e-05,
-      "loss": 1.4158,
+      "epoch": 3.32,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 1.3691,
       "step": 581
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 1.2977099236641221e-05,
-      "loss": 1.4954,
+      "epoch": 3.33,
+      "learning_rate": 3.337142857142857e-05,
+      "loss": 1.389,
       "step": 582
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 1.291348600508906e-05,
-      "loss": 1.473,
+      "epoch": 3.33,
+      "learning_rate": 3.334285714285714e-05,
+      "loss": 1.4565,
       "step": 583
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 1.2849872773536897e-05,
-      "loss": 1.6029,
+      "epoch": 3.34,
+      "learning_rate": 3.331428571428571e-05,
+      "loss": 1.3677,
       "step": 584
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 1.2786259541984732e-05,
-      "loss": 1.4873,
+      "epoch": 3.34,
+      "learning_rate": 3.3285714285714286e-05,
+      "loss": 1.3421,
       "step": 585
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 1.2722646310432571e-05,
-      "loss": 1.5771,
+      "epoch": 3.35,
+      "learning_rate": 3.325714285714286e-05,
+      "loss": 1.2984,
       "step": 586
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 1.2659033078880408e-05,
-      "loss": 1.5967,
+      "epoch": 3.35,
+      "learning_rate": 3.3228571428571434e-05,
+      "loss": 1.3738,
       "step": 587
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 1.2595419847328243e-05,
-      "loss": 1.4916,
+      "epoch": 3.36,
+      "learning_rate": 3.32e-05,
+      "loss": 1.4382,
       "step": 588
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 1.2531806615776082e-05,
-      "loss": 1.4802,
+      "epoch": 3.37,
+      "learning_rate": 3.3171428571428574e-05,
+      "loss": 1.4113,
       "step": 589
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 1.2468193384223919e-05,
-      "loss": 1.4806,
+      "epoch": 3.37,
+      "learning_rate": 3.314285714285714e-05,
+      "loss": 1.3727,
       "step": 590
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 1.2404580152671756e-05,
-      "loss": 1.517,
+      "epoch": 3.38,
+      "learning_rate": 3.3114285714285715e-05,
+      "loss": 1.4042,
       "step": 591
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 1.2340966921119595e-05,
-      "loss": 1.5302,
+      "epoch": 3.38,
+      "learning_rate": 3.308571428571429e-05,
+      "loss": 1.4052,
       "step": 592
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 1.227735368956743e-05,
-      "loss": 1.4838,
+      "epoch": 3.39,
+      "learning_rate": 3.305714285714286e-05,
+      "loss": 1.3701,
       "step": 593
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 1.2213740458015267e-05,
-      "loss": 1.54,
+      "epoch": 3.39,
+      "learning_rate": 3.302857142857143e-05,
+      "loss": 1.4521,
       "step": 594
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 1.2150127226463104e-05,
-      "loss": 1.4704,
+      "epoch": 3.4,
+      "learning_rate": 3.3e-05,
+      "loss": 1.3066,
       "step": 595
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 1.2086513994910942e-05,
-      "loss": 1.4974,
+      "epoch": 3.41,
+      "learning_rate": 3.2971428571428576e-05,
+      "loss": 1.4493,
       "step": 596
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 1.202290076335878e-05,
-      "loss": 1.4962,
+      "epoch": 3.41,
+      "learning_rate": 3.294285714285714e-05,
+      "loss": 1.3907,
       "step": 597
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 1.1959287531806616e-05,
-      "loss": 1.6077,
+      "epoch": 3.42,
+      "learning_rate": 3.291428571428572e-05,
+      "loss": 1.3197,
       "step": 598
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 1.1895674300254453e-05,
-      "loss": 1.4532,
+      "epoch": 3.42,
+      "learning_rate": 3.2885714285714284e-05,
+      "loss": 1.4457,
       "step": 599
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 1.1832061068702292e-05,
-      "loss": 1.4745,
+      "epoch": 3.43,
+      "learning_rate": 3.285714285714286e-05,
+      "loss": 1.4287,
       "step": 600
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 1.1768447837150127e-05,
-      "loss": 1.6156,
+      "epoch": 3.43,
+      "learning_rate": 3.282857142857143e-05,
+      "loss": 1.3399,
       "step": 601
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 1.1704834605597966e-05,
-      "loss": 1.5023,
+      "epoch": 3.44,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 1.4595,
       "step": 602
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 1.1641221374045803e-05,
-      "loss": 1.5043,
+      "epoch": 3.45,
+      "learning_rate": 3.277142857142858e-05,
+      "loss": 1.3931,
       "step": 603
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 1.1577608142493638e-05,
-      "loss": 1.4825,
+      "epoch": 3.45,
+      "learning_rate": 3.2742857142857145e-05,
+      "loss": 1.413,
       "step": 604
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 1.1513994910941477e-05,
-      "loss": 1.5165,
+      "epoch": 3.46,
+      "learning_rate": 3.271428571428571e-05,
+      "loss": 1.4462,
       "step": 605
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 1.1450381679389314e-05,
-      "loss": 1.5332,
+      "epoch": 3.46,
+      "learning_rate": 3.2685714285714285e-05,
+      "loss": 1.3365,
       "step": 606
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 1.1386768447837151e-05,
-      "loss": 1.4058,
+      "epoch": 3.47,
+      "learning_rate": 3.265714285714286e-05,
+      "loss": 1.3412,
       "step": 607
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 1.1323155216284988e-05,
-      "loss": 1.5201,
+      "epoch": 3.47,
+      "learning_rate": 3.262857142857143e-05,
+      "loss": 1.4854,
       "step": 608
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 1.1259541984732823e-05,
-      "loss": 1.4773,
+      "epoch": 3.48,
+      "learning_rate": 3.26e-05,
+      "loss": 1.3158,
       "step": 609
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 1.1195928753180662e-05,
-      "loss": 1.609,
+      "epoch": 3.49,
+      "learning_rate": 3.257142857142857e-05,
+      "loss": 1.4656,
       "step": 610
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 1.11323155216285e-05,
-      "loss": 1.4419,
+      "epoch": 3.49,
+      "learning_rate": 3.254285714285715e-05,
+      "loss": 1.4103,
       "step": 611
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 1.1068702290076336e-05,
-      "loss": 1.5836,
+      "epoch": 3.5,
+      "learning_rate": 3.2514285714285714e-05,
+      "loss": 1.4259,
       "step": 612
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 1.1005089058524173e-05,
-      "loss": 1.6614,
+      "epoch": 3.5,
+      "learning_rate": 3.248571428571429e-05,
+      "loss": 1.4604,
       "step": 613
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 1.0941475826972012e-05,
-      "loss": 1.6177,
+      "epoch": 3.51,
+      "learning_rate": 3.245714285714286e-05,
+      "loss": 1.3915,
       "step": 614
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 1.0877862595419848e-05,
-      "loss": 1.4421,
+      "epoch": 3.51,
+      "learning_rate": 3.242857142857143e-05,
+      "loss": 1.3336,
       "step": 615
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 1.0814249363867685e-05,
-      "loss": 1.542,
+      "epoch": 3.52,
+      "learning_rate": 3.24e-05,
+      "loss": 1.4415,
       "step": 616
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 1.0750636132315522e-05,
-      "loss": 1.492,
+      "epoch": 3.53,
+      "learning_rate": 3.2371428571428575e-05,
+      "loss": 1.426,
       "step": 617
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 1.0687022900763359e-05,
-      "loss": 1.5336,
+      "epoch": 3.53,
+      "learning_rate": 3.234285714285715e-05,
+      "loss": 1.4257,
       "step": 618
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 1.0623409669211198e-05,
-      "loss": 1.5533,
+      "epoch": 3.54,
+      "learning_rate": 3.2314285714285716e-05,
+      "loss": 1.392,
       "step": 619
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 1.0559796437659033e-05,
-      "loss": 1.6151,
+      "epoch": 3.54,
+      "learning_rate": 3.228571428571428e-05,
+      "loss": 1.3406,
       "step": 620
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 1.049618320610687e-05,
-      "loss": 1.5709,
+      "epoch": 3.55,
+      "learning_rate": 3.2257142857142856e-05,
+      "loss": 1.4017,
       "step": 621
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 1.0432569974554709e-05,
-      "loss": 1.5086,
+      "epoch": 3.55,
+      "learning_rate": 3.222857142857143e-05,
+      "loss": 1.4736,
       "step": 622
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 1.0368956743002544e-05,
-      "loss": 1.501,
+      "epoch": 3.56,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 1.3957,
       "step": 623
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 1.0305343511450383e-05,
-      "loss": 1.5721,
+      "epoch": 3.57,
+      "learning_rate": 3.217142857142858e-05,
+      "loss": 1.4527,
       "step": 624
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 1.024173027989822e-05,
-      "loss": 1.6042,
+      "epoch": 3.57,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 1.4024,
       "step": 625
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 1.0178117048346055e-05,
-      "loss": 1.4864,
+      "epoch": 3.58,
+      "learning_rate": 3.211428571428571e-05,
+      "loss": 1.3226,
       "step": 626
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 1.0114503816793894e-05,
-      "loss": 1.5107,
+      "epoch": 3.58,
+      "learning_rate": 3.2085714285714284e-05,
+      "loss": 1.4371,
       "step": 627
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 1.0050890585241731e-05,
-      "loss": 1.5386,
+      "epoch": 3.59,
+      "learning_rate": 3.205714285714286e-05,
+      "loss": 1.3084,
       "step": 628
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 9.987277353689568e-06,
-      "loss": 1.5453,
+      "epoch": 3.59,
+      "learning_rate": 3.202857142857143e-05,
+      "loss": 1.4082,
       "step": 629
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 9.923664122137405e-06,
-      "loss": 1.51,
+      "epoch": 3.6,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.3619,
       "step": 630
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 9.860050890585242e-06,
-      "loss": 1.5821,
+      "epoch": 3.61,
+      "learning_rate": 3.197142857142857e-05,
+      "loss": 1.3552,
       "step": 631
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 9.79643765903308e-06,
-      "loss": 1.5815,
+      "epoch": 3.61,
+      "learning_rate": 3.1942857142857146e-05,
+      "loss": 1.3593,
       "step": 632
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 9.732824427480917e-06,
-      "loss": 1.5071,
+      "epoch": 3.62,
+      "learning_rate": 3.191428571428571e-05,
+      "loss": 1.3934,
       "step": 633
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 9.669211195928754e-06,
-      "loss": 1.5823,
+      "epoch": 3.62,
+      "learning_rate": 3.1885714285714286e-05,
+      "loss": 1.4082,
       "step": 634
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 9.60559796437659e-06,
-      "loss": 1.4497,
+      "epoch": 3.63,
+      "learning_rate": 3.185714285714286e-05,
+      "loss": 1.2599,
       "step": 635
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 9.541984732824428e-06,
-      "loss": 1.5951,
+      "epoch": 3.63,
+      "learning_rate": 3.182857142857143e-05,
+      "loss": 1.5062,
       "step": 636
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 9.478371501272265e-06,
-      "loss": 1.5335,
+      "epoch": 3.64,
+      "learning_rate": 3.18e-05,
+      "loss": 1.4426,
       "step": 637
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 9.414758269720102e-06,
-      "loss": 1.5423,
+      "epoch": 3.65,
+      "learning_rate": 3.1771428571428574e-05,
+      "loss": 1.41,
       "step": 638
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 9.351145038167939e-06,
-      "loss": 1.5326,
+      "epoch": 3.65,
+      "learning_rate": 3.174285714285715e-05,
+      "loss": 1.4143,
       "step": 639
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 9.287531806615776e-06,
-      "loss": 1.3375,
+      "epoch": 3.66,
+      "learning_rate": 3.1714285714285715e-05,
+      "loss": 1.3948,
       "step": 640
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 9.223918575063615e-06,
-      "loss": 1.323,
+      "epoch": 3.66,
+      "learning_rate": 3.168571428571429e-05,
+      "loss": 1.3897,
       "step": 641
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 9.16030534351145e-06,
-      "loss": 1.4634,
+      "epoch": 3.67,
+      "learning_rate": 3.1657142857142855e-05,
+      "loss": 1.459,
       "step": 642
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 9.096692111959287e-06,
-      "loss": 1.426,
+      "epoch": 3.67,
+      "learning_rate": 3.162857142857143e-05,
+      "loss": 1.4066,
       "step": 643
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 9.033078880407126e-06,
-      "loss": 1.65,
+      "epoch": 3.68,
+      "learning_rate": 3.16e-05,
+      "loss": 1.4348,
       "step": 644
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 8.969465648854961e-06,
-      "loss": 1.485,
+      "epoch": 3.69,
+      "learning_rate": 3.1571428571428576e-05,
+      "loss": 1.3857,
       "step": 645
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 8.9058524173028e-06,
-      "loss": 1.6655,
+      "epoch": 3.69,
+      "learning_rate": 3.154285714285714e-05,
+      "loss": 1.3381,
       "step": 646
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 8.842239185750637e-06,
-      "loss": 1.372,
+      "epoch": 3.7,
+      "learning_rate": 3.1514285714285717e-05,
+      "loss": 1.4283,
       "step": 647
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 8.778625954198473e-06,
-      "loss": 1.5437,
+      "epoch": 3.7,
+      "learning_rate": 3.148571428571428e-05,
+      "loss": 1.4544,
       "step": 648
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 8.715012722646311e-06,
-      "loss": 1.5228,
+      "epoch": 3.71,
+      "learning_rate": 3.145714285714286e-05,
+      "loss": 1.434,
       "step": 649
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 8.651399491094148e-06,
-      "loss": 1.3835,
+      "epoch": 3.71,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 1.4212,
       "step": 650
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 8.587786259541985e-06,
-      "loss": 1.4312,
+      "epoch": 3.72,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 1.3627,
       "step": 651
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 8.524173027989823e-06,
-      "loss": 1.5355,
+      "epoch": 3.73,
+      "learning_rate": 3.137142857142857e-05,
+      "loss": 1.4421,
       "step": 652
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 8.46055979643766e-06,
-      "loss": 1.3893,
+      "epoch": 3.73,
+      "learning_rate": 3.1342857142857145e-05,
+      "loss": 1.4152,
       "step": 653
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 8.396946564885497e-06,
-      "loss": 1.3776,
+      "epoch": 3.74,
+      "learning_rate": 3.131428571428572e-05,
+      "loss": 1.2664,
       "step": 654
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 8.333333333333334e-06,
-      "loss": 1.4967,
+      "epoch": 3.74,
+      "learning_rate": 3.1285714285714285e-05,
+      "loss": 1.44,
       "step": 655
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 8.26972010178117e-06,
-      "loss": 1.3751,
+      "epoch": 3.75,
+      "learning_rate": 3.125714285714286e-05,
+      "loss": 1.3987,
       "step": 656
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 8.206106870229008e-06,
-      "loss": 1.5909,
+      "epoch": 3.75,
+      "learning_rate": 3.122857142857143e-05,
+      "loss": 1.4801,
       "step": 657
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 8.142493638676845e-06,
-      "loss": 1.6396,
+      "epoch": 3.76,
+      "learning_rate": 3.12e-05,
+      "loss": 1.292,
       "step": 658
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 8.078880407124682e-06,
-      "loss": 1.5689,
+      "epoch": 3.77,
+      "learning_rate": 3.117142857142857e-05,
+      "loss": 1.3902,
       "step": 659
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 8.015267175572519e-06,
-      "loss": 1.4298,
+      "epoch": 3.77,
+      "learning_rate": 3.114285714285715e-05,
+      "loss": 1.4487,
       "step": 660
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 7.951653944020356e-06,
-      "loss": 1.4689,
+      "epoch": 3.78,
+      "learning_rate": 3.111428571428572e-05,
+      "loss": 1.3712,
       "step": 661
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 7.888040712468193e-06,
-      "loss": 1.6027,
+      "epoch": 3.78,
+      "learning_rate": 3.108571428571429e-05,
+      "loss": 1.3813,
       "step": 662
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 7.824427480916032e-06,
-      "loss": 1.4128,
+      "epoch": 3.79,
+      "learning_rate": 3.1057142857142854e-05,
+      "loss": 1.3506,
       "step": 663
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 7.760814249363867e-06,
-      "loss": 1.5457,
+      "epoch": 3.79,
+      "learning_rate": 3.102857142857143e-05,
+      "loss": 1.3984,
       "step": 664
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 7.697201017811704e-06,
-      "loss": 1.5069,
+      "epoch": 3.8,
+      "learning_rate": 3.1e-05,
+      "loss": 1.465,
       "step": 665
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 7.633587786259543e-06,
-      "loss": 1.5511,
+      "epoch": 3.81,
+      "learning_rate": 3.0971428571428575e-05,
+      "loss": 1.3658,
       "step": 666
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 7.569974554707379e-06,
-      "loss": 1.4441,
+      "epoch": 3.81,
+      "learning_rate": 3.094285714285715e-05,
+      "loss": 1.372,
       "step": 667
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 7.506361323155217e-06,
-      "loss": 1.4732,
+      "epoch": 3.82,
+      "learning_rate": 3.0914285714285715e-05,
+      "loss": 1.3928,
       "step": 668
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 7.4427480916030536e-06,
-      "loss": 1.479,
+      "epoch": 3.82,
+      "learning_rate": 3.088571428571428e-05,
+      "loss": 1.406,
       "step": 669
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 7.379134860050891e-06,
-      "loss": 1.4266,
+      "epoch": 3.83,
+      "learning_rate": 3.0857142857142856e-05,
+      "loss": 1.4114,
       "step": 670
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 7.3155216284987285e-06,
-      "loss": 1.5745,
+      "epoch": 3.83,
+      "learning_rate": 3.082857142857143e-05,
+      "loss": 1.4176,
       "step": 671
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 7.251908396946565e-06,
-      "loss": 1.3644,
+      "epoch": 3.84,
+      "learning_rate": 3.08e-05,
+      "loss": 1.454,
       "step": 672
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 7.188295165394403e-06,
-      "loss": 1.4786,
+      "epoch": 3.85,
+      "learning_rate": 3.077142857142857e-05,
+      "loss": 1.3545,
       "step": 673
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 7.12468193384224e-06,
-      "loss": 1.5273,
+      "epoch": 3.85,
+      "learning_rate": 3.0742857142857144e-05,
+      "loss": 1.3091,
       "step": 674
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 7.061068702290078e-06,
-      "loss": 1.4906,
+      "epoch": 3.86,
+      "learning_rate": 3.071428571428572e-05,
+      "loss": 1.3078,
       "step": 675
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 6.997455470737914e-06,
-      "loss": 1.4297,
+      "epoch": 3.86,
+      "learning_rate": 3.068571428571429e-05,
+      "loss": 1.3847,
       "step": 676
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 6.93384223918575e-06,
-      "loss": 1.5863,
+      "epoch": 3.87,
+      "learning_rate": 3.065714285714286e-05,
+      "loss": 1.4205,
       "step": 677
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 6.870229007633589e-06,
-      "loss": 1.5445,
+      "epoch": 3.87,
+      "learning_rate": 3.062857142857143e-05,
+      "loss": 1.4477,
       "step": 678
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 6.806615776081425e-06,
-      "loss": 1.4213,
+      "epoch": 3.88,
+      "learning_rate": 3.06e-05,
+      "loss": 1.4563,
       "step": 679
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 6.743002544529263e-06,
-      "loss": 1.4672,
+      "epoch": 3.89,
+      "learning_rate": 3.057142857142857e-05,
+      "loss": 1.3703,
       "step": 680
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 6.679389312977099e-06,
-      "loss": 1.5799,
+      "epoch": 3.89,
+      "learning_rate": 3.0542857142857146e-05,
+      "loss": 1.3799,
       "step": 681
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 6.615776081424936e-06,
-      "loss": 1.5335,
+      "epoch": 3.9,
+      "learning_rate": 3.0514285714285716e-05,
+      "loss": 1.3707,
       "step": 682
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 6.552162849872774e-06,
-      "loss": 1.4691,
+      "epoch": 3.9,
+      "learning_rate": 3.048571428571429e-05,
+      "loss": 1.4205,
       "step": 683
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 6.4885496183206104e-06,
-      "loss": 1.504,
+      "epoch": 3.91,
+      "learning_rate": 3.0457142857142856e-05,
+      "loss": 1.4544,
       "step": 684
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 6.424936386768448e-06,
-      "loss": 1.4863,
+      "epoch": 3.91,
+      "learning_rate": 3.042857142857143e-05,
+      "loss": 1.3434,
       "step": 685
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 6.3613231552162854e-06,
-      "loss": 1.4737,
+      "epoch": 3.92,
+      "learning_rate": 3.04e-05,
+      "loss": 1.3884,
       "step": 686
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 6.297709923664122e-06,
-      "loss": 1.6219,
+      "epoch": 3.93,
+      "learning_rate": 3.0371428571428574e-05,
+      "loss": 1.3845,
       "step": 687
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 6.2340966921119596e-06,
-      "loss": 1.5618,
+      "epoch": 3.93,
+      "learning_rate": 3.0342857142857144e-05,
+      "loss": 1.4193,
       "step": 688
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 6.1704834605597975e-06,
-      "loss": 1.523,
+      "epoch": 3.94,
+      "learning_rate": 3.0314285714285718e-05,
+      "loss": 1.3968,
       "step": 689
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 6.106870229007634e-06,
-      "loss": 1.4841,
+      "epoch": 3.94,
+      "learning_rate": 3.0285714285714288e-05,
+      "loss": 1.3406,
       "step": 690
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 6.043256997455471e-06,
-      "loss": 1.5883,
+      "epoch": 3.95,
+      "learning_rate": 3.0257142857142855e-05,
+      "loss": 1.4414,
       "step": 691
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 5.979643765903308e-06,
-      "loss": 1.5221,
+      "epoch": 3.95,
+      "learning_rate": 3.022857142857143e-05,
+      "loss": 1.3637,
       "step": 692
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 5.916030534351146e-06,
-      "loss": 1.4583,
+      "epoch": 3.96,
+      "learning_rate": 3.02e-05,
+      "loss": 1.4607,
       "step": 693
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 5.852417302798983e-06,
-      "loss": 1.6034,
+      "epoch": 3.97,
+      "learning_rate": 3.0171428571428572e-05,
+      "loss": 1.4028,
       "step": 694
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 5.788804071246819e-06,
-      "loss": 1.4732,
+      "epoch": 3.97,
+      "learning_rate": 3.0142857142857146e-05,
+      "loss": 1.3849,
       "step": 695
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 5.725190839694657e-06,
-      "loss": 1.5974,
+      "epoch": 3.98,
+      "learning_rate": 3.0114285714285716e-05,
+      "loss": 1.3121,
       "step": 696
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 5.661577608142494e-06,
-      "loss": 1.5268,
+      "epoch": 3.98,
+      "learning_rate": 3.008571428571429e-05,
+      "loss": 1.3836,
       "step": 697
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 5.597964376590331e-06,
-      "loss": 1.4549,
+      "epoch": 3.99,
+      "learning_rate": 3.0057142857142857e-05,
+      "loss": 1.3917,
       "step": 698
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 5.534351145038168e-06,
-      "loss": 1.5012,
+      "epoch": 3.99,
+      "learning_rate": 3.0028571428571427e-05,
+      "loss": 1.3586,
       "step": 699
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 5.470737913486006e-06,
-      "loss": 1.4293,
+      "epoch": 4.0,
+      "learning_rate": 3e-05,
+      "loss": 1.3215,
       "step": 700
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 5.407124681933842e-06,
-      "loss": 1.6263,
+      "epoch": 4.01,
+      "learning_rate": 2.997142857142857e-05,
+      "loss": 1.3277,
       "step": 701
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 5.343511450381679e-06,
-      "loss": 1.542,
+      "epoch": 4.01,
+      "learning_rate": 2.9942857142857145e-05,
+      "loss": 1.3669,
       "step": 702
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 5.2798982188295165e-06,
-      "loss": 1.5289,
+      "epoch": 4.02,
+      "learning_rate": 2.9914285714285718e-05,
+      "loss": 1.3304,
       "step": 703
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 5.216284987277354e-06,
-      "loss": 1.4846,
+      "epoch": 4.02,
+      "learning_rate": 2.988571428571429e-05,
+      "loss": 1.3059,
       "step": 704
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 5.1526717557251914e-06,
-      "loss": 1.5461,
+      "epoch": 4.03,
+      "learning_rate": 2.9857142857142862e-05,
+      "loss": 1.3963,
       "step": 705
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 5.089058524173028e-06,
-      "loss": 1.516,
+      "epoch": 4.03,
+      "learning_rate": 2.982857142857143e-05,
+      "loss": 1.333,
       "step": 706
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 5.025445292620866e-06,
-      "loss": 1.5432,
+      "epoch": 4.04,
+      "learning_rate": 2.98e-05,
+      "loss": 1.2779,
       "step": 707
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 4.961832061068703e-06,
-      "loss": 1.4167,
+      "epoch": 4.05,
+      "learning_rate": 2.9771428571428573e-05,
+      "loss": 1.3823,
       "step": 708
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 4.89821882951654e-06,
-      "loss": 1.441,
+      "epoch": 4.05,
+      "learning_rate": 2.9742857142857143e-05,
+      "loss": 1.3253,
       "step": 709
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 4.834605597964377e-06,
-      "loss": 1.547,
+      "epoch": 4.06,
+      "learning_rate": 2.9714285714285717e-05,
+      "loss": 1.372,
       "step": 710
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 4.770992366412214e-06,
-      "loss": 1.5322,
+      "epoch": 4.06,
+      "learning_rate": 2.968571428571429e-05,
+      "loss": 1.3313,
       "step": 711
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 4.707379134860051e-06,
-      "loss": 1.4142,
+      "epoch": 4.07,
+      "learning_rate": 2.965714285714286e-05,
+      "loss": 1.3027,
       "step": 712
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 4.643765903307888e-06,
-      "loss": 1.4593,
+      "epoch": 4.07,
+      "learning_rate": 2.9628571428571428e-05,
+      "loss": 1.3682,
       "step": 713
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 4.580152671755725e-06,
-      "loss": 1.5256,
+      "epoch": 4.08,
+      "learning_rate": 2.96e-05,
+      "loss": 1.3231,
       "step": 714
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 4.516539440203563e-06,
-      "loss": 1.5131,
+      "epoch": 4.09,
+      "learning_rate": 2.957142857142857e-05,
+      "loss": 1.3922,
       "step": 715
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 4.4529262086514e-06,
-      "loss": 1.474,
+      "epoch": 4.09,
+      "learning_rate": 2.9542857142857145e-05,
+      "loss": 1.3762,
       "step": 716
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 4.389312977099236e-06,
-      "loss": 1.4454,
+      "epoch": 4.1,
+      "learning_rate": 2.9514285714285715e-05,
+      "loss": 1.4071,
       "step": 717
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 4.325699745547074e-06,
-      "loss": 1.6011,
+      "epoch": 4.1,
+      "learning_rate": 2.948571428571429e-05,
+      "loss": 1.3917,
       "step": 718
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 4.262086513994911e-06,
-      "loss": 1.608,
+      "epoch": 4.11,
+      "learning_rate": 2.9457142857142863e-05,
+      "loss": 1.33,
       "step": 719
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 4.198473282442748e-06,
-      "loss": 1.4159,
+      "epoch": 4.11,
+      "learning_rate": 2.9428571428571426e-05,
+      "loss": 1.3685,
       "step": 720
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 4.134860050890585e-06,
-      "loss": 1.4697,
+      "epoch": 4.12,
+      "learning_rate": 2.94e-05,
+      "loss": 1.299,
       "step": 721
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 4.0712468193384225e-06,
-      "loss": 1.4841,
+      "epoch": 4.13,
+      "learning_rate": 2.9371428571428573e-05,
+      "loss": 1.4026,
       "step": 722
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 4.0076335877862595e-06,
-      "loss": 1.508,
+      "epoch": 4.13,
+      "learning_rate": 2.9342857142857144e-05,
+      "loss": 1.3235,
       "step": 723
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 3.944020356234097e-06,
-      "loss": 1.4807,
+      "epoch": 4.14,
+      "learning_rate": 2.9314285714285717e-05,
+      "loss": 1.3566,
       "step": 724
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 3.880407124681934e-06,
-      "loss": 1.4928,
+      "epoch": 4.14,
+      "learning_rate": 2.9285714285714288e-05,
+      "loss": 1.3508,
       "step": 725
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 3.816793893129772e-06,
-      "loss": 1.5036,
+      "epoch": 4.15,
+      "learning_rate": 2.925714285714286e-05,
+      "loss": 1.3755,
       "step": 726
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 3.7531806615776087e-06,
-      "loss": 1.4223,
+      "epoch": 4.15,
+      "learning_rate": 2.9228571428571428e-05,
+      "loss": 1.3036,
       "step": 727
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 3.6895674300254453e-06,
-      "loss": 1.6322,
+      "epoch": 4.16,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 1.3768,
       "step": 728
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 3.6259541984732824e-06,
-      "loss": 1.4899,
+      "epoch": 4.17,
+      "learning_rate": 2.9171428571428572e-05,
+      "loss": 1.4108,
       "step": 729
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 3.56234096692112e-06,
-      "loss": 1.5485,
+      "epoch": 4.17,
+      "learning_rate": 2.9142857142857146e-05,
+      "loss": 1.2796,
       "step": 730
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 3.498727735368957e-06,
-      "loss": 1.4896,
+      "epoch": 4.18,
+      "learning_rate": 2.9114285714285716e-05,
+      "loss": 1.365,
       "step": 731
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 3.4351145038167944e-06,
-      "loss": 1.5388,
+      "epoch": 4.18,
+      "learning_rate": 2.908571428571429e-05,
+      "loss": 1.3352,
       "step": 732
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 3.3715012722646315e-06,
-      "loss": 1.5161,
+      "epoch": 4.19,
+      "learning_rate": 2.905714285714286e-05,
+      "loss": 1.306,
       "step": 733
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 3.307888040712468e-06,
-      "loss": 1.5903,
+      "epoch": 4.19,
+      "learning_rate": 2.9028571428571427e-05,
+      "loss": 1.3322,
       "step": 734
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 3.2442748091603052e-06,
-      "loss": 1.5086,
+      "epoch": 4.2,
+      "learning_rate": 2.9e-05,
+      "loss": 1.2859,
       "step": 735
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 3.1806615776081427e-06,
-      "loss": 1.5234,
+      "epoch": 4.21,
+      "learning_rate": 2.897142857142857e-05,
+      "loss": 1.3707,
       "step": 736
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 3.1170483460559798e-06,
-      "loss": 1.4371,
+      "epoch": 4.21,
+      "learning_rate": 2.8942857142857144e-05,
+      "loss": 1.3309,
       "step": 737
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 3.053435114503817e-06,
-      "loss": 1.4445,
+      "epoch": 4.22,
+      "learning_rate": 2.8914285714285714e-05,
+      "loss": 1.2728,
       "step": 738
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 2.989821882951654e-06,
-      "loss": 1.4871,
+      "epoch": 4.22,
+      "learning_rate": 2.8885714285714288e-05,
+      "loss": 1.3801,
       "step": 739
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 2.9262086513994914e-06,
-      "loss": 1.4505,
+      "epoch": 4.23,
+      "learning_rate": 2.885714285714286e-05,
+      "loss": 1.3491,
       "step": 740
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 2.8625954198473285e-06,
-      "loss": 1.4307,
+      "epoch": 4.23,
+      "learning_rate": 2.8828571428571432e-05,
+      "loss": 1.3162,
       "step": 741
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 2.7989821882951656e-06,
-      "loss": 1.5002,
+      "epoch": 4.24,
+      "learning_rate": 2.88e-05,
+      "loss": 1.2886,
       "step": 742
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 2.735368956743003e-06,
-      "loss": 1.4684,
+      "epoch": 4.25,
+      "learning_rate": 2.8771428571428572e-05,
+      "loss": 1.2986,
       "step": 743
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 2.6717557251908397e-06,
-      "loss": 1.4525,
+      "epoch": 4.25,
+      "learning_rate": 2.8742857142857143e-05,
+      "loss": 1.3365,
       "step": 744
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 2.608142493638677e-06,
-      "loss": 1.4024,
+      "epoch": 4.26,
+      "learning_rate": 2.8714285714285716e-05,
+      "loss": 1.4407,
       "step": 745
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 2.544529262086514e-06,
-      "loss": 1.4652,
+      "epoch": 4.26,
+      "learning_rate": 2.8685714285714286e-05,
+      "loss": 1.3204,
       "step": 746
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 2.4809160305343513e-06,
-      "loss": 1.4308,
+      "epoch": 4.27,
+      "learning_rate": 2.865714285714286e-05,
+      "loss": 1.3159,
       "step": 747
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 2.4173027989821884e-06,
-      "loss": 1.4592,
+      "epoch": 4.27,
+      "learning_rate": 2.8628571428571434e-05,
+      "loss": 1.2222,
       "step": 748
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 2.3536895674300255e-06,
-      "loss": 1.3863,
+      "epoch": 4.28,
+      "learning_rate": 2.86e-05,
+      "loss": 1.2952,
       "step": 749
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 2.2900763358778625e-06,
-      "loss": 1.4913,
+      "epoch": 4.29,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 1.3046,
       "step": 750
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 2.2264631043257e-06,
-      "loss": 1.5188,
+      "epoch": 4.29,
+      "learning_rate": 2.8542857142857144e-05,
+      "loss": 1.3444,
       "step": 751
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 2.162849872773537e-06,
-      "loss": 1.4573,
+      "epoch": 4.3,
+      "learning_rate": 2.8514285714285715e-05,
+      "loss": 1.4729,
       "step": 752
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 2.099236641221374e-06,
-      "loss": 1.5268,
+      "epoch": 4.3,
+      "learning_rate": 2.848571428571429e-05,
+      "loss": 1.3795,
       "step": 753
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 2.0356234096692112e-06,
-      "loss": 1.6501,
+      "epoch": 4.31,
+      "learning_rate": 2.845714285714286e-05,
+      "loss": 1.259,
       "step": 754
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 1.9720101781170483e-06,
-      "loss": 1.6106,
+      "epoch": 4.31,
+      "learning_rate": 2.8428571428571432e-05,
+      "loss": 1.4053,
       "step": 755
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 1.908396946564886e-06,
-      "loss": 1.463,
+      "epoch": 4.32,
+      "learning_rate": 2.84e-05,
+      "loss": 1.3372,
       "step": 756
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 1.8447837150127227e-06,
-      "loss": 1.6244,
+      "epoch": 4.33,
+      "learning_rate": 2.837142857142857e-05,
+      "loss": 1.3185,
       "step": 757
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 1.78117048346056e-06,
-      "loss": 1.4751,
+      "epoch": 4.33,
+      "learning_rate": 2.8342857142857143e-05,
+      "loss": 1.3754,
       "step": 758
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 1.7175572519083972e-06,
-      "loss": 1.4866,
+      "epoch": 4.34,
+      "learning_rate": 2.8314285714285717e-05,
+      "loss": 1.3666,
       "step": 759
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 1.653944020356234e-06,
-      "loss": 1.5802,
+      "epoch": 4.34,
+      "learning_rate": 2.8285714285714287e-05,
+      "loss": 1.2839,
       "step": 760
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 1.5903307888040714e-06,
-      "loss": 1.4904,
+      "epoch": 4.35,
+      "learning_rate": 2.825714285714286e-05,
+      "loss": 1.3134,
       "step": 761
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 1.5267175572519084e-06,
-      "loss": 1.3811,
+      "epoch": 4.35,
+      "learning_rate": 2.822857142857143e-05,
+      "loss": 1.3155,
       "step": 762
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 1.4631043256997457e-06,
-      "loss": 1.4883,
+      "epoch": 4.36,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 1.3315,
       "step": 763
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 1.3994910941475828e-06,
-      "loss": 1.5247,
+      "epoch": 4.37,
+      "learning_rate": 2.817142857142857e-05,
+      "loss": 1.3228,
       "step": 764
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 1.3358778625954198e-06,
-      "loss": 1.5334,
+      "epoch": 4.37,
+      "learning_rate": 2.814285714285714e-05,
+      "loss": 1.3601,
       "step": 765
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 1.272264631043257e-06,
-      "loss": 1.514,
+      "epoch": 4.38,
+      "learning_rate": 2.8114285714285715e-05,
+      "loss": 1.3179,
       "step": 766
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 1.2086513994910942e-06,
-      "loss": 1.4731,
+      "epoch": 4.38,
+      "learning_rate": 2.808571428571429e-05,
+      "loss": 1.3243,
       "step": 767
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 1.1450381679389313e-06,
-      "loss": 1.486,
+      "epoch": 4.39,
+      "learning_rate": 2.805714285714286e-05,
+      "loss": 1.247,
       "step": 768
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 1.0814249363867685e-06,
-      "loss": 1.5147,
+      "epoch": 4.39,
+      "learning_rate": 2.8028571428571433e-05,
+      "loss": 1.2989,
       "step": 769
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 1.0178117048346056e-06,
-      "loss": 1.4505,
+      "epoch": 4.4,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.3347,
       "step": 770
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 9.54198473282443e-07,
-      "loss": 1.4967,
+      "epoch": 4.41,
+      "learning_rate": 2.797142857142857e-05,
+      "loss": 1.3562,
       "step": 771
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 8.9058524173028e-07,
-      "loss": 1.5572,
+      "epoch": 4.41,
+      "learning_rate": 2.7942857142857143e-05,
+      "loss": 1.331,
       "step": 772
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 8.26972010178117e-07,
-      "loss": 1.5879,
+      "epoch": 4.42,
+      "learning_rate": 2.7914285714285714e-05,
+      "loss": 1.3048,
       "step": 773
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 7.633587786259542e-07,
-      "loss": 1.5198,
+      "epoch": 4.42,
+      "learning_rate": 2.7885714285714287e-05,
+      "loss": 1.3037,
       "step": 774
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 6.997455470737914e-07,
-      "loss": 1.4886,
+      "epoch": 4.43,
+      "learning_rate": 2.785714285714286e-05,
+      "loss": 1.339,
       "step": 775
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 6.361323155216285e-07,
-      "loss": 1.4612,
+      "epoch": 4.43,
+      "learning_rate": 2.782857142857143e-05,
+      "loss": 1.3303,
       "step": 776
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 5.725190839694656e-07,
-      "loss": 1.5806,
+      "epoch": 4.44,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.384,
       "step": 777
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 5.089058524173028e-07,
-      "loss": 1.5416,
+      "epoch": 4.45,
+      "learning_rate": 2.7771428571428572e-05,
+      "loss": 1.4252,
       "step": 778
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 4.4529262086514e-07,
-      "loss": 1.5734,
+      "epoch": 4.45,
+      "learning_rate": 2.7742857142857142e-05,
+      "loss": 1.3753,
       "step": 779
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 3.816793893129771e-07,
-      "loss": 1.5684,
+      "epoch": 4.46,
+      "learning_rate": 2.7714285714285716e-05,
+      "loss": 1.2363,
       "step": 780
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 3.1806615776081423e-07,
-      "loss": 1.4813,
+      "epoch": 4.46,
+      "learning_rate": 2.7685714285714286e-05,
+      "loss": 1.2956,
       "step": 781
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 2.544529262086514e-07,
-      "loss": 1.5316,
+      "epoch": 4.47,
+      "learning_rate": 2.765714285714286e-05,
+      "loss": 1.3182,
       "step": 782
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 1.9083969465648855e-07,
-      "loss": 1.4848,
+      "epoch": 4.47,
+      "learning_rate": 2.762857142857143e-05,
+      "loss": 1.3055,
       "step": 783
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 1.272264631043257e-07,
-      "loss": 1.5582,
+      "epoch": 4.48,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 1.3547,
       "step": 784
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 6.361323155216285e-08,
-      "loss": 1.4645,
+      "epoch": 4.49,
+      "learning_rate": 2.757142857142857e-05,
+      "loss": 1.2879,
       "step": 785
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 0.0,
-      "loss": 1.3783,
+      "epoch": 4.49,
+      "learning_rate": 2.7542857142857144e-05,
+      "loss": 1.3005,
       "step": 786
     },
     {
-      "epoch": 3.0,
-      "step": 786,
-      "total_flos": 1.2699738389348352e+16,
-      "train_loss": 1.6923205314095087,
-      "train_runtime": 822.2959,
-      "train_samples_per_second": 22.879,
-      "train_steps_per_second": 0.956
+      "epoch": 4.5,
+      "learning_rate": 2.7514285714285714e-05,
+      "loss": 1.3595,
+      "step": 787
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7485714285714288e-05,
+      "loss": 1.3856,
+      "step": 788
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7457142857142858e-05,
+      "loss": 1.3491,
+      "step": 789
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.742857142857143e-05,
+      "loss": 1.3044,
+      "step": 790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 1.2415,
+      "step": 791
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.737142857142857e-05,
+      "loss": 1.3315,
+      "step": 792
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7342857142857142e-05,
+      "loss": 1.3962,
+      "step": 793
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7314285714285716e-05,
+      "loss": 1.3741,
+      "step": 794
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7285714285714286e-05,
+      "loss": 1.3343,
+      "step": 795
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.725714285714286e-05,
+      "loss": 1.3336,
+      "step": 796
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.722857142857143e-05,
+      "loss": 1.4101,
+      "step": 797
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 1.3092,
+      "step": 798
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7171428571428574e-05,
+      "loss": 1.3327,
+      "step": 799
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.714285714285714e-05,
+      "loss": 1.3595,
+      "step": 800
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7114285714285715e-05,
+      "loss": 1.3042,
+      "step": 801
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7085714285714285e-05,
+      "loss": 1.2747,
+      "step": 802
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.705714285714286e-05,
+      "loss": 1.2625,
+      "step": 803
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7028571428571432e-05,
+      "loss": 1.3316,
+      "step": 804
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.3095,
+      "step": 805
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6971428571428576e-05,
+      "loss": 1.3831,
+      "step": 806
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6942857142857143e-05,
+      "loss": 1.3367,
+      "step": 807
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6914285714285713e-05,
+      "loss": 1.3466,
+      "step": 808
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6885714285714287e-05,
+      "loss": 1.2935,
+      "step": 809
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6857142857142857e-05,
+      "loss": 1.3602,
+      "step": 810
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.682857142857143e-05,
+      "loss": 1.3846,
+      "step": 811
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 1.2971,
+      "step": 812
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6771428571428575e-05,
+      "loss": 1.2417,
+      "step": 813
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.674285714285714e-05,
+      "loss": 1.3028,
+      "step": 814
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6714285714285715e-05,
+      "loss": 1.3421,
+      "step": 815
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6685714285714285e-05,
+      "loss": 1.3303,
+      "step": 816
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.665714285714286e-05,
+      "loss": 1.4346,
+      "step": 817
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.662857142857143e-05,
+      "loss": 1.3082,
+      "step": 818
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 1.2856,
+      "step": 819
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6571428571428576e-05,
+      "loss": 1.2811,
+      "step": 820
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.654285714285714e-05,
+      "loss": 1.4053,
+      "step": 821
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6514285714285714e-05,
+      "loss": 1.3435,
+      "step": 822
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6485714285714287e-05,
+      "loss": 1.2782,
+      "step": 823
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6457142857142857e-05,
+      "loss": 1.3065,
+      "step": 824
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.642857142857143e-05,
+      "loss": 1.3341,
+      "step": 825
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.64e-05,
+      "loss": 1.2693,
+      "step": 826
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6371428571428575e-05,
+      "loss": 1.2947,
+      "step": 827
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6342857142857142e-05,
+      "loss": 1.3146,
+      "step": 828
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6314285714285712e-05,
+      "loss": 1.289,
+      "step": 829
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6285714285714286e-05,
+      "loss": 1.3284,
+      "step": 830
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.625714285714286e-05,
+      "loss": 1.2655,
+      "step": 831
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.622857142857143e-05,
+      "loss": 1.4469,
+      "step": 832
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 1.2999,
+      "step": 833
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6171428571428574e-05,
+      "loss": 1.3227,
+      "step": 834
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6142857142857147e-05,
+      "loss": 1.337,
+      "step": 835
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6114285714285714e-05,
+      "loss": 1.3279,
+      "step": 836
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6085714285714284e-05,
+      "loss": 1.291,
+      "step": 837
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6057142857142858e-05,
+      "loss": 1.4128,
+      "step": 838
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.602857142857143e-05,
+      "loss": 1.3253,
+      "step": 839
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.343,
+      "step": 840
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5971428571428575e-05,
+      "loss": 1.2943,
+      "step": 841
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5942857142857146e-05,
+      "loss": 1.387,
+      "step": 842
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5914285714285713e-05,
+      "loss": 1.2921,
+      "step": 843
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5885714285714286e-05,
+      "loss": 1.4114,
+      "step": 844
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5857142857142856e-05,
+      "loss": 1.4293,
+      "step": 845
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.582857142857143e-05,
+      "loss": 1.2976,
+      "step": 846
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.58e-05,
+      "loss": 1.3084,
+      "step": 847
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5771428571428574e-05,
+      "loss": 1.3507,
+      "step": 848
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5742857142857148e-05,
+      "loss": 1.3024,
+      "step": 849
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 1.2941,
+      "step": 850
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5685714285714285e-05,
+      "loss": 1.3075,
+      "step": 851
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.565714285714286e-05,
+      "loss": 1.2879,
+      "step": 852
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.562857142857143e-05,
+      "loss": 1.3169,
+      "step": 853
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 1.3521,
+      "step": 854
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5571428571428572e-05,
+      "loss": 1.3113,
+      "step": 855
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5542857142857146e-05,
+      "loss": 1.3631,
+      "step": 856
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5514285714285713e-05,
+      "loss": 1.3454,
+      "step": 857
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5485714285714287e-05,
+      "loss": 1.3585,
+      "step": 858
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5457142857142857e-05,
+      "loss": 1.4251,
+      "step": 859
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.542857142857143e-05,
+      "loss": 1.4008,
+      "step": 860
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.54e-05,
+      "loss": 1.3459,
+      "step": 861
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5371428571428574e-05,
+      "loss": 1.3039,
+      "step": 862
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5342857142857145e-05,
+      "loss": 1.3746,
+      "step": 863
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5314285714285718e-05,
+      "loss": 1.3884,
+      "step": 864
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5285714285714285e-05,
+      "loss": 1.351,
+      "step": 865
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5257142857142855e-05,
+      "loss": 1.2113,
+      "step": 866
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.522857142857143e-05,
+      "loss": 1.3051,
+      "step": 867
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 1.3643,
+      "step": 868
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5171428571428573e-05,
+      "loss": 1.2819,
+      "step": 869
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5142857142857147e-05,
+      "loss": 1.3283,
+      "step": 870
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5114285714285717e-05,
+      "loss": 1.3209,
+      "step": 871
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5085714285714284e-05,
+      "loss": 1.3381,
+      "step": 872
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5057142857142857e-05,
+      "loss": 1.3348,
+      "step": 873
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5028571428571428e-05,
+      "loss": 1.3479,
+      "step": 874
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5e-05,
+      "loss": 1.3556,
+      "step": 875
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4971428571428575e-05,
+      "loss": 1.2632,
+      "step": 876
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4942857142857142e-05,
+      "loss": 1.2726,
+      "step": 877
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4914285714285715e-05,
+      "loss": 1.2451,
+      "step": 878
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4885714285714286e-05,
+      "loss": 1.2124,
+      "step": 879
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.485714285714286e-05,
+      "loss": 1.2437,
+      "step": 880
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.482857142857143e-05,
+      "loss": 1.259,
+      "step": 881
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.48e-05,
+      "loss": 1.3374,
+      "step": 882
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4771428571428573e-05,
+      "loss": 1.2851,
+      "step": 883
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4742857142857147e-05,
+      "loss": 1.3082,
+      "step": 884
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.4714285714285714e-05,
+      "loss": 1.3048,
+      "step": 885
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.4685714285714288e-05,
+      "loss": 1.2108,
+      "step": 886
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4657142857142858e-05,
+      "loss": 1.296,
+      "step": 887
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4628571428571428e-05,
+      "loss": 1.2827,
+      "step": 888
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.46e-05,
+      "loss": 1.271,
+      "step": 889
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4571428571428572e-05,
+      "loss": 1.3433,
+      "step": 890
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4542857142857146e-05,
+      "loss": 1.2282,
+      "step": 891
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4514285714285716e-05,
+      "loss": 1.3487,
+      "step": 892
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4485714285714286e-05,
+      "loss": 1.2905,
+      "step": 893
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.445714285714286e-05,
+      "loss": 1.3112,
+      "step": 894
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.442857142857143e-05,
+      "loss": 1.32,
+      "step": 895
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.44e-05,
+      "loss": 1.3555,
+      "step": 896
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4371428571428574e-05,
+      "loss": 1.231,
+      "step": 897
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4342857142857144e-05,
+      "loss": 1.3232,
+      "step": 898
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4314285714285714e-05,
+      "loss": 1.3095,
+      "step": 899
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4285714285714288e-05,
+      "loss": 1.2877,
+      "step": 900
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4257142857142858e-05,
+      "loss": 1.2923,
+      "step": 901
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4228571428571432e-05,
+      "loss": 1.1921,
+      "step": 902
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 1.2367,
+      "step": 903
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4171428571428572e-05,
+      "loss": 1.3188,
+      "step": 904
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4142857142857146e-05,
+      "loss": 1.2978,
+      "step": 905
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4114285714285713e-05,
+      "loss": 1.3123,
+      "step": 906
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4085714285714286e-05,
+      "loss": 1.2387,
+      "step": 907
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.405714285714286e-05,
+      "loss": 1.2904,
+      "step": 908
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.402857142857143e-05,
+      "loss": 1.258,
+      "step": 909
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4e-05,
+      "loss": 1.2735,
+      "step": 910
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.397142857142857e-05,
+      "loss": 1.2948,
+      "step": 911
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3942857142857144e-05,
+      "loss": 1.3037,
+      "step": 912
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3914285714285715e-05,
+      "loss": 1.3115,
+      "step": 913
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3885714285714285e-05,
+      "loss": 1.3237,
+      "step": 914
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.385714285714286e-05,
+      "loss": 1.3139,
+      "step": 915
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3828571428571432e-05,
+      "loss": 1.3182,
+      "step": 916
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.38e-05,
+      "loss": 1.2913,
+      "step": 917
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3771428571428573e-05,
+      "loss": 1.2651,
+      "step": 918
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3742857142857143e-05,
+      "loss": 1.3385,
+      "step": 919
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3714285714285717e-05,
+      "loss": 1.3538,
+      "step": 920
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3685714285714287e-05,
+      "loss": 1.3486,
+      "step": 921
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3657142857142857e-05,
+      "loss": 1.2768,
+      "step": 922
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.362857142857143e-05,
+      "loss": 1.2944,
+      "step": 923
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.36e-05,
+      "loss": 1.348,
+      "step": 924
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 1.3669,
+      "step": 925
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3542857142857145e-05,
+      "loss": 1.2474,
+      "step": 926
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3514285714285715e-05,
+      "loss": 1.4053,
+      "step": 927
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3485714285714285e-05,
+      "loss": 1.287,
+      "step": 928
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.345714285714286e-05,
+      "loss": 1.2992,
+      "step": 929
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.342857142857143e-05,
+      "loss": 1.277,
+      "step": 930
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 1.2658,
+      "step": 931
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3371428571428573e-05,
+      "loss": 1.2591,
+      "step": 932
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3342857142857143e-05,
+      "loss": 1.2843,
+      "step": 933
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3314285714285717e-05,
+      "loss": 1.2686,
+      "step": 934
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3285714285714287e-05,
+      "loss": 1.2783,
+      "step": 935
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3257142857142858e-05,
+      "loss": 1.2888,
+      "step": 936
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.322857142857143e-05,
+      "loss": 1.3207,
+      "step": 937
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.32e-05,
+      "loss": 1.2825,
+      "step": 938
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3171428571428572e-05,
+      "loss": 1.2808,
+      "step": 939
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3142857142857145e-05,
+      "loss": 1.2387,
+      "step": 940
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3114285714285716e-05,
+      "loss": 1.3441,
+      "step": 941
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3085714285714286e-05,
+      "loss": 1.3292,
+      "step": 942
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3057142857142856e-05,
+      "loss": 1.2735,
+      "step": 943
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.302857142857143e-05,
+      "loss": 1.2249,
+      "step": 944
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.2249,
+      "step": 945
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.297142857142857e-05,
+      "loss": 1.1812,
+      "step": 946
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2942857142857144e-05,
+      "loss": 1.3092,
+      "step": 947
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2914285714285718e-05,
+      "loss": 1.3148,
+      "step": 948
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2885714285714288e-05,
+      "loss": 1.1908,
+      "step": 949
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 1.2542,
+      "step": 950
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2828571428571428e-05,
+      "loss": 1.1672,
+      "step": 951
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 1.2224,
+      "step": 952
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.2771428571428572e-05,
+      "loss": 1.2576,
+      "step": 953
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.2742857142857142e-05,
+      "loss": 1.2776,
+      "step": 954
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2714285714285716e-05,
+      "loss": 1.3872,
+      "step": 955
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2685714285714286e-05,
+      "loss": 1.2355,
+      "step": 956
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.2657142857142857e-05,
+      "loss": 1.2203,
+      "step": 957
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.262857142857143e-05,
+      "loss": 1.2918,
+      "step": 958
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.26e-05,
+      "loss": 1.3314,
+      "step": 959
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.257142857142857e-05,
+      "loss": 1.23,
+      "step": 960
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2542857142857144e-05,
+      "loss": 1.2755,
+      "step": 961
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2514285714285715e-05,
+      "loss": 1.2348,
+      "step": 962
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2485714285714288e-05,
+      "loss": 1.3726,
+      "step": 963
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.245714285714286e-05,
+      "loss": 1.2435,
+      "step": 964
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.242857142857143e-05,
+      "loss": 1.3648,
+      "step": 965
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 1.3117,
+      "step": 966
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2371428571428573e-05,
+      "loss": 1.2374,
+      "step": 967
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2342857142857143e-05,
+      "loss": 1.2719,
+      "step": 968
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2314285714285717e-05,
+      "loss": 1.178,
+      "step": 969
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2285714285714287e-05,
+      "loss": 1.2103,
+      "step": 970
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2257142857142857e-05,
+      "loss": 1.2689,
+      "step": 971
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.222857142857143e-05,
+      "loss": 1.2282,
+      "step": 972
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.22e-05,
+      "loss": 1.2303,
+      "step": 973
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2171428571428575e-05,
+      "loss": 1.3065,
+      "step": 974
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.214285714285714e-05,
+      "loss": 1.2426,
+      "step": 975
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2114285714285715e-05,
+      "loss": 1.2496,
+      "step": 976
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.208571428571429e-05,
+      "loss": 1.2837,
+      "step": 977
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.205714285714286e-05,
+      "loss": 1.2821,
+      "step": 978
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.202857142857143e-05,
+      "loss": 1.2269,
+      "step": 979
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.2415,
+      "step": 980
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1971428571428573e-05,
+      "loss": 1.2911,
+      "step": 981
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1942857142857143e-05,
+      "loss": 1.2307,
+      "step": 982
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1914285714285714e-05,
+      "loss": 1.2983,
+      "step": 983
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1885714285714287e-05,
+      "loss": 1.3145,
+      "step": 984
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.185714285714286e-05,
+      "loss": 1.2768,
+      "step": 985
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1828571428571428e-05,
+      "loss": 1.2651,
+      "step": 986
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.18e-05,
+      "loss": 1.296,
+      "step": 987
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.177142857142857e-05,
+      "loss": 1.2726,
+      "step": 988
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1742857142857142e-05,
+      "loss": 1.2601,
+      "step": 989
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.1714285714285715e-05,
+      "loss": 1.261,
+      "step": 990
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.1685714285714286e-05,
+      "loss": 1.2642,
+      "step": 991
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.165714285714286e-05,
+      "loss": 1.2386,
+      "step": 992
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.162857142857143e-05,
+      "loss": 1.2405,
+      "step": 993
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.16e-05,
+      "loss": 1.3141,
+      "step": 994
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1571428571428574e-05,
+      "loss": 1.2544,
+      "step": 995
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1542857142857144e-05,
+      "loss": 1.3342,
+      "step": 996
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1514285714285714e-05,
+      "loss": 1.2918,
+      "step": 997
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1485714285714288e-05,
+      "loss": 1.2641,
+      "step": 998
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1457142857142858e-05,
+      "loss": 1.24,
+      "step": 999
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 1.3657,
+      "step": 1000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 1.2104,
+      "step": 1001
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1371428571428572e-05,
+      "loss": 1.3424,
+      "step": 1002
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1342857142857146e-05,
+      "loss": 1.2253,
+      "step": 1003
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1314285714285716e-05,
+      "loss": 1.26,
+      "step": 1004
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1285714285714286e-05,
+      "loss": 1.3178,
+      "step": 1005
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.125714285714286e-05,
+      "loss": 1.3444,
+      "step": 1006
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1228571428571427e-05,
+      "loss": 1.3016,
+      "step": 1007
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.12e-05,
+      "loss": 1.3347,
+      "step": 1008
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1171428571428574e-05,
+      "loss": 1.2548,
+      "step": 1009
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1142857142857144e-05,
+      "loss": 1.2495,
+      "step": 1010
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1114285714285714e-05,
+      "loss": 1.2293,
+      "step": 1011
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1085714285714288e-05,
+      "loss": 1.2683,
+      "step": 1012
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.105714285714286e-05,
+      "loss": 1.2343,
+      "step": 1013
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1028571428571432e-05,
+      "loss": 1.2642,
+      "step": 1014
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1e-05,
+      "loss": 1.2858,
+      "step": 1015
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0971428571428572e-05,
+      "loss": 1.2633,
+      "step": 1016
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0942857142857146e-05,
+      "loss": 1.2989,
+      "step": 1017
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0914285714285713e-05,
+      "loss": 1.2302,
+      "step": 1018
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0885714285714287e-05,
+      "loss": 1.2577,
+      "step": 1019
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0857142857142857e-05,
+      "loss": 1.2848,
+      "step": 1020
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.082857142857143e-05,
+      "loss": 1.3636,
+      "step": 1021
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.08e-05,
+      "loss": 1.2772,
+      "step": 1022
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.077142857142857e-05,
+      "loss": 1.2869,
+      "step": 1023
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0742857142857145e-05,
+      "loss": 1.2698,
+      "step": 1024
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0714285714285718e-05,
+      "loss": 1.2281,
+      "step": 1025
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0685714285714285e-05,
+      "loss": 1.2564,
+      "step": 1026
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.065714285714286e-05,
+      "loss": 1.32,
+      "step": 1027
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.062857142857143e-05,
+      "loss": 1.2551,
+      "step": 1028
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.06e-05,
+      "loss": 1.2259,
+      "step": 1029
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0571428571428573e-05,
+      "loss": 1.2862,
+      "step": 1030
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0542857142857143e-05,
+      "loss": 1.3579,
+      "step": 1031
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0514285714285717e-05,
+      "loss": 1.2287,
+      "step": 1032
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0485714285714287e-05,
+      "loss": 1.2708,
+      "step": 1033
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0457142857142857e-05,
+      "loss": 1.2862,
+      "step": 1034
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.042857142857143e-05,
+      "loss": 1.223,
+      "step": 1035
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.04e-05,
+      "loss": 1.3186,
+      "step": 1036
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.037142857142857e-05,
+      "loss": 1.2676,
+      "step": 1037
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0342857142857145e-05,
+      "loss": 1.251,
+      "step": 1038
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0314285714285715e-05,
+      "loss": 1.2164,
+      "step": 1039
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0285714285714286e-05,
+      "loss": 1.2819,
+      "step": 1040
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.025714285714286e-05,
+      "loss": 1.3483,
+      "step": 1041
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.022857142857143e-05,
+      "loss": 1.2706,
+      "step": 1042
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 1.3235,
+      "step": 1043
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0171428571428573e-05,
+      "loss": 1.261,
+      "step": 1044
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0142857142857144e-05,
+      "loss": 1.2001,
+      "step": 1045
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0114285714285717e-05,
+      "loss": 1.2315,
+      "step": 1046
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0085714285714284e-05,
+      "loss": 1.2189,
+      "step": 1047
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0057142857142858e-05,
+      "loss": 1.3158,
+      "step": 1048
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.002857142857143e-05,
+      "loss": 1.2285,
+      "step": 1049
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2e-05,
+      "loss": 1.2588,
+      "step": 1050
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9971428571428572e-05,
+      "loss": 1.2846,
+      "step": 1051
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9942857142857142e-05,
+      "loss": 1.2334,
+      "step": 1052
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9914285714285716e-05,
+      "loss": 1.2025,
+      "step": 1053
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9885714285714286e-05,
+      "loss": 1.1503,
+      "step": 1054
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9857142857142856e-05,
+      "loss": 1.2338,
+      "step": 1055
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.982857142857143e-05,
+      "loss": 1.2692,
+      "step": 1056
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.218,
+      "step": 1057
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.977142857142857e-05,
+      "loss": 1.1996,
+      "step": 1058
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9742857142857144e-05,
+      "loss": 1.1991,
+      "step": 1059
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.9714285714285714e-05,
+      "loss": 1.2326,
+      "step": 1060
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.9685714285714288e-05,
+      "loss": 1.2157,
+      "step": 1061
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9657142857142858e-05,
+      "loss": 1.2037,
+      "step": 1062
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.962857142857143e-05,
+      "loss": 1.2286,
+      "step": 1063
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 1.2453,
+      "step": 1064
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9571428571428572e-05,
+      "loss": 1.1621,
+      "step": 1065
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9542857142857143e-05,
+      "loss": 1.2732,
+      "step": 1066
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9514285714285716e-05,
+      "loss": 1.1938,
+      "step": 1067
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9485714285714286e-05,
+      "loss": 1.2676,
+      "step": 1068
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9457142857142857e-05,
+      "loss": 1.219,
+      "step": 1069
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 1.2599,
+      "step": 1070
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.94e-05,
+      "loss": 1.1898,
+      "step": 1071
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9371428571428574e-05,
+      "loss": 1.2677,
+      "step": 1072
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9342857142857144e-05,
+      "loss": 1.1982,
+      "step": 1073
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9314285714285715e-05,
+      "loss": 1.2091,
+      "step": 1074
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 1.2033,
+      "step": 1075
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9257142857142855e-05,
+      "loss": 1.2865,
+      "step": 1076
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.922857142857143e-05,
+      "loss": 1.2346,
+      "step": 1077
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.2273,
+      "step": 1078
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9171428571428573e-05,
+      "loss": 1.1595,
+      "step": 1079
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9142857142857143e-05,
+      "loss": 1.252,
+      "step": 1080
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9114285714285717e-05,
+      "loss": 1.2,
+      "step": 1081
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9085714285714287e-05,
+      "loss": 1.1577,
+      "step": 1082
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9057142857142857e-05,
+      "loss": 1.2526,
+      "step": 1083
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9028571428571427e-05,
+      "loss": 1.2231,
+      "step": 1084
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9e-05,
+      "loss": 1.2639,
+      "step": 1085
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8971428571428575e-05,
+      "loss": 1.4006,
+      "step": 1086
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.894285714285714e-05,
+      "loss": 1.2843,
+      "step": 1087
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8914285714285715e-05,
+      "loss": 1.3375,
+      "step": 1088
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.888571428571429e-05,
+      "loss": 1.2015,
+      "step": 1089
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 1.2693,
+      "step": 1090
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.882857142857143e-05,
+      "loss": 1.2676,
+      "step": 1091
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.88e-05,
+      "loss": 1.2354,
+      "step": 1092
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8771428571428573e-05,
+      "loss": 1.2377,
+      "step": 1093
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8742857142857143e-05,
+      "loss": 1.2322,
+      "step": 1094
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8714285714285714e-05,
+      "loss": 1.3219,
+      "step": 1095
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8685714285714287e-05,
+      "loss": 1.2194,
+      "step": 1096
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8657142857142858e-05,
+      "loss": 1.2076,
+      "step": 1097
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8628571428571428e-05,
+      "loss": 1.1731,
+      "step": 1098
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.86e-05,
+      "loss": 1.1856,
+      "step": 1099
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8571428571428572e-05,
+      "loss": 1.2499,
+      "step": 1100
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8542857142857142e-05,
+      "loss": 1.2047,
+      "step": 1101
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8514285714285716e-05,
+      "loss": 1.2324,
+      "step": 1102
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8485714285714286e-05,
+      "loss": 1.2061,
+      "step": 1103
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.845714285714286e-05,
+      "loss": 1.2276,
+      "step": 1104
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 1.2223,
+      "step": 1105
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.84e-05,
+      "loss": 1.2657,
+      "step": 1106
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8371428571428574e-05,
+      "loss": 1.2827,
+      "step": 1107
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8342857142857144e-05,
+      "loss": 1.3136,
+      "step": 1108
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8314285714285714e-05,
+      "loss": 1.2106,
+      "step": 1109
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 1.1694,
+      "step": 1110
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8257142857142858e-05,
+      "loss": 1.1735,
+      "step": 1111
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8228571428571428e-05,
+      "loss": 1.2469,
+      "step": 1112
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 1.275,
+      "step": 1113
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8171428571428572e-05,
+      "loss": 1.3411,
+      "step": 1114
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8142857142857146e-05,
+      "loss": 1.1947,
+      "step": 1115
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8114285714285713e-05,
+      "loss": 1.2564,
+      "step": 1116
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8085714285714286e-05,
+      "loss": 1.2547,
+      "step": 1117
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.805714285714286e-05,
+      "loss": 1.2101,
+      "step": 1118
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.802857142857143e-05,
+      "loss": 1.2641,
+      "step": 1119
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8e-05,
+      "loss": 1.3164,
+      "step": 1120
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.797142857142857e-05,
+      "loss": 1.1706,
+      "step": 1121
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7942857142857144e-05,
+      "loss": 1.1826,
+      "step": 1122
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7914285714285715e-05,
+      "loss": 1.268,
+      "step": 1123
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7885714285714285e-05,
+      "loss": 1.2823,
+      "step": 1124
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 1.2254,
+      "step": 1125
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7828571428571432e-05,
+      "loss": 1.3097,
+      "step": 1126
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.78e-05,
+      "loss": 1.2124,
+      "step": 1127
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7771428571428573e-05,
+      "loss": 1.1683,
+      "step": 1128
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7742857142857143e-05,
+      "loss": 1.1925,
+      "step": 1129
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7714285714285713e-05,
+      "loss": 1.2316,
+      "step": 1130
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7685714285714287e-05,
+      "loss": 1.314,
+      "step": 1131
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7657142857142857e-05,
+      "loss": 1.3132,
+      "step": 1132
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.762857142857143e-05,
+      "loss": 1.2391,
+      "step": 1133
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.76e-05,
+      "loss": 1.2648,
+      "step": 1134
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 1.2307,
+      "step": 1135
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7542857142857145e-05,
+      "loss": 1.3283,
+      "step": 1136
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7514285714285715e-05,
+      "loss": 1.1606,
+      "step": 1137
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7485714285714285e-05,
+      "loss": 1.1954,
+      "step": 1138
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.745714285714286e-05,
+      "loss": 1.2318,
+      "step": 1139
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 1.231,
+      "step": 1140
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.74e-05,
+      "loss": 1.1751,
+      "step": 1141
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7371428571428573e-05,
+      "loss": 1.2806,
+      "step": 1142
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7342857142857143e-05,
+      "loss": 1.1632,
+      "step": 1143
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7314285714285717e-05,
+      "loss": 1.1881,
+      "step": 1144
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7285714285714287e-05,
+      "loss": 1.189,
+      "step": 1145
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7257142857142857e-05,
+      "loss": 1.2705,
+      "step": 1146
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.722857142857143e-05,
+      "loss": 1.2645,
+      "step": 1147
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 1.2979,
+      "step": 1148
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.717142857142857e-05,
+      "loss": 1.2624,
+      "step": 1149
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7142857142857145e-05,
+      "loss": 1.2362,
+      "step": 1150
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7114285714285715e-05,
+      "loss": 1.2824,
+      "step": 1151
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7085714285714286e-05,
+      "loss": 1.2423,
+      "step": 1152
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7057142857142856e-05,
+      "loss": 1.1784,
+      "step": 1153
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.702857142857143e-05,
+      "loss": 1.2816,
+      "step": 1154
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.2822,
+      "step": 1155
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.697142857142857e-05,
+      "loss": 1.1859,
+      "step": 1156
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6942857142857144e-05,
+      "loss": 1.1932,
+      "step": 1157
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6914285714285717e-05,
+      "loss": 1.2268,
+      "step": 1158
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6885714285714284e-05,
+      "loss": 1.2129,
+      "step": 1159
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 1.2325,
+      "step": 1160
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6828571428571428e-05,
+      "loss": 1.3025,
+      "step": 1161
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.2879,
+      "step": 1162
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6771428571428572e-05,
+      "loss": 1.2264,
+      "step": 1163
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6742857142857142e-05,
+      "loss": 1.2469,
+      "step": 1164
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 1.2385,
+      "step": 1165
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6685714285714286e-05,
+      "loss": 1.2549,
+      "step": 1166
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6657142857142856e-05,
+      "loss": 1.2005,
+      "step": 1167
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.662857142857143e-05,
+      "loss": 1.2035,
+      "step": 1168
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.66e-05,
+      "loss": 1.2198,
+      "step": 1169
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.657142857142857e-05,
+      "loss": 1.2112,
+      "step": 1170
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6542857142857144e-05,
+      "loss": 1.2061,
+      "step": 1171
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6514285714285714e-05,
+      "loss": 1.2431,
+      "step": 1172
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6485714285714288e-05,
+      "loss": 1.2181,
+      "step": 1173
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.645714285714286e-05,
+      "loss": 1.296,
+      "step": 1174
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 1.2314,
+      "step": 1175
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 1.2493,
+      "step": 1176
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6371428571428572e-05,
+      "loss": 1.1707,
+      "step": 1177
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6342857142857143e-05,
+      "loss": 1.2368,
+      "step": 1178
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6314285714285716e-05,
+      "loss": 1.2585,
+      "step": 1179
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 1.1886,
+      "step": 1180
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6257142857142857e-05,
+      "loss": 1.2597,
+      "step": 1181
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.622857142857143e-05,
+      "loss": 1.1979,
+      "step": 1182
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.62e-05,
+      "loss": 1.2213,
+      "step": 1183
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6171428571428574e-05,
+      "loss": 1.2825,
+      "step": 1184
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.614285714285714e-05,
+      "loss": 1.2104,
+      "step": 1185
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6114285714285715e-05,
+      "loss": 1.2539,
+      "step": 1186
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.608571428571429e-05,
+      "loss": 1.2486,
+      "step": 1187
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6057142857142855e-05,
+      "loss": 1.1958,
+      "step": 1188
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.602857142857143e-05,
+      "loss": 1.2434,
+      "step": 1189
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2585,
+      "step": 1190
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5971428571428573e-05,
+      "loss": 1.1735,
+      "step": 1191
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5942857142857143e-05,
+      "loss": 1.19,
+      "step": 1192
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5914285714285713e-05,
+      "loss": 1.2522,
+      "step": 1193
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5885714285714287e-05,
+      "loss": 1.2143,
+      "step": 1194
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 1.1572,
+      "step": 1195
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5828571428571428e-05,
+      "loss": 1.1969,
+      "step": 1196
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.58e-05,
+      "loss": 1.1789,
+      "step": 1197
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.577142857142857e-05,
+      "loss": 1.2163,
+      "step": 1198
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.574285714285714e-05,
+      "loss": 1.2201,
+      "step": 1199
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 1.2394,
+      "step": 1200
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5685714285714286e-05,
+      "loss": 1.1889,
+      "step": 1201
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.565714285714286e-05,
+      "loss": 1.2453,
+      "step": 1202
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.562857142857143e-05,
+      "loss": 1.2449,
+      "step": 1203
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.56e-05,
+      "loss": 1.2958,
+      "step": 1204
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5571428571428573e-05,
+      "loss": 1.1921,
+      "step": 1205
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5542857142857144e-05,
+      "loss": 1.1871,
+      "step": 1206
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5514285714285714e-05,
+      "loss": 1.2974,
+      "step": 1207
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5485714285714287e-05,
+      "loss": 1.1678,
+      "step": 1208
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5457142857142858e-05,
+      "loss": 1.3504,
+      "step": 1209
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5428571428571428e-05,
+      "loss": 1.2325,
+      "step": 1210
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.54e-05,
+      "loss": 1.1954,
+      "step": 1211
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5371428571428572e-05,
+      "loss": 1.2511,
+      "step": 1212
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5342857142857146e-05,
+      "loss": 1.2351,
+      "step": 1213
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5314285714285716e-05,
+      "loss": 1.2587,
+      "step": 1214
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5285714285714286e-05,
+      "loss": 1.2194,
+      "step": 1215
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5257142857142858e-05,
+      "loss": 1.1658,
+      "step": 1216
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5228571428571428e-05,
+      "loss": 1.187,
+      "step": 1217
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.52e-05,
+      "loss": 1.197,
+      "step": 1218
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5171428571428572e-05,
+      "loss": 1.2339,
+      "step": 1219
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 1.2621,
+      "step": 1220
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5114285714285714e-05,
+      "loss": 1.2397,
+      "step": 1221
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5085714285714286e-05,
+      "loss": 1.2295,
+      "step": 1222
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5057142857142858e-05,
+      "loss": 1.2575,
+      "step": 1223
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5028571428571428e-05,
+      "loss": 1.1525,
+      "step": 1224
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.2887,
+      "step": 1225
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4971428571428572e-05,
+      "loss": 1.2473,
+      "step": 1226
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4942857142857144e-05,
+      "loss": 1.1196,
+      "step": 1227
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4914285714285715e-05,
+      "loss": 1.2251,
+      "step": 1228
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4885714285714286e-05,
+      "loss": 1.1242,
+      "step": 1229
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 1.1919,
+      "step": 1230
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.482857142857143e-05,
+      "loss": 1.1957,
+      "step": 1231
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.48e-05,
+      "loss": 1.2005,
+      "step": 1232
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4771428571428573e-05,
+      "loss": 1.179,
+      "step": 1233
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4742857142857144e-05,
+      "loss": 1.2359,
+      "step": 1234
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4714285714285713e-05,
+      "loss": 1.259,
+      "step": 1235
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4685714285714287e-05,
+      "loss": 1.2717,
+      "step": 1236
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.4657142857142859e-05,
+      "loss": 1.0816,
+      "step": 1237
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.462857142857143e-05,
+      "loss": 1.2102,
+      "step": 1238
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 1.181,
+      "step": 1239
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 1.2051,
+      "step": 1240
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4542857142857145e-05,
+      "loss": 1.1448,
+      "step": 1241
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4514285714285713e-05,
+      "loss": 1.1301,
+      "step": 1242
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4485714285714285e-05,
+      "loss": 1.2258,
+      "step": 1243
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.4457142857142857e-05,
+      "loss": 1.23,
+      "step": 1244
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.442857142857143e-05,
+      "loss": 1.1806,
+      "step": 1245
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.44e-05,
+      "loss": 1.2021,
+      "step": 1246
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4371428571428571e-05,
+      "loss": 1.268,
+      "step": 1247
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4342857142857143e-05,
+      "loss": 1.238,
+      "step": 1248
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4314285714285717e-05,
+      "loss": 1.1767,
+      "step": 1249
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 1.2127,
+      "step": 1250
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4257142857142857e-05,
+      "loss": 1.2837,
+      "step": 1251
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.422857142857143e-05,
+      "loss": 1.1183,
+      "step": 1252
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.42e-05,
+      "loss": 1.161,
+      "step": 1253
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4171428571428572e-05,
+      "loss": 1.1957,
+      "step": 1254
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4142857142857143e-05,
+      "loss": 1.1916,
+      "step": 1255
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4114285714285715e-05,
+      "loss": 1.1816,
+      "step": 1256
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.4085714285714286e-05,
+      "loss": 1.2454,
+      "step": 1257
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4057142857142858e-05,
+      "loss": 1.2272,
+      "step": 1258
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.402857142857143e-05,
+      "loss": 1.1294,
+      "step": 1259
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.2474,
+      "step": 1260
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3971428571428572e-05,
+      "loss": 1.1345,
+      "step": 1261
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3942857142857144e-05,
+      "loss": 1.1249,
+      "step": 1262
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.3914285714285716e-05,
+      "loss": 1.1739,
+      "step": 1263
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.3885714285714286e-05,
+      "loss": 1.2192,
+      "step": 1264
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.3857142857142858e-05,
+      "loss": 1.2375,
+      "step": 1265
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.382857142857143e-05,
+      "loss": 1.1861,
+      "step": 1266
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.1637,
+      "step": 1267
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3771428571428572e-05,
+      "loss": 1.1864,
+      "step": 1268
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3742857142857144e-05,
+      "loss": 1.2389,
+      "step": 1269
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 1.1863,
+      "step": 1270
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3685714285714284e-05,
+      "loss": 1.1982,
+      "step": 1271
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3657142857142858e-05,
+      "loss": 1.2099,
+      "step": 1272
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.362857142857143e-05,
+      "loss": 1.2451,
+      "step": 1273
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.2528,
+      "step": 1274
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.357142857142857e-05,
+      "loss": 1.1299,
+      "step": 1275
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3542857142857142e-05,
+      "loss": 1.1795,
+      "step": 1276
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3514285714285716e-05,
+      "loss": 1.2197,
+      "step": 1277
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3485714285714288e-05,
+      "loss": 1.1641,
+      "step": 1278
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3457142857142857e-05,
+      "loss": 1.1615,
+      "step": 1279
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 1.2236,
+      "step": 1280
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 1.1911,
+      "step": 1281
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.337142857142857e-05,
+      "loss": 1.2689,
+      "step": 1282
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.3342857142857143e-05,
+      "loss": 1.1873,
+      "step": 1283
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3314285714285715e-05,
+      "loss": 1.1661,
+      "step": 1284
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3285714285714288e-05,
+      "loss": 1.1758,
+      "step": 1285
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3257142857142857e-05,
+      "loss": 1.2637,
+      "step": 1286
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3228571428571429e-05,
+      "loss": 1.2458,
+      "step": 1287
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.32e-05,
+      "loss": 1.2376,
+      "step": 1288
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3171428571428571e-05,
+      "loss": 1.22,
+      "step": 1289
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3142857142857143e-05,
+      "loss": 1.1867,
+      "step": 1290
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3114285714285715e-05,
+      "loss": 1.1729,
+      "step": 1291
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3085714285714287e-05,
+      "loss": 1.2337,
+      "step": 1292
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3057142857142857e-05,
+      "loss": 1.1378,
+      "step": 1293
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3028571428571429e-05,
+      "loss": 1.2629,
+      "step": 1294
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.1673,
+      "step": 1295
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.2971428571428573e-05,
+      "loss": 1.2138,
+      "step": 1296
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.2942857142857143e-05,
+      "loss": 1.1845,
+      "step": 1297
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2914285714285715e-05,
+      "loss": 1.1703,
+      "step": 1298
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2885714285714287e-05,
+      "loss": 1.2146,
+      "step": 1299
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 1.2226,
+      "step": 1300
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.282857142857143e-05,
+      "loss": 1.2902,
+      "step": 1301
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.2114,
+      "step": 1302
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2771428571428573e-05,
+      "loss": 1.1652,
+      "step": 1303
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2742857142857143e-05,
+      "loss": 1.1793,
+      "step": 1304
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2714285714285715e-05,
+      "loss": 1.2063,
+      "step": 1305
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2685714285714287e-05,
+      "loss": 1.248,
+      "step": 1306
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2657142857142859e-05,
+      "loss": 1.2503,
+      "step": 1307
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2628571428571428e-05,
+      "loss": 1.2242,
+      "step": 1308
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 1.1112,
+      "step": 1309
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2571428571428573e-05,
+      "loss": 1.1516,
+      "step": 1310
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2542857142857142e-05,
+      "loss": 1.2051,
+      "step": 1311
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2514285714285714e-05,
+      "loss": 1.1921,
+      "step": 1312
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2485714285714287e-05,
+      "loss": 1.1793,
+      "step": 1313
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2457142857142858e-05,
+      "loss": 1.2241,
+      "step": 1314
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 1.2178,
+      "step": 1315
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.24e-05,
+      "loss": 1.2603,
+      "step": 1316
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2371428571428574e-05,
+      "loss": 1.2178,
+      "step": 1317
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2342857142857144e-05,
+      "loss": 1.2263,
+      "step": 1318
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2314285714285714e-05,
+      "loss": 1.1786,
+      "step": 1319
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2285714285714286e-05,
+      "loss": 1.2788,
+      "step": 1320
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2257142857142858e-05,
+      "loss": 1.1498,
+      "step": 1321
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.222857142857143e-05,
+      "loss": 1.2405,
+      "step": 1322
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.22e-05,
+      "loss": 1.164,
+      "step": 1323
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2171428571428572e-05,
+      "loss": 1.2008,
+      "step": 1324
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2142857142857144e-05,
+      "loss": 1.2171,
+      "step": 1325
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2114285714285716e-05,
+      "loss": 1.2831,
+      "step": 1326
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2085714285714286e-05,
+      "loss": 1.1482,
+      "step": 1327
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2057142857142856e-05,
+      "loss": 1.1889,
+      "step": 1328
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.202857142857143e-05,
+      "loss": 1.1867,
+      "step": 1329
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.2e-05,
+      "loss": 1.1787,
+      "step": 1330
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.1971428571428572e-05,
+      "loss": 1.2058,
+      "step": 1331
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.1942857142857142e-05,
+      "loss": 1.0851,
+      "step": 1332
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1914285714285716e-05,
+      "loss": 1.1733,
+      "step": 1333
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1885714285714286e-05,
+      "loss": 1.198,
+      "step": 1334
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1857142857142858e-05,
+      "loss": 1.187,
+      "step": 1335
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1828571428571429e-05,
+      "loss": 1.2155,
+      "step": 1336
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.18e-05,
+      "loss": 1.2444,
+      "step": 1337
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1771428571428572e-05,
+      "loss": 1.191,
+      "step": 1338
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1742857142857143e-05,
+      "loss": 1.1728,
+      "step": 1339
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1714285714285715e-05,
+      "loss": 1.1744,
+      "step": 1340
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1685714285714287e-05,
+      "loss": 1.2009,
+      "step": 1341
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1657142857142859e-05,
+      "loss": 1.2131,
+      "step": 1342
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.1628571428571429e-05,
+      "loss": 1.2086,
+      "step": 1343
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.16e-05,
+      "loss": 1.181,
+      "step": 1344
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 1.1559,
+      "step": 1345
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1542857142857143e-05,
+      "loss": 1.2186,
+      "step": 1346
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1514285714285715e-05,
+      "loss": 1.0891,
+      "step": 1347
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1485714285714285e-05,
+      "loss": 1.2447,
+      "step": 1348
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1457142857142859e-05,
+      "loss": 1.1734,
+      "step": 1349
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.1485,
+      "step": 1350
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.237,
+      "step": 1351
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1371428571428571e-05,
+      "loss": 1.2379,
+      "step": 1352
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1342857142857143e-05,
+      "loss": 1.232,
+      "step": 1353
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1314285714285715e-05,
+      "loss": 1.2107,
+      "step": 1354
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1285714285714285e-05,
+      "loss": 1.2678,
+      "step": 1355
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.1257142857142857e-05,
+      "loss": 1.1586,
+      "step": 1356
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.122857142857143e-05,
+      "loss": 1.1271,
+      "step": 1357
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.2166,
+      "step": 1358
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1171428571428571e-05,
+      "loss": 1.1643,
+      "step": 1359
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 1.2628,
+      "step": 1360
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1114285714285715e-05,
+      "loss": 1.2173,
+      "step": 1361
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1085714285714287e-05,
+      "loss": 1.1893,
+      "step": 1362
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.1057142857142858e-05,
+      "loss": 1.2319,
+      "step": 1363
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.102857142857143e-05,
+      "loss": 1.183,
+      "step": 1364
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.2344,
+      "step": 1365
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0971428571428572e-05,
+      "loss": 1.1551,
+      "step": 1366
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0942857142857144e-05,
+      "loss": 1.1417,
+      "step": 1367
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.0914285714285714e-05,
+      "loss": 1.232,
+      "step": 1368
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.0885714285714286e-05,
+      "loss": 1.2065,
+      "step": 1369
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 1.2284,
+      "step": 1370
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.082857142857143e-05,
+      "loss": 1.1259,
+      "step": 1371
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.08e-05,
+      "loss": 1.2002,
+      "step": 1372
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0771428571428572e-05,
+      "loss": 1.1922,
+      "step": 1373
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0742857142857144e-05,
+      "loss": 1.1551,
+      "step": 1374
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 1.2329,
+      "step": 1375
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.0685714285714286e-05,
+      "loss": 1.2197,
+      "step": 1376
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0657142857142858e-05,
+      "loss": 1.1409,
+      "step": 1377
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.062857142857143e-05,
+      "loss": 1.1619,
+      "step": 1378
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.06e-05,
+      "loss": 1.1285,
+      "step": 1379
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 1.2532,
+      "step": 1380
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.0542857142857144e-05,
+      "loss": 1.218,
+      "step": 1381
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0514285714285716e-05,
+      "loss": 1.2115,
+      "step": 1382
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0485714285714286e-05,
+      "loss": 1.1195,
+      "step": 1383
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0457142857142856e-05,
+      "loss": 1.1154,
+      "step": 1384
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0428571428571428e-05,
+      "loss": 1.1892,
+      "step": 1385
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.04e-05,
+      "loss": 1.222,
+      "step": 1386
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0371428571428572e-05,
+      "loss": 1.0729,
+      "step": 1387
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0342857142857143e-05,
+      "loss": 1.2224,
+      "step": 1388
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.0314285714285715e-05,
+      "loss": 1.1833,
+      "step": 1389
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.0285714285714286e-05,
+      "loss": 1.248,
+      "step": 1390
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0257142857142858e-05,
+      "loss": 1.1831,
+      "step": 1391
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0228571428571429e-05,
+      "loss": 1.1667,
+      "step": 1392
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.02e-05,
+      "loss": 1.2026,
+      "step": 1393
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0171428571428573e-05,
+      "loss": 1.172,
+      "step": 1394
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0142857142857143e-05,
+      "loss": 1.1752,
+      "step": 1395
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.0114285714285715e-05,
+      "loss": 1.1991,
+      "step": 1396
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.0085714285714287e-05,
+      "loss": 1.1911,
+      "step": 1397
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0057142857142859e-05,
+      "loss": 1.1975,
+      "step": 1398
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0028571428571429e-05,
+      "loss": 1.2783,
+      "step": 1399
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1e-05,
+      "loss": 1.3125,
+      "step": 1400
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.971428571428571e-06,
+      "loss": 1.2041,
+      "step": 1401
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.942857142857143e-06,
+      "loss": 1.0592,
+      "step": 1402
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.914285714285715e-06,
+      "loss": 1.1759,
+      "step": 1403
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.885714285714285e-06,
+      "loss": 1.1795,
+      "step": 1404
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.857142857142857e-06,
+      "loss": 1.1374,
+      "step": 1405
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.828571428571429e-06,
+      "loss": 1.1302,
+      "step": 1406
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.1266,
+      "step": 1407
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.771428571428571e-06,
+      "loss": 1.2986,
+      "step": 1408
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.742857142857143e-06,
+      "loss": 1.1811,
+      "step": 1409
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 1.1399,
+      "step": 1410
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.685714285714287e-06,
+      "loss": 1.1527,
+      "step": 1411
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.657142857142857e-06,
+      "loss": 1.2195,
+      "step": 1412
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.628571428571428e-06,
+      "loss": 1.1753,
+      "step": 1413
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.1934,
+      "step": 1414
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.571428571428572e-06,
+      "loss": 1.2099,
+      "step": 1415
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.542857142857143e-06,
+      "loss": 1.2931,
+      "step": 1416
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.514285714285714e-06,
+      "loss": 1.1303,
+      "step": 1417
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.485714285714287e-06,
+      "loss": 1.173,
+      "step": 1418
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.457142857142858e-06,
+      "loss": 1.1857,
+      "step": 1419
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 1.2425,
+      "step": 1420
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.4e-06,
+      "loss": 1.2034,
+      "step": 1421
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.371428571428572e-06,
+      "loss": 1.1445,
+      "step": 1422
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.342857142857144e-06,
+      "loss": 1.1803,
+      "step": 1423
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.314285714285714e-06,
+      "loss": 1.17,
+      "step": 1424
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.285714285714286e-06,
+      "loss": 1.1949,
+      "step": 1425
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.257142857142858e-06,
+      "loss": 1.2051,
+      "step": 1426
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.22857142857143e-06,
+      "loss": 1.2093,
+      "step": 1427
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.2e-06,
+      "loss": 1.2029,
+      "step": 1428
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.171428571428572e-06,
+      "loss": 1.1428,
+      "step": 1429
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 1.1722,
+      "step": 1430
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.114285714285714e-06,
+      "loss": 1.2178,
+      "step": 1431
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.085714285714286e-06,
+      "loss": 1.2454,
+      "step": 1432
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.057142857142856e-06,
+      "loss": 1.1531,
+      "step": 1433
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.02857142857143e-06,
+      "loss": 1.0534,
+      "step": 1434
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9e-06,
+      "loss": 1.2329,
+      "step": 1435
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.971428571428572e-06,
+      "loss": 1.1228,
+      "step": 1436
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.942857142857142e-06,
+      "loss": 1.1843,
+      "step": 1437
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.914285714285716e-06,
+      "loss": 1.1364,
+      "step": 1438
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.885714285714286e-06,
+      "loss": 1.1261,
+      "step": 1439
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.857142857142857e-06,
+      "loss": 1.119,
+      "step": 1440
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.828571428571429e-06,
+      "loss": 1.1867,
+      "step": 1441
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.8e-06,
+      "loss": 1.0948,
+      "step": 1442
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.771428571428572e-06,
+      "loss": 1.0973,
+      "step": 1443
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.742857142857143e-06,
+      "loss": 1.1451,
+      "step": 1444
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.714285714285715e-06,
+      "loss": 1.1849,
+      "step": 1445
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.685714285714287e-06,
+      "loss": 1.1819,
+      "step": 1446
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.657142857142858e-06,
+      "loss": 1.1684,
+      "step": 1447
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.628571428571429e-06,
+      "loss": 1.1166,
+      "step": 1448
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 1.1608,
+      "step": 1449
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.571428571428573e-06,
+      "loss": 1.1536,
+      "step": 1450
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.542857142857143e-06,
+      "loss": 1.0794,
+      "step": 1451
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.514285714285715e-06,
+      "loss": 1.187,
+      "step": 1452
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.485714285714285e-06,
+      "loss": 1.1096,
+      "step": 1453
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.457142857142859e-06,
+      "loss": 1.2126,
+      "step": 1454
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.428571428571429e-06,
+      "loss": 1.1528,
+      "step": 1455
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.1233,
+      "step": 1456
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.371428571428571e-06,
+      "loss": 1.1533,
+      "step": 1457
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.342857142857143e-06,
+      "loss": 1.1637,
+      "step": 1458
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.314285714285715e-06,
+      "loss": 1.1799,
+      "step": 1459
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.285714285714285e-06,
+      "loss": 1.2575,
+      "step": 1460
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.257142857142857e-06,
+      "loss": 1.1258,
+      "step": 1461
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.22857142857143e-06,
+      "loss": 1.2118,
+      "step": 1462
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 1.202,
+      "step": 1463
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.171428571428571e-06,
+      "loss": 1.1769,
+      "step": 1464
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.142857142857143e-06,
+      "loss": 1.1704,
+      "step": 1465
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.114285714285715e-06,
+      "loss": 1.1668,
+      "step": 1466
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.085714285714287e-06,
+      "loss": 1.1993,
+      "step": 1467
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.057142857142857e-06,
+      "loss": 1.1184,
+      "step": 1468
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.028571428571428e-06,
+      "loss": 1.2608,
+      "step": 1469
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1715,
+      "step": 1470
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.971428571428572e-06,
+      "loss": 1.1536,
+      "step": 1471
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.942857142857144e-06,
+      "loss": 1.1514,
+      "step": 1472
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.914285714285714e-06,
+      "loss": 1.1995,
+      "step": 1473
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.885714285714286e-06,
+      "loss": 1.1469,
+      "step": 1474
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 1.1635,
+      "step": 1475
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.82857142857143e-06,
+      "loss": 1.0898,
+      "step": 1476
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.8e-06,
+      "loss": 1.1769,
+      "step": 1477
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.771428571428572e-06,
+      "loss": 1.147,
+      "step": 1478
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.742857142857144e-06,
+      "loss": 1.2148,
+      "step": 1479
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.714285714285714e-06,
+      "loss": 1.239,
+      "step": 1480
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.685714285714286e-06,
+      "loss": 1.1689,
+      "step": 1481
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.657142857142858e-06,
+      "loss": 1.1541,
+      "step": 1482
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.628571428571429e-06,
+      "loss": 1.1101,
+      "step": 1483
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.6e-06,
+      "loss": 1.0905,
+      "step": 1484
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.571428571428572e-06,
+      "loss": 1.202,
+      "step": 1485
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.542857142857143e-06,
+      "loss": 1.1799,
+      "step": 1486
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.514285714285714e-06,
+      "loss": 1.2425,
+      "step": 1487
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.485714285714286e-06,
+      "loss": 1.1827,
+      "step": 1488
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.457142857142857e-06,
+      "loss": 1.1214,
+      "step": 1489
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 1.1768,
+      "step": 1490
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.4e-06,
+      "loss": 1.1479,
+      "step": 1491
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.371428571428572e-06,
+      "loss": 1.1493,
+      "step": 1492
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.342857142857143e-06,
+      "loss": 1.1698,
+      "step": 1493
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.314285714285715e-06,
+      "loss": 1.163,
+      "step": 1494
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.285714285714286e-06,
+      "loss": 1.1981,
+      "step": 1495
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.257142857142857e-06,
+      "loss": 1.1761,
+      "step": 1496
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.228571428571429e-06,
+      "loss": 1.225,
+      "step": 1497
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.2e-06,
+      "loss": 1.1828,
+      "step": 1498
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.171428571428572e-06,
+      "loss": 1.1605,
+      "step": 1499
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 1.2068,
+      "step": 1500
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.114285714285715e-06,
+      "loss": 1.1226,
+      "step": 1501
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.085714285714286e-06,
+      "loss": 1.1504,
+      "step": 1502
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.057142857142858e-06,
+      "loss": 1.2027,
+      "step": 1503
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.028571428571429e-06,
+      "loss": 1.2045,
+      "step": 1504
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.1675,
+      "step": 1505
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.971428571428572e-06,
+      "loss": 1.1905,
+      "step": 1506
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.942857142857143e-06,
+      "loss": 1.159,
+      "step": 1507
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.914285714285715e-06,
+      "loss": 1.1726,
+      "step": 1508
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.885714285714286e-06,
+      "loss": 1.1606,
+      "step": 1509
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 1.143,
+      "step": 1510
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.828571428571429e-06,
+      "loss": 1.1589,
+      "step": 1511
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.1389,
+      "step": 1512
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.771428571428571e-06,
+      "loss": 1.2338,
+      "step": 1513
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.742857142857144e-06,
+      "loss": 1.2066,
+      "step": 1514
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.714285714285714e-06,
+      "loss": 1.1679,
+      "step": 1515
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.685714285714285e-06,
+      "loss": 1.1704,
+      "step": 1516
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.657142857142857e-06,
+      "loss": 1.0813,
+      "step": 1517
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.628571428571428e-06,
+      "loss": 1.1983,
+      "step": 1518
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.6e-06,
+      "loss": 1.1327,
+      "step": 1519
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.5714285714285714e-06,
+      "loss": 1.1415,
+      "step": 1520
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.542857142857143e-06,
+      "loss": 1.1415,
+      "step": 1521
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.5142857142857145e-06,
+      "loss": 1.1915,
+      "step": 1522
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.485714285714286e-06,
+      "loss": 1.2378,
+      "step": 1523
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.4571428571428575e-06,
+      "loss": 1.1808,
+      "step": 1524
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.428571428571429e-06,
+      "loss": 1.1028,
+      "step": 1525
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.0962,
+      "step": 1526
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.371428571428572e-06,
+      "loss": 1.1545,
+      "step": 1527
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.342857142857144e-06,
+      "loss": 1.2055,
+      "step": 1528
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.314285714285714e-06,
+      "loss": 1.154,
+      "step": 1529
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.285714285714287e-06,
+      "loss": 1.1113,
+      "step": 1530
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.257142857142857e-06,
+      "loss": 1.1994,
+      "step": 1531
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.228571428571429e-06,
+      "loss": 1.1248,
+      "step": 1532
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.2e-06,
+      "loss": 1.1793,
+      "step": 1533
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.171428571428572e-06,
+      "loss": 1.1289,
+      "step": 1534
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.142857142857143e-06,
+      "loss": 1.1168,
+      "step": 1535
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.114285714285715e-06,
+      "loss": 1.2153,
+      "step": 1536
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.085714285714286e-06,
+      "loss": 1.1003,
+      "step": 1537
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.057142857142858e-06,
+      "loss": 1.1924,
+      "step": 1538
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.028571428571428e-06,
+      "loss": 1.1365,
+      "step": 1539
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6e-06,
+      "loss": 1.1494,
+      "step": 1540
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.971428571428571e-06,
+      "loss": 1.1155,
+      "step": 1541
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.942857142857143e-06,
+      "loss": 1.2853,
+      "step": 1542
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.914285714285714e-06,
+      "loss": 1.2212,
+      "step": 1543
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.885714285714286e-06,
+      "loss": 1.1777,
+      "step": 1544
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.857142857142857e-06,
+      "loss": 1.1711,
+      "step": 1545
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.828571428571429e-06,
+      "loss": 1.1916,
+      "step": 1546
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.8e-06,
+      "loss": 1.2016,
+      "step": 1547
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.7714285714285715e-06,
+      "loss": 1.1548,
+      "step": 1548
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.7428571428571426e-06,
+      "loss": 1.0878,
+      "step": 1549
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.1708,
+      "step": 1550
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.685714285714286e-06,
+      "loss": 1.1818,
+      "step": 1551
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.6571428571428576e-06,
+      "loss": 1.1823,
+      "step": 1552
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.628571428571429e-06,
+      "loss": 1.2061,
+      "step": 1553
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.161,
+      "step": 1554
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.571428571428572e-06,
+      "loss": 1.1154,
+      "step": 1555
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.542857142857144e-06,
+      "loss": 1.1852,
+      "step": 1556
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.514285714285715e-06,
+      "loss": 1.2232,
+      "step": 1557
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.485714285714286e-06,
+      "loss": 1.1364,
+      "step": 1558
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.457142857142857e-06,
+      "loss": 1.1736,
+      "step": 1559
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 1.1854,
+      "step": 1560
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.4e-06,
+      "loss": 1.1587,
+      "step": 1561
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.371428571428572e-06,
+      "loss": 1.1903,
+      "step": 1562
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.342857142857143e-06,
+      "loss": 1.2218,
+      "step": 1563
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.314285714285715e-06,
+      "loss": 1.0914,
+      "step": 1564
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.285714285714286e-06,
+      "loss": 1.1373,
+      "step": 1565
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.257142857142858e-06,
+      "loss": 1.1615,
+      "step": 1566
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.228571428571428e-06,
+      "loss": 1.0707,
+      "step": 1567
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.2e-06,
+      "loss": 1.2612,
+      "step": 1568
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.171428571428571e-06,
+      "loss": 1.1822,
+      "step": 1569
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.142857142857143e-06,
+      "loss": 1.2074,
+      "step": 1570
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.114285714285714e-06,
+      "loss": 1.196,
+      "step": 1571
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.085714285714286e-06,
+      "loss": 1.1075,
+      "step": 1572
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.057142857142857e-06,
+      "loss": 1.0347,
+      "step": 1573
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.028571428571429e-06,
+      "loss": 1.2361,
+      "step": 1574
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5e-06,
+      "loss": 1.1735,
+      "step": 1575
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.9714285714285715e-06,
+      "loss": 1.1728,
+      "step": 1576
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.942857142857143e-06,
+      "loss": 1.0923,
+      "step": 1577
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.9142857142857145e-06,
+      "loss": 1.0906,
+      "step": 1578
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.885714285714286e-06,
+      "loss": 1.1076,
+      "step": 1579
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 1.2176,
+      "step": 1580
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.828571428571429e-06,
+      "loss": 1.1989,
+      "step": 1581
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.2039,
+      "step": 1582
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.771428571428572e-06,
+      "loss": 1.147,
+      "step": 1583
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.742857142857144e-06,
+      "loss": 1.1326,
+      "step": 1584
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.714285714285715e-06,
+      "loss": 1.1783,
+      "step": 1585
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.685714285714286e-06,
+      "loss": 1.1103,
+      "step": 1586
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.657142857142857e-06,
+      "loss": 1.1544,
+      "step": 1587
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.628571428571429e-06,
+      "loss": 1.1774,
+      "step": 1588
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.6e-06,
+      "loss": 1.107,
+      "step": 1589
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 1.1337,
+      "step": 1590
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.542857142857143e-06,
+      "loss": 1.1428,
+      "step": 1591
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.514285714285715e-06,
+      "loss": 1.1712,
+      "step": 1592
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.485714285714286e-06,
+      "loss": 1.2818,
+      "step": 1593
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.457142857142858e-06,
+      "loss": 1.1271,
+      "step": 1594
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.428571428571428e-06,
+      "loss": 1.2166,
+      "step": 1595
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.4e-06,
+      "loss": 1.0997,
+      "step": 1596
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.371428571428571e-06,
+      "loss": 1.1687,
+      "step": 1597
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.342857142857143e-06,
+      "loss": 1.2076,
+      "step": 1598
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.314285714285714e-06,
+      "loss": 1.0787,
+      "step": 1599
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.285714285714286e-06,
+      "loss": 1.1682,
+      "step": 1600
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.257142857142857e-06,
+      "loss": 1.1755,
+      "step": 1601
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.228571428571429e-06,
+      "loss": 1.1749,
+      "step": 1602
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.2322,
+      "step": 1603
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.1714285714285715e-06,
+      "loss": 1.1602,
+      "step": 1604
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.142857142857143e-06,
+      "loss": 1.1559,
+      "step": 1605
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.114285714285715e-06,
+      "loss": 1.2234,
+      "step": 1606
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.085714285714286e-06,
+      "loss": 1.1566,
+      "step": 1607
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.057142857142858e-06,
+      "loss": 1.1324,
+      "step": 1608
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.028571428571429e-06,
+      "loss": 1.2169,
+      "step": 1609
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.1903,
+      "step": 1610
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.971428571428572e-06,
+      "loss": 1.1301,
+      "step": 1611
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.942857142857143e-06,
+      "loss": 1.1581,
+      "step": 1612
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.914285714285715e-06,
+      "loss": 1.1489,
+      "step": 1613
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.885714285714286e-06,
+      "loss": 1.244,
+      "step": 1614
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.857142857142857e-06,
+      "loss": 1.1807,
+      "step": 1615
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.828571428571429e-06,
+      "loss": 1.098,
+      "step": 1616
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.8e-06,
+      "loss": 1.1864,
+      "step": 1617
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.7714285714285716e-06,
+      "loss": 1.1447,
+      "step": 1618
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.742857142857143e-06,
+      "loss": 1.1248,
+      "step": 1619
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 1.0697,
+      "step": 1620
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.685714285714286e-06,
+      "loss": 1.1363,
+      "step": 1621
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.6571428571428576e-06,
+      "loss": 1.128,
+      "step": 1622
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 3.6285714285714283e-06,
+      "loss": 1.1087,
+      "step": 1623
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.6e-06,
+      "loss": 1.184,
+      "step": 1624
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 1.1856,
+      "step": 1625
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.542857142857143e-06,
+      "loss": 1.1032,
+      "step": 1626
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.5142857142857144e-06,
+      "loss": 1.1943,
+      "step": 1627
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.485714285714286e-06,
+      "loss": 1.144,
+      "step": 1628
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4571428571428574e-06,
+      "loss": 1.173,
+      "step": 1629
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 1.1487,
+      "step": 1630
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.1552,
+      "step": 1631
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.371428571428572e-06,
+      "loss": 1.1868,
+      "step": 1632
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.3428571428571427e-06,
+      "loss": 1.1577,
+      "step": 1633
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.314285714285714e-06,
+      "loss": 1.1659,
+      "step": 1634
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.2857142857142857e-06,
+      "loss": 1.2075,
+      "step": 1635
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2571428571428572e-06,
+      "loss": 1.1657,
+      "step": 1636
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.2285714285714288e-06,
+      "loss": 1.1275,
+      "step": 1637
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.1619,
+      "step": 1638
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.171428571428572e-06,
+      "loss": 1.0813,
+      "step": 1639
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1428571428571433e-06,
+      "loss": 1.1533,
+      "step": 1640
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.1142857142857144e-06,
+      "loss": 1.1754,
+      "step": 1641
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.085714285714286e-06,
+      "loss": 1.1399,
+      "step": 1642
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.0571428571428575e-06,
+      "loss": 1.1924,
+      "step": 1643
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.028571428571429e-06,
+      "loss": 1.1303,
+      "step": 1644
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3e-06,
+      "loss": 1.2072,
+      "step": 1645
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.9714285714285716e-06,
+      "loss": 1.1663,
+      "step": 1646
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.942857142857143e-06,
+      "loss": 1.1577,
+      "step": 1647
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.9142857142857146e-06,
+      "loss": 1.1738,
+      "step": 1648
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.8857142857142857e-06,
+      "loss": 1.0478,
+      "step": 1649
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.0933,
+      "step": 1650
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.8285714285714288e-06,
+      "loss": 1.1358,
+      "step": 1651
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.094,
+      "step": 1652
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.771428571428572e-06,
+      "loss": 1.1383,
+      "step": 1653
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.742857142857143e-06,
+      "loss": 1.2338,
+      "step": 1654
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.7142857142857144e-06,
+      "loss": 1.1278,
+      "step": 1655
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.685714285714286e-06,
+      "loss": 1.123,
+      "step": 1656
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6571428571428575e-06,
+      "loss": 1.1989,
+      "step": 1657
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.628571428571429e-06,
+      "loss": 1.1712,
+      "step": 1658
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.6e-06,
+      "loss": 1.1172,
+      "step": 1659
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.5714285714285716e-06,
+      "loss": 1.1932,
+      "step": 1660
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.542857142857143e-06,
+      "loss": 1.0686,
+      "step": 1661
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.5142857142857147e-06,
+      "loss": 1.1874,
+      "step": 1662
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.4857142857142858e-06,
+      "loss": 1.1438,
+      "step": 1663
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.4571428571428573e-06,
+      "loss": 1.179,
+      "step": 1664
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.428571428571429e-06,
+      "loss": 1.1445,
+      "step": 1665
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.0935,
+      "step": 1666
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.371428571428572e-06,
+      "loss": 1.2171,
+      "step": 1667
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.342857142857143e-06,
+      "loss": 1.2176,
+      "step": 1668
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.3142857142857145e-06,
+      "loss": 1.0836,
+      "step": 1669
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 1.1739,
+      "step": 1670
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.2571428571428575e-06,
+      "loss": 1.1265,
+      "step": 1671
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.228571428571429e-06,
+      "loss": 1.1278,
+      "step": 1672
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.2e-06,
+      "loss": 1.1497,
+      "step": 1673
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.1714285714285716e-06,
+      "loss": 1.1593,
+      "step": 1674
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.142857142857143e-06,
+      "loss": 1.0416,
+      "step": 1675
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.1142857142857147e-06,
+      "loss": 1.19,
+      "step": 1676
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.0857142857142858e-06,
+      "loss": 1.1379,
+      "step": 1677
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.0571428571428573e-06,
+      "loss": 1.1397,
+      "step": 1678
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.028571428571429e-06,
+      "loss": 1.1533,
+      "step": 1679
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.1168,
+      "step": 1680
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.9714285714285714e-06,
+      "loss": 1.1514,
+      "step": 1681
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.942857142857143e-06,
+      "loss": 1.143,
+      "step": 1682
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.9142857142857145e-06,
+      "loss": 1.2329,
+      "step": 1683
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.8857142857142858e-06,
+      "loss": 1.1127,
+      "step": 1684
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.8571428571428573e-06,
+      "loss": 1.1063,
+      "step": 1685
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.8285714285714288e-06,
+      "loss": 1.1945,
+      "step": 1686
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.8e-06,
+      "loss": 1.1076,
+      "step": 1687
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7714285714285714e-06,
+      "loss": 1.1526,
+      "step": 1688
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.742857142857143e-06,
+      "loss": 1.1742,
+      "step": 1689
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 1.1234,
+      "step": 1690
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.685714285714286e-06,
+      "loss": 1.1422,
+      "step": 1691
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.657142857142857e-06,
+      "loss": 1.1065,
+      "step": 1692
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.6285714285714286e-06,
+      "loss": 1.2324,
+      "step": 1693
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.1745,
+      "step": 1694
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.5714285714285717e-06,
+      "loss": 1.1504,
+      "step": 1695
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.542857142857143e-06,
+      "loss": 1.164,
+      "step": 1696
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.5142857142857145e-06,
+      "loss": 1.147,
+      "step": 1697
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.4857142857142858e-06,
+      "loss": 1.1331,
+      "step": 1698
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4571428571428573e-06,
+      "loss": 1.1321,
+      "step": 1699
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.1113,
+      "step": 1700
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.0999,
+      "step": 1701
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.3714285714285715e-06,
+      "loss": 1.1162,
+      "step": 1702
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.342857142857143e-06,
+      "loss": 1.1185,
+      "step": 1703
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3142857142857145e-06,
+      "loss": 1.1674,
+      "step": 1704
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.2857142857142858e-06,
+      "loss": 1.1942,
+      "step": 1705
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2571428571428573e-06,
+      "loss": 1.1559,
+      "step": 1706
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2285714285714286e-06,
+      "loss": 1.1529,
+      "step": 1707
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.1976,
+      "step": 1708
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1714285714285715e-06,
+      "loss": 1.1535,
+      "step": 1709
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 1.1287,
+      "step": 1710
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1142857142857145e-06,
+      "loss": 1.2045,
+      "step": 1711
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.0857142857142858e-06,
+      "loss": 1.1061,
+      "step": 1712
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0571428571428573e-06,
+      "loss": 1.1602,
+      "step": 1713
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0285714285714286e-06,
+      "loss": 1.1806,
+      "step": 1714
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.1134,
+      "step": 1715
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.714285714285715e-07,
+      "loss": 1.0925,
+      "step": 1716
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.428571428571429e-07,
+      "loss": 1.1441,
+      "step": 1717
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.142857142857144e-07,
+      "loss": 1.1646,
+      "step": 1718
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 8.857142857142857e-07,
+      "loss": 1.2171,
+      "step": 1719
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 1.1863,
+      "step": 1720
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.285714285714285e-07,
+      "loss": 1.1377,
+      "step": 1721
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.1671,
+      "step": 1722
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.714285714285715e-07,
+      "loss": 1.1242,
+      "step": 1723
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.428571428571429e-07,
+      "loss": 1.192,
+      "step": 1724
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.1033,
+      "step": 1725
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.857142857142857e-07,
+      "loss": 1.1654,
+      "step": 1726
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.571428571428572e-07,
+      "loss": 1.2146,
+      "step": 1727
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.285714285714287e-07,
+      "loss": 1.0838,
+      "step": 1728
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.1062,
+      "step": 1729
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 1.1258,
+      "step": 1730
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.428571428571429e-07,
+      "loss": 1.1799,
+      "step": 1731
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.142857142857143e-07,
+      "loss": 1.1261,
+      "step": 1732
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.857142857142857e-07,
+      "loss": 1.1668,
+      "step": 1733
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.571428571428572e-07,
+      "loss": 1.0714,
+      "step": 1734
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.285714285714286e-07,
+      "loss": 1.1479,
+      "step": 1735
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.2092,
+      "step": 1736
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.7142857142857145e-07,
+      "loss": 1.1544,
+      "step": 1737
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.4285714285714286e-07,
+      "loss": 1.1102,
+      "step": 1738
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 3.1428571428571433e-07,
+      "loss": 1.162,
+      "step": 1739
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 1.1376,
+      "step": 1740
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.5714285714285716e-07,
+      "loss": 1.1805,
+      "step": 1741
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.285714285714286e-07,
+      "loss": 1.1403,
+      "step": 1742
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.1657,
+      "step": 1743
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.7142857142857143e-07,
+      "loss": 1.1106,
+      "step": 1744
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.4285714285714287e-07,
+      "loss": 1.256,
+      "step": 1745
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.142857142857143e-07,
+      "loss": 1.1122,
+      "step": 1746
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.571428571428572e-08,
+      "loss": 1.1415,
+      "step": 1747
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.714285714285715e-08,
+      "loss": 1.1179,
+      "step": 1748
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.8571428571428575e-08,
+      "loss": 1.1105,
+      "step": 1749
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0,
+      "loss": 1.1718,
+      "step": 1750
+    },
+    {
+      "epoch": 10.0,
+      "step": 1750,
+      "total_flos": 4.233246129782784e+16,
+      "train_loss": 1.389875247819083,
+      "train_runtime": 2378.4813,
+      "train_samples_per_second": 26.366,
+      "train_steps_per_second": 0.736
     }
   ],
-  "max_steps": 786,
-  "num_train_epochs": 3,
-  "total_flos": 1.2699738389348352e+16,
+  "max_steps": 1750,
+  "num_train_epochs": 10,
+  "total_flos": 4.233246129782784e+16,
   "trial_name": null,
   "trial_params": null
 }